Common startup advice of the B2B SaaS era says:

pick your ICP and focus

That was right.

But it’s not anymore.

The Constraint was never the Strategy

Let’s be precise about what that advice was actually doing, because the reason it worked matters.

Meaningful personalization required human effort—researchers who found the relevant detail on a prospect, writers who crafted a message for a specific pain, SDRs who spent 40 minutes per contact making the outreach feel considered rather than templated. A five-person team running that work across 20 different segments simultaneously would collapse under the weight of it. So the strategic advice was: pick your best bet, go deep, don’t dilute.

The data validated it. Companies with tightly-defined ICPs spent 50% less on sales and marketing and see 24% shorter CAC payback. The advice worked because the constraint and results were real.

But here’s the thing — the constraint was doing all the work, not the principle. The goal was never narrowness for its own sake. It was finding the customers who genuinely want what you’re building and serving them well. ICP focus was the most resource-efficient path to that goal given what was operationally possible. When you couldn’t afford to personalize at scale, picking one segment and going deep was the right call.

Now? The operational ceiling changed. So the advice needs to change.

GTM as a “Yes, and” machine

There’s a principle in improv called “Yes, and.” When your scene partner makes a move, you don’t block it—you accept what they’ve given you and build on it. That’s how scenes go somewhere. The moment someone says “No” or “but” the scene dies.

Traditional GTM is the same. It runs on “No, but.” The ICP is a filter, and the filter does most of the work before any real conversation starts. Wrong industry, wrong company size, wrong title—the answer is no, before the prospect ever sees a message tailored to their situation. The filter was necessary because building something relevant for the people it filtered out would have cost more than it was worth.

When the cost of personalization approaches zero, the filter becomes the liability.

Every visitor can get a landing page built for their industry, their role, their specific framing of the problem. Every keyword gets an ad written for the intent behind it. Every outbound prospect gets a sequence that speaks to what their company is actually working through right now. The filter disappears because the reason for having it disappears.

The conventional wisdom was “you can’t boil the ocean—pick your cup.” The real constraint was that heating any cup took expensive human effort, so you could only afford to heat a few.

But now you can heat every cup precisely, simultaneously, for nearly nothing—and when you do, you find customers that the company still running the old filter never knew existed.

Can you hear me now?

Introducing Frequency, a tool for modern GTM.

The ICP was always a declaration. You sit in a room, look at your first handful of customers, and announce who your ideal customer is. Then you build a filter from that declaration—the profile drives the channel, the channel drives the content—three sequential stages, each derived from the last. Months of work optimized around one hypothesis before the market responds.

What AI makes possible is a different atomic unit entirely: the GTM Frequency.

A Frequency is defined by three elements, collapsed into one thing:

1. WHO: the individual it’s tuned for: their role, their company context, their individual signals.

2. WHAT: the value proposition, message, or call to action carried to that person

3. HOW: the channel it travels through: outbound email, paid ad, landing page, LinkedIn, or an onboarding flow.

In the old model, these were separate decisions made in sequence. The profile (who), determined the content (what), determined the channel (how). Three stages, each dependent on the last. But when the cost of each drops to zero, new opportunities arise.

A Frequency collapses them into a single horizontal element—one testable thing that either resonates or doesn’t.

The permutation space this opens is enormous.

With AI creating and executing them simultaneously, you can test all 1,000 in the time it used to take to run one campaign for one declared ICP. And you don’t declare which ones will work. You can just… find out.

You don’t declare a Frequency before testing. You generate thousands of them, broadcast simultaneously across every GTM touchpoint, and let the market tell you which ones hit.

The ones that work aren’t a “segment” you’ve identified or a “profile” you’ve declared. They’re just the frequencies that resonate, for the individuals they resonate with. The market did the ranking.

What does that change?

When we were building Tarka’s 2nd product, an AI SDR, we ran a manual version of this without having the framework to name it. We had a product with real horizontal value — AI-powered professional relationship management — and instead of picking one ICP and building everything around it, we ran parallel framings simultaneously. Customer discovery teams, market researchers, product teams running continuous discovery, people building expert networks — each was a real use case, each needed completely different language for the same underlying capability.

We built all of them.

Custom landing pages, custom sequences, custom messaging for each framing. Individuals responded differently to each one. The language that worked for a product manager had no resonance with a sales leader.

What came back wasn’t just conversion data — it was a map of which frequencies were hitting and which were broadcasting into silence. The market told us who our customers were, and how they thought about what we’d built. Then we built the product dynamically to match what the signal confirmed.

That was a manual process, run by people, tracked in spreadsheets.

What’s possible now is that system running automatically — generating and testing new frequencies continuously, retiring the ones that don’t resonate, surfacing the ones that do, and feeding that signal directly into product decisions. The gap between “declare an ICP” and “run a Frequency portfolio” is the gap between one upfront guess and a continuous market truth-telling system.

Broadcasting at Scale

Most teams who discover AI personalization treat it as a single technique. They adopt Clay and start running personalized outbound. Or they implement Mutiny and start varying landing pages by segment. They pick one, measure it, and call that their AI GTM motion.

But that’s just one frequency.

The shift is broadcasting across the full spectrum simultaneously — every channel generating Frequencies in parallel, across every WHO hypothesis you have. The WHO varies by job title, company type, intent signal. The HOW varies by channel — paid, outbound, landing page, onboarding. The WHAT varies by value proposition, message angle, and call to action. AI writes the variants in seconds and executes them across every touchpoint at once as fast as there are individuals.

The reason this compounds is that signal from every channel feeds the same discovery loop. A Frequency that converts in paid acquisition but falls flat in outbound tells you something specific about timing and intent. A landing page variant that holds attention but doesn’t convert tells you the framing is landing but the offer isn’t. When every touchpoint is instrumented and running simultaneously, you learn faster than any sequential experiment on a single declared ICP could produce.

The numbers from companies who’ve adopted parts of this out are not marginal improvements on the old model.

LaunchDarkly built more than 100 account-specific microsites in 60 days and hit 150% of their quarterly pipeline goal.

Zapier built 50,000 programmatic landing pages and now drives 16 million organic visitors a month.

Teams running AI-personalized outbound at the right level of specificity are hitting 15–25% reply rates against an industry average below 5%. These results come from a categorically different operating model — one where the market is constantly responding to you, and you have the infrastructure to hear it at the individual level.

It doesn’t stop at growth

This Changes What “Product” Means

When your Frequency portfolio is running continuously — when the market is giving you real-time signal about which combinations of who, what, and how are converting — that signal doesn’t just tell you who to target. It tells you what the product should be for the individuals who respond. The GTM discovery loop and the product development loop start to merge.

Product-market fit stops being a milestone you declare after enough iteration and becomes something you’re reading and responding to continuously, at the individual level.

What this enables over time is building more like a platform than a single-use SaaS product.

SaaS says: declare your ICP, build the features they need, hold that surface stable enough to sell and support.

Platform says: build a flexible core, let your Frequency portfolio tell you how different individuals and markets want to use it, and let the product evolve in direct response to what you’re learning. You’re not building features for a profile you declared. You’re building in response to the frequencies that are actually resonating.

“Sell to everyone” used to describe an unfocused company that hadn’t done the hard work of finding product-market fit. We think it’s becoming a description of an AI-native company that has built the infrastructure to find fit continuously — across many markets, at the individual level — and serve each of them well.

