How to find your ICP with no customers (a 13-criteria scorecard)

With no customers, your ICP (ideal customer profile) is a guess, and that's fine. What helps is writing the guess down in a way you can compare, so you can put segments side by side and then check the best one quickly.

I score every segment I'm considering on 13 criteria, weighted, out of 120. Below is the scorecard, and the actual scores I gave the segments for one of my products.

Step 1: write down what your product does best

Before scoring anyone, list what your product is good at and what each of those says about the customer.

That's my table for Mailyond, an inbox for all your domains. Yours will look different, but it keeps the scoring tied to what you can actually deliver.

Step 2: score each segment on 13 criteria

Rate each segment from 1 to 5 on every criterion, multiply by the weight, and add it up. The weights sum to 24, so the maximum is 120.

Pain: does it hurt?

Fit: can your product solve it?

Business: can they buy?

Value: is it worth it?

How I read the total:

• 95 or more is a strong fit, start there
• 75 to 94 is a good fit and worth testing
• 55 to 74 is moderate, only try it if the top ones fail
• under 55, move on

A worked example with real scores

For Mailyond I scored six segments. Three of them show how this plays out:

Agencies had the top score, and I still started with solo founders at 79.

The reason is accessibility, which the total can hide. Solo founders are the only group I can actually reach, every pain signal I had came from them, and I'm one of them myself. Agencies scored higher but they come with expectations (uptime guarantees, maybe white-label) that I'd want to check with a few of them before building anything for them.

So use the total to rank, then read the accessibility row again before you pick.

Price is a filter on who shows up

Your price also decides who contacts you. Customers on a low price tend to send more support requests per dollar, while the ones paying more usually just want it to work.

When two segments score close on pain, weight the one that pays more. And do the math at scale before committing: if 100 customers at $9/month send 200 support tickets a month, that's more of a job than a business.

Qualify on stage as well as fit

A segment can match on paper and still not have the problem yet. Founders still building their first product don't need a support inbox, because nobody is emailing them. They'll say it looks useful and never sign up.

So add the stage to the ICP, meaning the moment the pain starts. For Mailyond it's "shipped products with real users", because no users means no support email.

Step 3: check the guess fast

The scorecard is still educated guessing, and my own scores show it. Mailyond's segments scored well and it still had 0 paying customers two months after launch. Scoring ranks the segments, but it can't tell you whether anyone in them will switch from what they use today.

To check the top segment:

1. Look at who your competitors show in their case studies, because that's who pays them.
2. Have about 10 conversations inside the segment, following The Mom Test (ask what they do today, not whether they like your idea), and ask about the switch as well as the pain.
3. Send the first 20 messages by hand, and if nobody replies, debug the outreach in order before you change the copy.

If you'd rather not do the weighting in a spreadsheet, the free client qualification tool runs a shorter 6-criteria version in the browser, and you can add your own criteria.

The full 13-criteria version is part of the Distribution Framework, the playbook I use to find customers for my own products.