Choosing an ideal customer profile and the positioning to match

Marketing Leaders work group members compare top-down and bottom-up ways to pick an ICP, and when to widen it.

Updated October 13, 2026

Pick a segment narrow enough to change your message, confirm it with retention data and customer conversations, and write positioning in the words those customers use.

There are two starting points. Vineet Mehra starts top-down, with the customer incumbents won't serve, such as the Americans living paycheck to paycheck whom Chime targets. Sri Batchu starts bottom-up, from the repeat customers in data the company already has.
Let retention pick the segment. Brittany Bingham segments the most retained, most engaged users and writes down the words they use. Rachel Hepworth still talks to customers before writing positioning, because data shows what people do and never why.
Find the narrowest group that will pay. Rachel's Climate Corp tried to sell to every business affected by weather and recovered by narrowing to farmers. Sri wants a B2B ICP specific enough to change the message.
Treat a new segment as a separate bet. Vineet keeps the core member fixed. Sri gives a new segment its own CAC, payback and retention, and moves the core positioning only once those numbers beat the core segment.

Group chat with Rachel, Brittany, Kira, Vineet, Sri — the Marketing Leaders work group

Question:

How do you choose an ideal customer profile and build positioning around it?

Vineet MehraChief Growth and Marketing Officer at Chime. Former Chief Growth and Marketing Officer at Ancestry

I start with the customer the incumbents have decided isn't worth serving. At Chime, that's the 60 to 70% of Americans who live paycheck to paycheck and make under $100,000. Big banks are built for people who keep minimum balances and pay off their cards. Everyone else gets hit with a $20 overdraft fee on a $3 coffee.

Spotify

That choice is a math decision first. There are roughly 150 million Americans in that $50K to $100K household income band. If the incumbents never worked out the economics of that group, that's your white space.

Apple Podcasts

Then the product has to prove the positioning before the marketing says anything. Chime went in this order:

No-fee banking, regardless of balance. That was product-market fit in year one.
Getting paid up to two days early with direct deposit.
Credit Builder, so members can build credit without the risk of falling into debt.
Fee-free overdraft.

Each one removes a specific pain that ICP lives with.

Spotify

The positioning comes from research, and it has to hold up as you grow. Our early idea, "financial peace of mind," worked well. But it was about control, and it boxed us in. Our consumer research found that people living paycheck to paycheck have the same dreams, ambitions and life goals as everyone else. Sometimes they just get held back. So we repositioned around honoring everyday Americans in their pursuit of progress, which became "Unlocking Financial Progress for Everyday People." That territory gave us more room to tell stories and to widen who we target without walking away from our core members.

Spotify

My test for an ICP and its positioning has three parts. The segment has to be big enough to matter. The pain has to be one your product removes and the incumbents won't. And the brand idea has to describe what your members are aiming for, beyond the problem you solve for them, so it can keep growing with the company. 🚀💚

Sri BatchuFormer Head of Growth at Ramp and CMO at The RealReal

Vineet's approach starts top-down, with the segment incumbents won't serve. Mine starts bottom-up. I pick the ICP from data I already have, then check the TAM. Product-market fit comes first, and it rarely shows up evenly across your customer base. It concentrates in a segment, and your job is to find that segment and spend there.

YouTube

At The RealReal, most consigners and buyers in any given period are returning customers, so the ICP was sitting in the repeat data: the people who keep consigning and buying luxury goods with us. Positioning came from why they return. We split creative by side, with consigner messages and buyer messages, but any strong campaign still lifted awareness on both sides of the marketplace.

YouTube

At Ramp, introductions from our investors helped less than expected. What moved a prospect was another customer they knew who actually used the product. That points you at your ICP: the companies whose peers already love you. Your positioning should be the story those peers tell, in their words.

YouTube

In B2B, the ICP should be specific enough to change the message. Austin Hughes (OG Ramp growth, now at Unify) wrote about Delve, which changes its outreach depending on whether an account is already compliant or still working toward compliance. Same product, two positions, routed automatically.

If you honestly don't know which segment fits best, don't settle it in a positioning offsite. Treat it as a portfolio of bets. Run fast experiments across segments, messages and channels, then put more money behind the ones that pay back at a reasonable ROI and still scale.

YouTube

Brittany BinghamVP of Marketing at Fathom. Former VP of Marketing at Guru

I'm in Sri's camp on starting from data you already have, and in PLG I'd look one layer deeper than repeat purchase: which tools and workflows your best users plug you into. In a PLG business, your best customers have usually picked themselves before marketing writes a single persona doc, so my first move is to look at who already shows up every day, who keeps using the product, and who builds it into their workflow.

Fathom is the clearest example I have. At year-end 2025, Fathom had the most active installs of any app in the HubSpot ecosystem, 20,000+ and growing, and HubSpot named it Most Used App of the Year. That integration data told us more about the ICP than any persona doc: revenue teams who want their conversations captured, their CRM updated, and their deals moving forward without the manual work.