⚠️ This doesn’t mean every company should attempt to serve every market immediately. If the individuals responding to different frequencies genuinely need different core products with no shared foundation, that’s multiple products or even multiple companies, not one platform.

But if you have a product with real horizontal value and you’ve been conservative about your ICP because personalization at scale seemed operationally impossible — the thing holding you back isn’t your product. It’s an assumption about what’s feasible that AI has already made obsolete.

So what do you do, today?

Get started.

Your competitors with five-person GTM teams might already be running hundreds of Frequencies.

Custom experience for every visitor type. Custom message for every meaningful keyword combination. Custom sequence for every individual’s context and signals. And when certain Frequencies start resonating — when the market starts responding — they’re already building the product toward what those individuals actually need, not what a declared ICP would have predicted.

The ICP declaration made sense when personalization was expensive. When running one GTM motion well required the full capacity of a small team. The mental model it created — pick one customer type, declare them ideal, filter everything else out — was a rational response to a real constraint.

The constraint is gone. The mental model is the last thing holding most teams back. Prove to your investors and stakeholders that this is the future or prove that you lost.

We don’t fully know yet what the ceiling is for companies that get this right — that build the infrastructure to generate, test, and refine thousands of Frequencies continuously, with the product loop feeding back into the GTM loop feeding back into the product. We think it’s significantly higher than what the ICP model could ever reach.

Let’s find out.

Tarka builds AI-first infrastructure for GTM teams — including the discovery loops that make Frequency-driven GTM possible. See our playbook and more at tarka.ai.

Today, there are dozens of transcription companies.

Just off the top of my head, in alphabetical order:

• Clarify integrates with my CRM.

• Dialpad does a great job with phone calls.

• Fathom quickly creates video clips on the call to automatically so you can easily reflect on them later with Voice of the Customer.

• Google Meet does live transcriptions and language translations.

• Granola does it without recording the audio.

• Krisp does audio recording and noise canceling without a meeting bot.

• Otter does a great job on mobile.

• Recall allows anyone to build a meeting bot, easily.

• Zoom even has it built in now.

I absolutely love them because they make my workflow feel like a superhuman.

And while they all supposedly deliver “summary reports”—something is wrong—and it’s none of their faults. It’s the underlying technology. The math behind the language models.

Summarization is more than just the average weight of commonality of words.

Human verbal and and non-verbal communication implies a fuller experience we’ve yet to encode in models. To understand what this means, lets take a super specific example. Let’s assume just verbal communication. Let’s assume just English only. Let’s assume that we can capture every word that is said, perfectly, with just transcription, and use the example:

The sentence"You're going to wear that?" can mean a genuine question, a surprised inquiry, a skeptical comment, a critical observation, or a playful tease—all depending on where the emphasis is placed.

Rising intonation on "that": "You're going to wear that?" - This sounds like a genuine question, asking for clarification about the clothing choice.

Falling intonation on "that": "You're going to wear that?" - This could sound critical or skeptical, implying the speaker disapproves of the outfit.

Emphasis on "you": "You're going to wear that?" - This could suggest the speaker is surprised or questioning the person's decision-making.

Emphasis on "wear": "You're going to wear that?" - This might imply the speaker is questioning the practicality or suitability of the clothing item.

Playful intonation: "You're going to wear that?" - With a lighter, teasing tone, this could be a playful comment about someone's outfit choice.

This is also easily seen when asking an LLM to summarize a negative proof research paper or satirical news article—where LLMs have limits similar to the human limitation of Poe’s law. The paper may spend dozens of pages on proposing a counter-proof dripping in satire and a single paragraph on the actual result, and easily miss the entire meaning of the work.

This is one of the foundational reasons why we’re way off from a fully-agentic world.

Further Reading:

• The Future Will Be Brief, John Herrman

• The LLMentalist Effect, Baldur Bjarnason

• When ChatGPT summarises, it actually does nothing of the kind, R&A IT Strategy & Architecture

TL;DR

AI Agents need to perform 5 things to be comparable to humans:

1. Do things.

2. Do things better, by learning.

3. Reason over weeks and months of time.

4. Collaborate with non-agents towards outcomes.

5. Have autonomy within resource constraints.

What is an agent?

There’s a lot of debate right now on what an AI agent actually is.

Today’s definition of agents are an LLM with reasoning, and tool-calling. While this appears to act like a human does, this couldn’t be further from the what non-programmers expect. While you could imagine these systems getting better, faster, and cheaper—that simple framework will never compare to human agents.

We suggest that this is a functional view of a agent-like system, not a systematic view required for relating AI agents to humans. The framework we propose will redefine the term “Agent” and break down a conceptual model for what it will take to get to an equivalent to human agency by moving entire cognitive tasks onto computers holistically.

So here’s our first pass to share our thinking, terminology, and full maturity model we’ve adopted while building our Agent platform, Tarka.

A layered approach

Taking inspiration from the OSI model of networking, let’s start with our building block: A Layer

A Layer is a logical separation from other layers above and below it. It receives input and returns output. The layer itself doesn’t usually understand the full context of the layer above it (the parent), but just enough to do its work, often by creating and managing lower layers (children). On top of input, each layer has some unique properties that separate it from other layers.

Usually multiple duplicates or different instances of the same layer make up the work of the parent layer. A better contextual visual of a single layer might be this:

The Agent Maturity Model (AMM)

The layer construct allow us to conceptually build a 5-layer system that makes up an Agent and construct unique grammar for the entire system that we can use as a shared glossary.

Let’s break each layer down, starting from the bottom.

Throughout this, let's compare it to how humans themselves mature.
I have 2 young children (5 and 7 years old) and even though I’m not a professional teacher, I am often reminded how simple things I take for granted can be quite difficult. Let's look at my fleshy little agents along the way.

Functions

fka tools or API calls

The first layer (starting from the bottom) is the most simplistic building block: A Function.

They are black boxes to the agent—so from the agent’s perspective they only appear to work or not. It might be idempotent, but it might not. Importantly a Function cannot reason.

A Function is directed by the Command input and returns a Response output. Used by Skills, Functions are the most granular operation, usually a single service (code) or API call that can be run in sequence or parallel with other Functions.

If you’re interested in building AI agents, MCP is probably the leading Open-Source implementation of this layer. Read more: https://modelcontextprotocol.io

In the real world, Functions are things like inanimate objects. They are tools that Agents use to do things. A chair can be pulled out from under a table to sit on or pushed under the table to clean up. It can also be turned upside-down to be used as a fort—or to store toys.

For my kids, they’re learning how to brush their teeth. They’re learning how a toothbrush works and different types of toothpaste—both Functions. They don’t understand how the molecular layer of toothpaste works to clean our teeth, but they don’t need to in order to get the desired result from the toothpaste.

Learning functions? Yes! Agents develop their own set of Functions from the vast array of the possible. How? Through skills.

Skills

fka actions, tasks, activities, steps, memory, or stories.

Skills are the second layer that makes up an AI agent—and one that most “AI agents” currently do not have. It takes no reasoning to execute and, to most LLMs models, looks like a functional “tool” call. Skills make up a composite set of Functions that are joined together by an Instruction input to deliver an Execution output. Cognitive reasoning is not required.

