Most of the founders I work with are smart, technical, and starting their go-to-market from a blank canvas. I’ve spent years telling them roughly the same things. Find the number your customer actually cares about. Don’t build on a proxy for it. Talk to people before you build.
This year I co-founded a company and got to find out how much of that I could follow with my own money and my own weekends on the line.
The thing we built
My co-founder is my wife, Dayna. She’s an independent college counselor, and families hire her for the whole arc — planning from freshman year, building the list, the essays, the applications, the aid strategy, the decision in April. It’s a relationship that runs for years, and that’s what she’s paid for.
Inside that engagement there’s one specific piece. Finding a student schools that are genuinely good matches takes her two or three hours. Then another hour sits with the parents on what each of those schools would actually cost them — because the published price is close to fiction, and the generous schools are rarely the ones you’d guess.
That piece is not the cheap part of her job. It’s the foundational part: nothing else starts until a family knows which schools are worth their attention and what they’d really pay. Get it wrong and every hour after it is spent on the wrong list.
And it’s the piece that families who never hire anyone are left to do alone. They go back to guessing, and some cross off schools they could have afforded.
So we built that piece and made it free. Aiko takes a student’s grades or test scores and what the family can realistically spend, searches about 2,500 four-year US institutions on federal data — IPEDS, Clery, the College Scorecard — and returns a list sorted into Reach, Match and Likely, with what each school would likely cost after aid. It’s a free college list builder — in beta, and rough in places.
It is not a replacement for what Dayna does. It’s the first foundation, for people who were never going to get one.
Three things I’d tell a founder now that I’ve been on the receiving end of my own advice.
1. “We don’t have it” and “it doesn’t exist” are different problems
Ask a hundred families what decides college admission odds and most will say GPA.
So we went looking for GPA, found almost none of it, and I drew the wrong conclusion.
Of the 2,561 four-year institutions in our database, 17 had an average high-school GPA — and one of those weights it, which makes it incomparable to the other sixteen. Test scores were better and still only covered about 39%.
I read that as a fact about the world. It is a fact about my database.
Colleges do report GPA. The Common Data Set carries an average in section C12 and a band-by-band distribution in C11, and several hundred colleges publish one each year — a minority of the 2,500-odd, but far more than the 17 we had. It arrives as a PDF per college, which is why we hadn’t ingested it: the pipeline was an open ticket, not an impossibility.
And it is very much for sale. My co-founder has had GPA averages and ranges on screen the whole time, in College Planner Pro, which licenses the Common Data Set from Peterson’s. Counselors can buy this. Families mostly can’t, and the better-resourced high schools layer Naviance or Scoir on top for their own applicants’ scattergrams.
So the real finding wasn’t “this number doesn’t exist.” It was “this number is fenced off, and we hadn’t paid or built to get over the fence.” Those lead to completely different roadmaps.
The transferable version: confirm the input exists at the coverage you’d need before you commit a quarter to a feature. Then confirm what “we don’t have it” actually means — nobody has it, it’s licensed, or you haven’t fetched it yet. Those are three different problems and from inside your own schema they look identical. I reported the third as the first, in public, which is how you end up correcting a post about data rigour.
2. A domain-expert co-founder is not market research
This is the one I’d least have predicted, and the reason this post isn’t titled after the data.
Founding with the practitioner feels like it solves discovery. Dayna has done this work hundreds of times. Every question I had, she could answer immediately and correctly. So I stopped asking anyone else, for months, without ever deciding to.
The problem is that she is one expert with one client population — families who found her, could afford a multi-year engagement, and had already decided professional help was worth paying for. Aiko’s users are mostly the opposite: families who never got that far. Those two groups want overlapping but genuinely different things, and the differences are invisible from inside the practice. A family who has committed to a counselor asks different questions than a family who has never spoken to one and doesn’t know what to ask.
Her expertise told us what good work looks like. It could not tell us what a stranger needs in the first ninety seconds, and I spent months treating the first answer as though it were the second.
The transferable version: an expert co-founder is an enormous advantage at solution quality and a hidden liability at problem definition. Their sample is their clients. If your market is the people who aren’t their clients, you still have all of your discovery ahead of you — and you will feel like you’ve done it, which is the dangerous part.
3. Saying what the product won’t do is the cheapest trust available
Aiko launches into a market saturated with AI tools, where the generic pitch is exhausted and the default assumption is that the thing is confidently making things up. So it’s explicit about three refusals:
- It won’t write your essay. It will help you plan one and pull a draft apart with you. The writing stays the student’s.
- It won’t predict an admission decision. “Match” means your GPA or test scores are in the range the school typically admits, where we have that figure. Where we don’t, it reflects how selective the school is. That’s where you stand, not a forecast — anything stronger is a lie with a number attached.
- It won’t quote a price it can’t stand behind. Costs start as averages and are labeled as averages until a family enters its own numbers.
Every one of those is a feature someone will request. Every one is also the reason a parent might believe the rest of it. In this category, the constraint is the differentiator.
The confession
I tell founders that go-to-market is an iterative learning engine rather than a launch plan. Then I built for considerably longer than I’d have let any client get away with, while my co-founder had a waitlist of exactly the people I needed to talk to.
I don’t think the product would be better if we’d built less. I think it would be the same product, sooner — which is the entire argument I make for a living.
If you have a junior or senior at home
Aiko is free. No card, no trial. It’s in beta and I’d rather fifty families tell us what’s broken than five thousand sign up and bounce. If you try it, tell me what annoyed you. That’s worth more to me than a compliment.