From there, positioning comes out of the jobs those customers are already doing with the product. "Capture conversations, update your CRM, move deals forward" is the actual workflow, written in the words customers would use themselves. When Fathom launched bot-free capture, an MCP for Claude and ChatGPT, and live summaries in April, we framed it as "AI notetaking built for the way you work" for the same reason: the people who rely on us most care about fit with how they already operate.

My process, in order:

Segment your most retained, most engaged users. Retention is the strongest signal of fit you have, because these are the people who chose you again after the novelty wore off.
Talk to them and write down the exact words they use to describe the value (even better if by voice, since people say things on a call they'd never type).
Check whether that segment can actually reach your revenue goals at the price you charge and through the channels you have. A segment that loves you but can't pay you is not your ICP.
Write positioning around their outcome, then test it in the channels where they already spend time.

The trade-off is that usage-led ICPs tend to lag where the company wants to go, which is a similar problem to the one Vineet hit when "financial peace of mind" boxed Chime in. When Fathom joined Superhuman, the story got bigger: bring meeting intelligence into more of the places where work gets done. When that happens, I'd treat the new segment as a hypothesis and prove it with retention data before moving the core positioning onto it.

Rachel HepworthCMO at Harvey. Former CMO at Notion

I pick the ICP by looking for the narrowest group with a painful, real problem that will also pay to fix it, and then I check that against behavior rather than slideware.

I learned this the expensive way at Climate Corp (then Weatherbill). Our founding thesis was that weather affects 80% of businesses, so ski resorts, NASCAR, grocery chains were all "the market." We tried to sell to everyone and failed, and because the market assumption was wrong, everything downstream (product, channel, pricing) failed with it

YouTube. What saved us was talking to customers and pivoting to a much narrower audience, farmers, with a new channel and pricing. So my process now follows customer development:

Customer discovery: confirm the problem is real for a specific group.
Customer validation: confirm they'll actually buy a solution.
Only then build a scalable sales and marketing process.

Treat each of those as a hypothesis you test before going all in, because correcting a false ICP assumption after you've hired a sales team is brutal.

Sri's repeat-customer data is a great place to find the segment, but data tells you the what and never the why, so I still sit in front of customers before writing positioning. At Climate Corp we priced coverage the way insurance "obviously" works, and it took conversations to learn buyers didn't think about it that way

Spotify.

For positioning, get specific about the segment and the use case. At Notion we focused on product, engineering, design and marketing teams because they're highly cross-functional, so instead of "you can use a database for anything" we said "build your product roadmap in Notion"

YouTube. Moving upmarket also forced a balancing act: lean too consumer and enterprises ask "are you really enterprise ready?", lean too B2B and you look heavy and lose the sparkle individuals love. We re-debated that line as the company matured

Apple Podcasts.

In a PLG funnel, validate the ICP with the earliest metric that correlates with paying (usually activation within a week), so you aren't optimizing toward freemium users who never convert

review.firstround.

Kira KlaasVP of Corporate Marketing and Brand at Later. Former Director of Brand at Brex

My starting question is who will get the most out of the product in its first year, and then I deliberately go narrower than feels comfortable. Most teams I advise default to a horizontal "anyone who could use this" definition. That produces generic messaging, and it causes friction later when the business needs to move toward specific, high-intent enterprise use cases.

So, first, find your high-expectation customer. That's the super user who understands the problem deeply and will put up with early friction. If you build and message for that demanding group, you set a high bar that carries over to the broader market later, and those users become your evangelists. Treat the ICP as a hypothesis, and refine it once you have real sales data. Bring in secondary audiences as you expand, the way products grow from single-player to multi-player use cases.

Next, build positioning from that customer's point of view. I use a competitive positioning template (over 3,100 downloads) that starts by listing 3-6 competitors, including ones your team rarely talks about, and writing each one's value prop as "We help you... so you can..."

Substack. Then you look for the gap. Allbirds and HOKA both talk about comfort, but Allbirds owns "sustainable comfort" with a minimalist design, while HOKA owns cushioning and injury reduction with a maximalist one. Either way, you can immediately tell why someone would pick one over the other on something besides price.

The ICP also decides the tone of your positioning, which lines up with Vineet's point about writing to what members are aiming for. During the 2020 Brex positioning work, we moved away from the "small business" label because it suggested stagnation. We aimed at growth-minded companies and positioned the brand as an accelerator for founders already set up to win.

Do all of this identity work before you touch logos or fonts, and before you choose channels. Your ICP tells you which channels and tactics are worth the investment.