Agents may turn reasoning into repeatable Skills with code workflows for execution time similar to how humans “crystallize Intelligence”—also called System 1 processing in the Dual-process theory. They are things that we take for granted, but each had to develop at one time. They’re usually functional even with neurocognitive impairments such as ADHD or dementia—but some may never develop them.

The reason we’ve added this as a second layer separate to Functions is that it can be improved upon. An Agent, just like a human, has to be able to develop its own Skills. It’s “saved” into code and optimized (driven by the three pillars of cost, speed, and quality for the desired use case).

To “get ready for bed” my kids have to do different Skills like “Take a shower” or “Brush teeth” or “calm our bodies down”. These all take cognition and little variables can easily throw off a routine Skill (what happens when there’s no hot water or the toothpaste is empty?) as they face Functions they don’t have yet (buying toothpaste) or don’t yet know how to develop new skills through reason (get toothpaste from another bathroom).

Importantly they can improve the “brush teeth” skill over time. They can create “conditional logic” and save it so they don’t spend mental load on it in the future.

So should all Agents be able to code? Yes! All agents should have the “learn” Skill to write and save code. Just like how human agents can improve their own Skills, AI Agents need to be able to improve their Skills—their substrate is just code. How? with the next Layer: Abilities.

Abilities

fka reasoning, workflows, processes, state, or chains

Abilities are the largest unit of work that a single Agent can complete independently—usually with reasoning and as a non-deterministic workflow of multiple Skills. It receives a Specification input and delivers a Result output. This is what current AI is simply calling “reasoning”.

In humans, this is “Fluid Intelligence” or System 2 processes in the dual-process theory. It is executive cognition that requires mental memory, cognitive flexibility, and inhibitory control that all happens in the prefrontal cortex.

Additionally, just like traditional code is saved into workflows or processes, Agents can be optimized to reason through their Skills—to choose the right Skills in the moment to produce a result. If an Agent is performing the same set of Skills in a flow, then the Agent should consider crystallizing that intelligence into deterministic code.

If you’re interested in building AI agents, Letta (fka MemGPT)’s State is probably the leading Open-Source implementation of this layer. Read more: https://docs.letta.com/stateful-agents

Let’s look at how my kids “Get Ready for Bed” on their own. This requires high-level reasoning and knowledge to create a workflow to stitch them together depending on different contexts. My kids each reason through the process differently. My son prefers to eat a red vitamin before a shower and after he brushes his teeth while my daughter prefers to shower, eat an orange vitamin, then brush teeth.

From an agent perspective, the Ability is the same, as the results are the same. But their reasoning and workflow are very differently encoded.

What about chains of chains? Chains of Abilities? Abilities of Abilities? We just call those Abilities—they’re no different—unless the Agent cannot accomplish it by itself. Then it’s called a Project.

Projects

fka as programs or plans

A structured collection of work comprising a series of Abilities aimed at reaching a specific Outcome output based on its Objective input. Importantly, Projects require outside collaboration—for example between Agents or things outside of their control (humans, systems, etc) to accomplish Outcome. When within a single agent without an interface to others, it can appear intrinsic but cannot reach an Outcome on its own.

If your agent requires an extended period of time or is waiting on additional input, then it’s a good sign it’s operating a Project—not just a self-contained Ability.

We have yet to find a leading provider of the Project level. If you know of one, we would love to review them. The closest thing we’ve found is different attempts at Multi-Agent Systems (eg: https://docs.letta.com/guides/agents/multi-agent), but this is missing non-agent operations.

The reason my kids “get ready for bed” at a certain time is to keep them in a routine and make our lives easier. They can then wake up happier and be productive at school or play the next day. So their Project for getting to bed is to “continue developing” where the Objective resets daily—ie “learn a little bit tomorrow”—and the Outcome depends on the success of their “get ready for bed” Ability’s Result the night before. If they do not correctly get their Result of 8 hours of sleep then they’ll be cranky and less likely to “continue developing” the next day.

So with outside collaboration, we just need one more thing—autonomy.

Agents

fka workers, employees, entities, people, or actors.

An individual entity with autonomy by an Assignment input that outputs Work product. Responsible for sharing resources between Projects that execute Abilities to accomplish Work. Can communicate (and interact) with other Agent(s). Has autonomy (decision making authority) and control over their own Projects within their Assignments. They have a set of Skills that they have learned and developed across their entire experience. They deftly can handle changing context, vague assignments, and random variables to varying degrees of success, often based on resource constraints (ie: capacity).

Building an autonomous layer for AI? We’d love to evaluate it.

My kids are Agents. They have autonomy (as they learn Abilities, they’ll get more Assignments and manage more Projects—every day). Their current Assignment is to learn to be the best human they can be within our family’s “child rearing” budget (resources).

So where does this leave us?

Putting it all together

We now have a full-view maturity model of what layers it takes to reach agency.

So what does a real world AI agent look like? Let’s look at an example through the lens of “Go to Market”, where Tarka is building today:

Today, so-called “agents” can “reason” through a set of “tools” based on “context”. While this gives the appearance of agency, it’s a very limited definition of such—and is very far from human agency.

It leaves open the huge gap of learnable Skills, adaptable Abilities, outcome-driven Projects, and autonomous Agents that humans perform that completely under delivers what Silicon Valley has rushed to deem “agents”. In short: We’re still in the toddler stage of AI agents.

Beyond standalone agents

But the world isn’t just made up of individual agents. Agents can work together and be coordinated—just like humans. That’s why we’ve envisioned a future based on Organization Design that encompasses 3 layers above agents:

Teams

fka. groups, collectives, squads, crews, rosters, clubs, or agencies

A collaborative unit within a Company, made up of Agents who work together on Projects that are driven by a Purpose (input) and produce Product (output)—ideally contributing to the company’s objective (ie. Value creation). Usually led by a specific manager or lead/head Agent, but not always.

My family is a Team. It’s made up of the children agents and the parent agents. We collaborate (and plenty of times don’t), but we’re sharing Purpose “Thrive” (input) and working towards the same Product of “Survived 2025” (output).

Companies

fka. organizations, clans, firms, businesses, corporations, or cooperatives

A distinct organization that uses 1 or multiple Teams working towards a unified mission and leverages shared resources. It is driven by Resources (input) and returns Value (output) It functions within the Ecosystem.

My family is superseded by my lineage to form the Moore Clan. The current grandparents have their own Team and often interact within our company. The entire family tree is a company that has constantly withstood the test of time within an Ecosystem.

Ecosystems

fka. markets, industries, or environments

The broad competitive environment where multiple companies vie for market share, resources, and strategic advantage by creating and capturing value. It has constraints in the form of Environment input and produces a collective value “GDP” as output. There are many Ecosystems, all overlapping, within the entire World.

Another human example

For another fun example, how do you teach someone, with just written instructions, how to make a PB&J sandwich? If they were born yesterday, it’s actually quite difficult:

Where does [agent-company] fit?

We’d wager that all AI agents today are not up to our standard of “Agent” in the Agent Maturity Model definition. We’ve only begun to scratch the surface of AI agents and so far none are autonomous and most certainly do not build off of Projects or even Abilities. While GPT-4o and similar “reasoning models” offer glimpses into reasoning Abilities, they’re often integrated directly into functions and then called “agents” (ie: LLMs that can “call tools”).

While “natural language reasoning” is one approach to reasoning, it’s not actually what we consider cognitive reasoning, but better thought of as chained prompting. Why? It has to do with how language is encoded.

So what?

AI is Tech’s “Quartz Crisis” moment. VC-funded companies are throwing stuff at the wall to see what sticks. But intelligence, like human agents, takes a long-term strategic view to get it right.

In true Silicon Valley fashion the market has jumped to the wrong definition too quickly and has over-promised and under-delivered.

We don’t suggest that development should stop or slow down—just that there is a large difference between the written word and the human-experience we’re emulating with LLMs.

So lets stop just throwing dumb things together and hope for the best. Let’s start delivering with intentionality and delivering intelligence for agentic work. Let’s get this right.

Related readings

• Antonio Damasio’s Descartes’ Error

• Anthony Diamond’s Guide to Identifying High-Leverage AI Agent Applications

• Dan Ariely’s Predictably Irrational

• Daniel Kahneman’s Thinking, Fast and Slow

• Herbert Simon’s Bounded Rationality

• Jeff Patton’s User Story Mapping

• Jerry Fodor’s The Modular Mind

• Julian Jaynes’ The Origin of Consciousness in the Breakdown of the Bicameral Mind

• NfX’s Onboarding AI Agents

• Teresa Torres’ Opportunity Solution Trees

Related topics

• Behavioral Economics

• Business Process Modeling

• Cognitive Reserve

• Executive Functions

• Machiavellian Intelligence / Social Brain Hypothesis

• Multi-agent systems

• Neural Efficiency Hypothesis

• Objectives & Key Results

• Organizational Architecture

• OSI Model

• Service-oriented architecture

• Systems Design

Last updated March 1, 2025

All feedback and criticism is welcome.

Thanks to A and E for teaching me how to develop agency.

This guides details Stage 3 of our founder-driven sales framework. Be sure to see the overview to check your assumptions and use Stage 1, Prospects to Problems, and Stage 2, Problems to Pilots, before you try these techniques.

Getting your first pilot signed is huge. It was not an easy journey to get here—but we’re still early. Before you can begin to scale your business, you’ll want to use this opportunity to build the right business, and prove to yourself that it’s scalable.

We break it down into 3 steps of a cyclical process that never stops:

1. Deliver value

2. Deduce what works and destroy what doesn’t

3. Double down

We’ll also use a few stories 📖 based on our experiences to help make these perspectives more concrete.

1. Deliver value

You’ve signed a commitment to deliver some value in the form of a pilot. But how do you deliver it?

Prioritize learnings

The point of these pilots are to go deep with your customers. Use these engagements first and foremost as a learning opportunity. We like to bake it into the customer agreement that they’ll meet with you on a regular cadence in exchange for the discount or free value. Their time is 10x more valuable than any amount of money they can initially pay.

Here’s one way you can use consulting to develop the right solution:

📖 Consulting a solution instead of self-service

For Vistaly, product managers and business consultants have many steps to ultimately improve the alignment and do better product discovery. While Vistaly is a great product for experienced leaders that have already figured out their business metrics and how they map together, helping them operationalize their strategy to build consensus and keep it updated—it’s difficult for the rest of us—for people that have never documented a strategy in the first place.

To learn how to best deliver an experience to do this, Matt runs workshops—multi-hour consulting calls that allows Vistaly to perform this work with them. While expensive, this gives Matt the time to really understand how people perform this job today, see what questions they have, and where they struggle.

It turns out that there’s a 2-step process here:

1. Laying out everything they’re working on and what they care about first

2. THEN figure out how they’re all connected.

His live prompts during step 1 gets immediate feedback, further questions, and better results.

So this is one way you can maximize your learning. Try to perform this work with them. A single 4-hour session might seem like a lot of work, but that’s a lot cheaper than building software for 4-6 weeks to realize you built the wrong thing.

Do things that don’t scale

These pilots are your minimal viable products, the least amount of rigid product (work, service, or program) that can maximize what you need to learn. 

If you have to build, build as little as possible. Code is not an asset, it’s a liability. These are code spikes. None of this may work and you’ll likely have to throw it all away once you learn something else.

Create interactive “fake doors”. Publish content to a URL to make it feel more “real”. Add on a mouse tracker and you can “see” what they do. It doesn’t matter how the content and data got there.

Retool, Notion, Airtable are some of our favorite applications to deliver value in a “fake app” way that all minimize the hard-content creation.

A note on speed

Manual MVPs are slower than automated software. Yes, speed is one aspect of value, but it’s not the only aspect. If your delivery is slower than a fully automated solution (since it’s a fake pilot) you may often hear that “this would be valuable if it were faster”.

One way to test this feedback is to try to figure out a pilot version that matches their time needs, but limit the amount of the pilot. For example, if your solution is an AI-driven Zoom-based BDR coach, see if you can hop on a sales call as a fake recorder and live-coach the BDR over slack. But only offer that for 1 call a day. The rest are delivered slower.

If there’s a substantial difference to value in this case, see if you can measure it or if it’s just perceived-value (ie: look for proof that there’s a material outcome difference between the live coaching and after-the-fact coaching). It may not be there.

2. Deduce what works and destroy what doesn’t

Once you’re in pilots you’re in learning mode. Your job is to understand everything. Every little detail of the process should be clear to you.

Map it out

Using Miro or a whiteboard, create a map of the entire process and all the sub-steps. You can be as rigid as a User Story Map or as messy as sticky notes. Sort through each piece you don’t understand and prioritize what you need to learn and understand next.

Challenge yourself

Playing devils advocate can really help bolster your original position. You don’t have to believe what you are saying to raise these objections—but to do so genuinely can help you dive deeper into your learning.

We personally find this difficult to accomplish as an individual and like to bring in a partner or coach to help. With the advent of ChatGPT, GenAI LLMs are an accessible method at helping with this as well. Whichever way you go, here are some questions that can help challenge your thinking:

“[Learning]. What exceptions are there to this?”
“[Learning]. Is this biased in any way?”
“[Learning]. Who else might this impact? What does this mean to them?”
“[Learning]. What scenarios might this not be true in?”
“[Learning]. What’s a hypothetical scenario that refutes this?”
“[Learning]. What alternative explanation might there be for this?”
“[Learning]. What’s the absurd extreme version of this?”
“[Learning]. What flaws might exist within this?”
“[Learning]. Why won’t this work?”
“[Learning]. Who is hurt by this?”
“[Learning]. What ambiguity exists in this?”
“[Learning]. What assumptions are hidden in this?”

Experiment

If you don’t think there’s value, try skipping a step or piece of a deliverable to see if the customer misses it and demands it back.

Try different versions for each step. You’re not shooting for perfect delivery, you’re shooting for perfect understanding. Figure out what’s really working here and get rid of everything else. Yes, you’ll need to do some pre-work and steps to deliver that ultimate value, but everything else is icing and extra.

All of these techniques can help you build your understanding. Here’s one way we corrected our assumptions while learning how the customers thought:

📖 Improving existing software? Ride along.

For another startup, we were learning how to optimize complex staffing payroll. A payroll controller ran reports to gather information from their systems. By sitting next to them at their computer, we saw how cumbersome the process was to coordinate values from different systems. How they tried to “touch” the screen at times and how they thought about their data flow in steps—independently from the features of the software today.