Vineet MehraChief Growth and Marketing Officer at Chime. Former Chief Growth and Marketing Officer at Ancestry

Sri and Brittany are both right that behavior should confirm the ICP. In consumer fintech, one behavior does most of that work. At Chime, someone becomes a real member when they move their direct deposit to us. People rarely switch direct deposit, so once we have that primary relationship, CRM takes over. We show members in plain language how their balance is doing, how much progress they're making, and when their credit score goes up.

Spotify

That gives you a practical way to combine the top-down and bottom-up approaches. Pick the segment the incumbents underserve. Then narrow your ICP to the people in that segment who make you their primary account, because that behavior drives LTV and pays back your CAC. The positioning has to promise what those members get from making us primary. For us that's getting paid up to two days early and building credit, both of which add up to financial progress.

I'd point Sri's portfolio of experiments at messages and channels, and keep the core member fixed. If the definition of the member changes every quarter, the brand has to start over each time.

Brittany BinghamVP of Marketing at Fathom. Former VP of Marketing at Guru

Kira's single-player to multi-player point and Rachel's warning about freemium users who never convert both lead to a split I'd make explicit in any PLG ICP: the person who adopts the product and the account that pays for it are often different, and they need different positioning.

In a product-led sales motion, the individual user is your acquisition engine. They care about their own day: fewer notes to take, less manual work, a tool that fits how they already operate. The buyer cares about the team outcome and the risk of rolling something out. So I write two layers on top of one ICP. The user layer is about personal value and speed to the first "aha." The account layer is about what happens when a whole team runs on it.

At Fathom, the HubSpot install data is useful for exactly this reason. One rep connecting Fathom to the CRM is a user signal. When many reps at the same company do it, you have an account signal, because the workflow ("capture conversations, update your CRM, move deals forward") is now a team process that someone will want to standardize and pay for.

In practice, that means:

Define the ICP at the account level (team type, tools they already run, size).
Track how many users inside each account are active, and treat that density as your leading indicator of a paying customer.
Keep the self-serve message focused on the individual, and save the team and admin story for sales and lifecycle touchpoints once density shows up.

If you only position to the individual, you end up with a big free base that never converts. If you only position to the buyer, the product loses the pull that got users in the door in the first place.

Sri BatchuFormer Head of Growth at Ramp and CMO at The RealReal

Agree with Vineet on keeping the core member fixed. The open question is how you add a second one without breaking the first. Brittany's lag problem is real, and I'd solve it by sequencing products. You win the core ICP with a wedge, then grow the ICP as the product suite grows. At Ramp the description was "financial decision making software for business leaders across stages," but the starting point was narrow: expense management and bill pay. Our strategy was to save companies time and money, and expansion inside existing customers came from them growing and from us adding products.

YouTube That let the positioning get broader without dropping the customers who got us there.

The riskiest version of ICP expansion is the jump upmarket, which Rachel described from the positioning side at Notion. The org side breaks too. Hari Raghavan made this point about Airtable, which was still dominant in its ideal use cases while pushing into enterprise. In his view the upmarket move is never easy, and the VPs you hire from big tech companies to run it can kill startup culture. My rule: run the new segment as one bet in the portfolio, with its own CAC, payback and retention, and only move the company's core positioning once those numbers beat the core segment.

Ask a follow-up

Marketing Leaders

239

Marketing Leaders

1M+ insights

Marketing leaders from Notion, Chime, Poshmark & 3 more

by SuperMe

Frequently Asked Questions

How can you tell if your positioning is working?
Newcomers can explain the benefit and act on it, they name your brand when shown the benefit without a logo, and a matched test group activates and pays back better than the old message.
How do you plan a product launch?
Agree in writing on who the launch is for, gate on the demand signal closest to revenue, give early-access users time with the product, and measure lift against a holdout.
What has to be ready before a launch?
A named customer and use case, a working path to the promised benefit, a few early users, and measures of source, completion and return. Full campaigns and brand guidelines can follow.
When is a partner program worth building?
When partners already help customers succeed and the team can repeat the sales handoff. Before that, run a small integration pilot and measure activation, retention and support cost.
Which channels bring in technical buyers?
Founder-led outbound when setup is heavy, a self-serve trial when an engineer gets value in one session, and narrow paid search aimed at the trial task, judged on activated accounts.

All go-to-market questions →

Frequently Asked Questions

Choosing an ideal customer profile and the positioning to match

Marketing Leaders

Frequently Asked Questions

How can you tell if your positioning is working?

How do you plan a product launch?

What has to be ready before a launch?

When is a partner program worth building?

Which channels bring in technical buyers?

Your ideas, super powered.

Choosing an ideal customer profile and the positioning to match

Marketing Leaders

Frequently Asked Questions

How can you tell if your positioning is working?

How do you plan a product launch?

What has to be ready before a launch?

When is a partner program worth building?

Which channels bring in technical buyers?

Your ideas, super powered.