For our own understanding purposes, we diagrammed out the steps that person was going through, with a focus on what data they needed when, what questions they were asking, and what confidence levels they had.

Our first take, to simplify the entire process into a single button, didn’t deliver the real value they were looking for.

They’d rather have what we thought of as bad—more friction. That taught us that a middle-ground solution that was able to showcase the work with preview spot-checks delivered the value of confidence over a simplistic “do it for me” approach. Sure, the CFO would rather payroll gets done correctly and faster—but the controller needs to guarantee that outcome, which requires a high level of confidence.

Get more pilots

While you’re delivering value and understanding it, don’t stop selling. Net-new research conversations give you the opportunity to try out learnings with other perspectives to learn objections and challenges from your pilots in other environments. You should still be spending more than 50% of your time talking to new prospects and continue to sell. This is where having a cofounder can come in handy—as it’s difficult to do both at the same time.

If you need to pivot to another stakeholder, keep the original pilot happy—make them whole by giving their money back or asking what they need in exchange for the learning. They might not be in your target market, but they’re another connection that can make or break your future work.

Beware your biases

Every founder has biases that they need to work through and check at the door.

For example, technical founders enjoy building and like to over-engineer solutions. While building a scalable product is great for later-stage products, it actually hurts us now—because we’re much more likely to build a solution that the target market doesn’t understand or know how to use.

Founders with a sales (AE/closing) background, on the other hand, will likely find it easier to build for a single customer instead of generalizing across their entire customer base—and often expand features and value to compete based on objections, instead of prioritizing ROI of each feature developed.

Whichever bias you have, push yourself out of that default mode of operating and strive for better, conscious investments into the next stage—when you double down on your bets.

3. Double down

Off the success of our first few pilots, let’s get to building more and creating more value.

Start to automate core pieces to what is working. Prioritize those features based on a ROI calculation—high returns with small investments. At some point you’ll need to build a lot more, but here you have access to the resources to make it possible.

Surprise! You’ve got traction

These pilots are proof—proof to yourself, investors, and customers that you can really deliver value. Now your goal is to scale that up.

If you can, continue to use that initial traction with customers and keep the relationship just delivering value with them. Convert those existing pilots into full-blown customers. That’s the easiest form of business.

If you don’t have the balance sheet to grow, look for alternative options. Reduce employee wages by offering them equity or alternative benefits. Reduce vendor expenses with payment terms. Look for loans on AR (long-term contracts) or leveraged assets. This calm growth pays big in the long run because it’s compounding.

But it’s not always feasible—and venture capital might be a good partner for you. Pre-seed investors, angels, and friend and family are a good source of capital now that you have initial traction. Just beware that these silent partners are looking for returns—and you’ll enter the comparison game of all investments—likely leading to decisions that don’t best serve the owner-operator.

Growth from pilots

One of the most valuable assets you have as an actual business is your customer base. Turn their success stories into yours—you can use their logos as marketing collateral and their testimonials as social proof to your outreach and sales.

Get referrals to their friends and competitors. Learn from them what community they live and operate in professionally. Go to the conferences they attend, publish ads in the articles they read, and host webinars with them. Create look-a-like profiles for social media ad-spend or find other ways to learn where to find 10 others just like them.

Each growth playbook is unique per market and changes over time. Here’s a story of one that worked for us:

📖 Improving the physical world? Go there.

For Awaiting, early on in the pandemic contact-tracing was all the rage. We theorized that just by collecting phone numbers we’d be able to deliver a manual form of contract tracing for businesses that might be on the hook to perform it for regulation purposes.

Turns out that regulation never came to pass in America (Canada was a different story)—but what was needed was less crowds. By creating a virtual waitlist service we were able to capture all sorts of scenarios where crowds were an issue. We helped BTS concerts line-up while staying far apart, Job fairs regulate their lines, resorts handle check-in on popular (and cold) weekends, and even daycares handle parents picking up their kids.

After failing to sell to many “look alike” companies, we had to go to these events to really understand the value we created and be able to find other opportunities. Physically experiencing the lines, proximity, and visual capabilities that a physical space had showcased where a virtual waitlist would really prosper—and how it was incredibly different from curbside service (Ala drive thru) at Best Buy.

Without seeing the whole picture, we would’ve missed the surge-event opportunity, which was ultimately where the majority of our revenue came from.

Some things you can’t do from behind a desk. Sure, travel costs money, but it doesn’t cost as much as the missed learning or opportunity.

Now what?

Congratulations, you’ve gotten some market traction and can re-run your projections to see what kind of business you’ve created.

Look at your goals and plan what you need to get there.  You should now be feeling the burn of the candle at both ends, and might want to consider raising funds to scale faster or grow at your comfortable rate.

There’s several directions you can head now, and several alternative playbooks. We’ll cover some growth options like outbound sales, enterprise sales, and product-led growth models soon, but with initial recurring revenue, continue to support customers while you spend on acquiring the next customer. At this point in the public SaaS market, you’re not just looking for CLTV > CAC, but things like:

• 90% profit margins

• acquiring new customers in a scalable way

• some referral engine for new customers

• framing a market category

• constantly evolving to a changing world

🎉 Congratulations, you’ve come a far way—we’re excited to see where you’ll grow from here!

When using generative AI chat interfaces, have you ever been stumped on what to ask?

Yeah, you’ve probably read something about prompt engineering—HOW to write prompts—that is how to write a prompt right—but what about the RIGHT prompt?

Yes, we have creative techniques like suggestions and chat history. We could even just duplicate the abusive and often dangerous Surveillance Capitalism playbook and use contextual data, that companies like Google have, to guess what we’re trying to accomplish—based on our locations, previous behaviors, and what’s going on in the world.

But what if there was a better way?

Let’s try another perspective…

As an operator with a huge propensity to act, I’m very good at responding to things. If I were a creative professional, I’d be much better at editing than generating content. That’s probably why I like to practice Inbox Zero. I’ve trained myself to be incredibly effective at doing—not thinking.

Could generative AI help us think better?

What if AI could hold our goals in mind?

What if it could see and understand everything that we experience?

What if it could ask us the right questions to move us towards our own goals?

What if, instead of having to type out a prompt, it gave us content to consume, think about, learn from, and the ability to react?

Could AI send me a message to keep me focused and thinking about the right next thing?

How would it need to work?

It couldn’t be stupid, or else I’d find it a waste of time. It would have to be smart, clever, and clear. It would have to provide us value, knowing where we were today and pushing us at a healthy pace of improvement.

It couldn’t be rigid, or else I would ignore it. It would have to be flexible to our behaviors, forgiving us when we messed up, and being patient when we needed it.

It couldn’t be ignorant, or else I wouldn’t get what I need. It would have to be critical and revealing of blind spots. It couldn’t ignore our shortcomings or shy away from where we need help.

It would have to be kind.

It would have to do all of this at the exact levels I would want it to be—which is different than what you’d want it to be—so it would have to be personalized.

What then?

Assuming we could design an AI coach to do this, how would that change our behaviors? How would each of us excel beyond where we were today? What would we be able to accomplish with a personal coach. A customized mentor. The ideal business partner...

This guides details Stage 2 of our founder-driven sales framework. Be sure to see the overview to check your assumptions and use Stage 1, Prospects to Problems, before you try these techniques.

With a bunch of validated problem hypotheses in hand, it’s time to go build!

Hold up, wait a minute. Building is still one of the most expensive activities we can do with the lowest confidence returns. Before we get there, let’s be sure we understand what customers are willing to spend their extremely limited time and money on, via pilots in 3 steps.

1. Validate the value

2. Validate an offer

3. Convert the offer into a pilot

1. Validate the value

There’s a ton of ways to create value propositions and even more ways to test them. To name a few, Strategyzer has a Value Proposition Canvas, Pragmatic Institute uses Product Positioning, the Lean Product Playbook uses benefits lists and competitor tables, The Unicorn Within uses Storyboards, Sprint uses Prototypes, Unbiased Insights uses Product Concepts, and Lean Customer Development uses MVPs. 

Write 3 value proposition statements

After years of delivering MVPs in record time and trying everything but, we’ve honed in on one of the simplest versions—a few words on a slide, following this template:

Here’s a consumer example that is easily understood:

Uber is for city dwellers with smart phones
who need a reliable way to get a taxi
that enables the user to hail a cab, feel in control, and pay for it automatically.

Here’s a B2B business example to Flywheel, an Uber platform for Taxi Companies:

Flywheel is for taxi dispatchers
who need an easy way to provide reliable and convenient service to customers
that tracks all active drivers, allows customers to book a cab online, and collects payments automatically.

This is absolutely an art and is very difficult to master. No matter where you are, try some of these techniques to improve each value proposition:

1. Write, rewrite, remove words, make it simple, be clear and straight forward.

2. Focus on the WHAT. A value proposition doesn’t tell you HOW.

3. Leave the clever marketing writing for later. Clear wins out over clever, every time.

4. Draw a picture. Act it out. The simpler the better

5. Write out what it DOESN’T solve. Clarity is about saying NO to everything else.

6. How do your benefits compare to the current competition?

Produce at least 3 value propositions. If you produce more than 3, keep others for later, but winnow down the selection based on your previous interviews and next conversations you have lined up. No need to be statistically significant here, you can and should mix choose the 3 most appropriate for each interview.

The value proposition you’ve created is yet another hypothesis—it’s the first time those words have been combined in that specific way. That means it’s time to test it again. But before we do that, let’s prioritize who we should talk to.

Estimate what your conversations will think

Go through all your interviews when you were searching for a problem.

Listen to the recordings and read the transcripts to see if this person has a problem that your value proposition will solve.

Based on what they said last time, estimate their answer to this question:

On a scale of 0 to 10, how important is this product to my role at company today?

If it’s above a 7, that is an 8, 9, or 10, add them to your prioritized list to contact.

Now you need to validate it with the market.

Talk to customers, again

Reach out to those contacts that you assumed to be above a 7 for a new call, using some easy language:

Hey <Name>, thanks again for sharing your experience with me! I’ve now talked with X people and come up with a lot of potential ideas, a few I think might be really applicable to you. I’d love to get your perspective on them.

You’ll also want to contact some net-new conversations. Using the same outreach flow as last time, you might adjust the template a few ways:

I have some ideas, but don’t know if it’s valuable… can we connect?

Get their quantitative and qualitative feedback

All our other protocols still apply to this second call. You’ll want to get permission to record the conversation, confirm any qualifications you’ve made of them, and thank them for their time.

During this call, you’ll share 3 of the value propositions with the language exactly the way you’ve written it down. Putting each on a single slide in big font and sharing your screen is a great way to stay consistent.

Ask them to read it out loud. You’ll hear if they struggle with any specific words or the wording is too complex

Ask them to describe it to a friend. See if they use the exact same words or different words and if they understood your intention.

For each value, you can then ask:

If you had $100 to spend on everything you need at work, how much would you spend on this solution out of that $100?

Ask them to explain their thinking and why they gave that number.

Go on to the next value and repeat that question.

After you ask all 3, reflect their numbers back to them and see if they still feel this way or want to change any of their numbers.

Then ask them to rank the value propositions together. I like to put up a visual scale of 0-10 and ask them to place the 3 value propositions on that scale, using the product names, with the question:

On a scale of 0 to 10, how important is this product to your role at company today?

Follow up with the usual qualitative “Why?” to get context on their rankings. Discussion here will continue to build rapport as well as provide additional insights that you can continue to ideate on.

With this foundational quantitative data, you can then ask them:

Are there any other products that are more important to you today.

Return to your consistent wrap-up protocol here, thanking them for their time, asking if you should’ve asked them anything else and a desired call to action.

Your next call to action depends on how they responded to the value propositions:

1. If they’re interested in one value more than another (10 vs 5 vs 1):
   ask them to schedule another call to review the pilot offer you’re putting together. If you’re not sure this is the right value proposition, just keep it general—you’re collecting interest for the first pilot of this value and if they’d like to be included.

2. If they rated things poorly or aren’t keen on these options that you really still want to pursue:
   Share that you have more learning to do. Let them know you’ll reach out, if it’s OK, if you make progress on other ideas or other values based on their feedback.

Either way, focus this next ask as a reach out to them again, not any connections that they might make. You’re starting to build a relationship with them and will remain top of mind to them—they’ll start to connect you with others if it makes sense.

Analyze your results

Take the data from all these conversations and compare them to what you predicted. You’re looking for high signal that one of these value propositions is standing out among the others. This comparison delta will help you filter through all positives and all negative feedback, as you ultimately want an outlier.

Ideally you’ll have 5 potential people to pilot this with (think of this as the minimum requirement to get to your first pilot, with a 20% success conversion rate).

2. Validate an offer

With the outlier value proposition in hand, it’s time to create an offer to pilot.

Plan an offer

Starting with the end value that you just validated, go through the user journey you think happens leading up to that value. Rely on the contextual information you learned from these qualitative calls, but don’t sweat too much—it’s just a thinking framework for the first version. Map out the user story map of how a manual pilot would operate. Think about all the nitty gritty details to the point where you know exactly how each step will work.

You can document the really expensive stuff you want to do, but ideally think of manual, cheaper slices—ways that you can deliver the same value without building anything—you want to maximize value and minimize investment, even if it won’t scale as-is.

For example, in order for Uber to test their feasibility pilot, they gave a small app to 5 of their friends in San Francisco with a map and a single button. Every time their friend pressed the button, they would get an alert and their friends’ location. Once they got the alert, they would call a taxi company that they would pay to send a cab out to pick them up and take them to where they wanted to go. This is a long way off from the first release of the Uber app.

If you want to test feasibility, make the offer free. This validates that you are able to connect with customers that have this problem and are willing to give you their time, data, or other resources in order to see it done.

If you want to test viability, make the offer commensurate with the value.

What’s commensurate value? Here’s a consultant’s answer: it depends. Typically products are anchored against 3 variables: Time, Cost, or Quality. If you’re anchoring on cost, go 10x cheaper than the alternatives. If you’re anchoring on time, estimate a price for an intern or new college graduate to do the work manually. If you’re anchoring on quality, match the price of a competitor.

No matter how you determine initial value, you’ll need to round out the offer a bit:

1. Provide a 50% “pilot discount” for your first round of pilots (3 or so)

2. Come up with a small and a large alternative. What would an offer 2x smaller look like. What would an offer 2x bigger look like?

3. Come up with a list of requirements that you need from them, with the hardest call-outs first. You want them to tell you NO.

Note: With these 3 offers in hand, do some quick financial modeling. Gut check to make sure IF you were to build this offer into a product it would generate returns at some basic scale.

Document the offer as a pilot

Once you’re comfortable with how it’ll work and what the offer is, combine this into a pilot deck that includes:

1. The problem you’re solving

2. The pilot value you’re delivering

3. The general pilot process

4. Any call-out requirements you have of the pilot goers

5. The 3 different versions of the pilot

6. The 50% off deal

Now lets go get punched.

3. Convert them to a pilot

Your plans are beautiful. Your pitch is perfect… to learn from. So let’s go learn what “the market” thinks.

Reach back out to people you’ve been building relationships with that validated the value they want to see—anyone that rated a value proposition you’ve codified into a pilot a 9 or 10.

On this 45-minute call, you’ll want to walk through the pitch deck and ask for their feedback. A good sign is when they’re saying YES, AND in response to things. 

Either way, ask them to co-design the pilot with you. What would they change? Ask them their requirements if they were to run this pilot.

Showcase your flexibility, speed, and ability to listen—3 powerful benefits this pilot will bring them that competitors with static plans and large teams can’t. These are your unfair competitive advantage.

If they haven’t already, directly ask them what it’ll take to work on this pilot together.

See who else they need to introduce you to at their company to get this pilot started.

The easy way, or the better way

The easiest way to get a pilot with a customer is to say yes to whatever they ask for, including all the ridiculous demands or requirements they have.

• HIPAA compliance? Yes.

• Business Insurance? Absolutely.

• Do they want to pay with NET90 terms? Of course.

• 50 page legal contract? Signed.

The better way to get the right pilot is to ask more questions. Discuss each of their objections and questions.

• Why is that a requirement?

• What about that is absolutely a requirement or is there any way we can work around that?

• When was the last time your procurement department made an exception to this?

• Well that level of insurance costs X, would you be willing to pay extra for that?

• What if we were to put that on a credit card or bill it differently, what would change?

• Does time zones matter?

• Can we deliver that faster or slower?

• What does this clause in the contract mean to you, exactly?

• Can I talk to Jim in IT or Jennifer in Legal directly?

It doesn’t hurt to ask your potential pilots these questions—in fact, it hurts you not to ask these questions.

🎯 It adds friction, but your point here is not to create the most frictionless experience possible. Your goal is to learn—to understand their perspective, process, and motivations—to maximize your learning.

Remember, this pilot isn’t a long-term solution that you’ll tell anyone else the details about. It’s just for you and them. It’s OK if none of this is ideal.

Only two things matter:

1. That you deliver value

2. That you keep the relationship healthy

These are the core results that you’ll need for the Stage 3, Pilots to Proof, on how to move from one-off pilots to selling a scalable product.

This guide details Stage 1 of our founder-driven sales framework. Be sure to see the overview and check your assumptions before you use this.

Let’s turn your network of contacts into market validated problems you can and should solve.

1. Find your target potential customers on LinkedIn

2. Connect with them

3. Discovering their problems

1. Find your target potential customers on LinkedIn

Using LinkedIn Sales Navigator is a must—create a list of “prospects” using crucial filters to increase your response success and uses your 60 connection requests/week wisely. I always create a target list of companies (“accounts”) and add that as a filter. This ANDs the people’s profiles with the companies they work for (instead of ORs, based on the LinkedIn filter). Active on LinkedIn in 30 days, person’s location, 2nd connection, work experience, current title (and time in role) are also table stakes for a well-focused target.

🔎 Look at the number of prospects. If this number is lower than 200, you probably need to expand. If it’s above 1,000, you probably need to refine down more.

2. Connect with your prospects on LinkedIn

So now, with each of those prospects, you’ll send a connection request with a custom message. This puts  your requests in their main app—and a custom message helps you stand out. Connecting adds immediate value to both sides of the connection—opening your network and profile to them. Some people don’t accept connection requests from people they don’t know, but then they’ll be much harder to sell—so let’s not worry about those privacy-focused individuals.

Here’s a great message template that works ~80% of the time for me:

“Hey <First Name>, I’m working on <problem/struggle> and would love to learn from your experience as a <Title> at <Company>. Can we connect?”

This keeps the ask immediate (there are yes/no buttons in the app) and small.

Note: If you’re getting a blocker requiring an email address, it’s because they get too many requests and have turned on this feature. You can try to find their work email via hunter.io, or just move on to the next person, as most people’s LinkedIn profiles are built off of their personal email.

Scheduling a call with your new connection

As soon as they accept your request on LinkedIn, a few things will happen:

1. You’ll get a new message thread with your custom message.

2. They might respond—typically with a question.

3. They might not respond, leaving it open for you to continue the conversation.

They can respond to you without accepting the connection request. That shouldn’t change any of the workflow—since we’re just keeping it in chat and they still “connected” with you by talking to you.

🎯 Your goal here is to get them on a zoom call for a qualitative discovery call. I usually just send 1-2 messages, as I don’t like to pester people, but I’ve seen different techniques work here.

Here’s a great message template that fits my personality:

“Hey <First Name>, thanks for connecting. <Answer any of their questions/references>. As I mentioned, I’m exploring  <problem/struggle> and would love to learn from your perspective since you’ve been in the space for so long. Are you open to a 30min zoom?”

At this point, ideally they say sure. If not, you may have to get through an objection or two. Here’s some common responses that keep the conversation going:

“I’m not selling anything—just looking to learn to see if I’m even working on a painful problem or not.”

“I’d really like to verbally hear your responses to my questions, but here’s a few of the things I’m trying to learn: <hypothesis 1, hypothesis 2>. Would a phone call work?”

If they say yes, send:

“Great! If it’s easy for you, here are some times that work for me: <calendly link>. Otherwise, happy to work around your schedule.”

🗓️ Keep times as open as possible on your Calendly (5am? Sure! 10pm? Absolutely.) and have an option for them as early as tomorrow—anything sooner than that doesn’t give you time to prepare—or 2 weeks from now—but if they’re not available in the next 2 weeks, it’s probably not important enough for them.

3. Discovering their problems

Remember that sales objection we solved? It’s true. Your goal on this call is NOT to sell them something. Ideally, you don’t have something to even sell them now. Your goal is to learn from them. Ultimately, you’ll want to test a hypothesis with them. The earliest hypothesis is do they have a problem that’s cursing them.

CURSEs are: Crucial, Ubiquitous, Recurring, Specific, and Extreme problems.

This handy acronym checks to see if a problem is:

1. Crucial enough to the organization that real money is being spent on it
   (have they already spent money trying to solve it?)

2. Ubiquitous enough that there is a real market size to solving this problem
   (everyone that you talk to, within a segment, has this problem)

3. Recurring enough that this will stay top of mind for them
   (you’ll retain them as users)

4. Specific enough that they don’t feel like they have a great solution
   (there’s room for a pull)

5. Extreme enough that they’re willing to do something about it
   (they’re pushed to find a solution)

There’s many different kinds of ways to get at these hypotheses in early discovery calls. Mach49’s Linda Yates published a great pain-script in her book, The Unicorn Within, but I personally like to run a Jobs to be Done (JTBD.org) discovery script here—which is a guided storytelling conversation that can be used to validate or invalidate your current hypothesis. Getting them to frame their experiences as a story is the best proxy for experiencing it and seeing what they’re thinking about/remember about that experience, without being biased about their hopes and desires or your experiences and ideas.

RECORD THESE CALLS
Just for your own note-taking purposes, but you’ll want to revisit these calls dozens if not hundreds of times in the future. Listening to the exact terminology someone uses, their tone of voice and even their body language. I personally like Fathom as my current zoom and google meet recorder of choice. You’ll want to get permission, and always be willing to turn it off, but 90% of people say yes if asked politely. (I like to send them the recording afterwards as well, as it gives them something of value, even if tiny).

At the end of the call, don’t forget next steps. I like to always ask 3 questions:

1. “Were there any questions I should’ve asked you, but didn’t?”
   (you’ll be surprised what they’ll say sometimes—but most of the time this is a recap of whatever is top of mind for them for the conversation)

2. “Can I reach back out as I make progress on this?”
   (I’ve never been told no here, but this primes the pump for a second conversation, which you need to get to a pilot)

3. “I’m trying to learn about <problem>. Is there anyone else that I should talk to? Maybe <NAME>?”
   To get a warm referral to someone else. Be sure to reiterate what you’re trying to learn and who you want to talk to—and bring up the name of a person they might’ve talked about on the call.

After the call, see if it aligns with other problems you’ve received from other calls. Focus in on their emotions/perspectives. It’s important that you don’t try to solve it—yet.

If you’ve done a handful (5-10) of these calls and haven’t found a CURSEd problem, pivot to a different person. A pivot can be a refining of your current stakeholder (maybe it’s only software developers in their first month of learning a new language) or a reframing (it’s actually the first time someone learns anything new, not just software developers).

When you’ve found a problem that meets the CURSE criteria, turn this into a list of hypotheses of problems for your back burner. You’ll want to come back to this step a few times, and it’s easier to create a list before you’ve moved on too far from these conversations.

While this is more of an art than a science, this format might help:

“We Believe That…. <person> has <problem> during <context>, which causes them to feel <emotion>. They currently do <bad solution>, which costs them <quantified expense>.

If you’re unsure of the others, just write down all the different emotions that were expressed related to this one problem in your conversations. Write down the impacts that this problem has. That’s a good start to keep a variety of words as a list of different problems.

Once you have found a CURSEd problem, it’s time to validate it, in it’s simplest form—the value proposition. Let’s move into Stage 2, or how we use these Problems to develop the right Pilot with our first paying customers.

Here’s a simple playbook that works incredibly well for your first sales, using your LinkedIn profile. Note that even though this is simple, it’s not easy. There’s a ton of nuance that is easily missed—and even well experienced founders miss. Even in our experience working with serial entrepreneurs, most founders try to outsource this work too soon—not learning the nuances of their market themselves.

In general, the framework runs in 3 main stages:

1️⃣ Prospects to problems (read here)

1. Find your target potential customers on LinkedIn

2. Connect with them

3. Discovering their problems

2️⃣ Problems to pilots (read here)

1. Validate your discovered value proposition

2. Validate a pilot offer

3. Convert them to a signed pilot

3️⃣ Pilots to proof (read here)

1. Deliver value

2. Deduce what works and destroy what doesn’t

3. Double down

⚠️ This doesn’t always work

There are a ton of assumptions baked into this framework. If your efforts aren’t aligned with these assumptions, trying this framework will likely waste your time.

Here are some assumptions to check, before you get started.

1. It assumes your target has a LinkedIn profile and is active on LinkedIn. B2B Sales to professions like Teachers and Nurses, who don’t use LinkedIn frequently or during active business hours, won’t work. Consider alternative channels where these personas may be active and struggling. Not sure what channel they use? Go shadow a friend to see what they do on a daily basis.

2. It assumes that you’re targeting roles within companies with a specific JTBD—if you’re looking for consumers based on some psychographic or JTBD based on their current state—LinkedIn won’t help you there. Consider Google Ads or Facebook/TikTok ad placements to catch their eyes.

3. It assumes that you’re connected into your target market. Don’t have secondary connections in the industry yet? Target their thought-leaders, recruiters, and sales staff. Those are the most active accounts on LinkedIn with the most connections.

4. It assumes that this role has problems you’re able to solve. Selling to nurses during COVID was impossible—it was simply all-encompassing for them to deal with discovery calls. That said, many other roles within healthcare companies were interested in talking—as they were all working remotely and looking for a lot of help.

5. It assumes that they’re interested in learning or being considered an expert. Some roles attract more introverts than extroverts. You’ll always find a few accountants that love to share their knowledge—but most don’t.

Linked in this post are the details behind each stage of this framework—we encourage you to try this on your next startup! If you’re running a studio and need assistance, check out our AI offering that helps you automate the mundane parts of this work.

At Tarka, we’re trying to solve the cofounder problem—that is that nothing can surpass the value that a cofounder provides your startup. There are a lot of venture building programs, like accelerators and incubators, that can help you create a startup—but none of them are better than the right cofounder.

But why is that the case? We think it’s because most programs offer a small subset of what a cofounder offers. Here are just a few things that a great cofounder will do for, and with, you.

A cofounder that challenges you

Ensure that your vision and execution are both thoroughly vetted and refined. Healthy debate leads to more robust strategies and solutions, making your work more resilient.A cofounder who respectfully challenges you can help you avoid blind spots and encourage better thinking through diverse perspectives.

A cofounder that complements you

No one has all the skills and expertise needed to build a successful startup. A complementary cofounder fills in your knowledge gaps and strengthen the overall team. For instance, if you're a technical expert, having a cofounder with strong marketing or sales skills will elevate your work. Their different perspective and abilities can help your startup navigate challenges and seize opportunities you might otherwise miss.

A cofounder that adapts with you

The startup journey is unpredictable, filled with pivots, market shifts, and unexpected challenges. Having a cofounder who can adapt and evolve alongside you is crucial. If they follow too closely to a rigid process, then they’ll fail to learn. They should be resilient, open to change, and willing to learn from failures. This adaptability ensures that your startup remains agile and responsive to changing market conditions.

A cofounder that you can delegate to

As a founder, you can't do it all, and you ultimately need people, processes, and systems to take things off your plate. Delegating tasks and responsibilities is essential for scaling your business. A cofounder you can trust to take on critical aspects of the company allows you to focus on your core strengths and strategic decision-making. This division of labor enhances efficiency and enables your startup to grow faster.

A cofounder that prompts you

A cofounder should serve as a source of motivation and accountability. They can prompt you to stay focused on your goals and maintain a strong work ethic. When you hit rough patches or face setbacks, a supportive cofounder can provide the encouragement needed to keep pushing forward. Their presence can be a powerful motivator to help you stay on track.

A cofounder that works as hard as you

Building a startup is a demanding endeavor that often requires long hours and unwavering dedication. Having a cofounder who matches your level of commitment is essential.

A cofounder that’s invested with you

Your cofounder should share incentives and outcomes with you. It creates a stronger resilience sense of purpose that can propel your startup forward, even when faced with adversity. This commitment is evident to employees, investors, and customers, inspiring confidence and trust in your venture.

Without these traits, a partnership is not fully formed and will always be lesser than the alternative. Can anything except a cofounder provide all of these things? Let’s find out…