Gabriel AbreuWhich of your new signups will become your new best clients?
Which of your new signups look most like your best customers? By reading this article, you'll learn a simple 3-table way to rank them with AI.
I've been thinking about a specific problem lately, and I want to share how I'd solve it. Maybe it hits home for some of y'all as well.
But first, as an exercise, I invite y'all to think of your 3 best clients. Pause for a sec here while you think about them.
Okay, now let's compare last month's signups and ask: which ones are most like them?
Chances are you can name only one or two & the rest will probably blur together, right? And that blur makes us a lil bit nervous, right?
I feel you. Esp. because one of your next great clients is probably in there. They just haven't raised their hand yet (and they probably won't ever).
How AI can help you out
Before I show you how to fix that, let's first understand how AI works. In short, AI is v v good at grouping things that are alike.
A cool & classic example is: King - Man + Woman = Queen. At the end of the day, it's probability all the way. Math speaking for itself.
Sooo, hell yeah, you can def. use the same math to check, verify, and spot new signups that look like the clients you want more of.
And btw, we're living in a new era where A LOT of your best potential customers will first show up through an AI agent using your product for them.
An AI agent could find your API docs, test your integration in a sandbox, and start building it into a workflow before anyone on the buyer’s team even books a call ++ teh clients you want usually won't take the initiative to book a call with you. They'll just show up and start building around your infra, which is fine as long as you can adjust to that.
The fix is pretty simple
You're building each client's list based on what each client does, yeah?
So, that list of numbers is a vector. Clients that act alike should get vectors that sit close together, got it?
And when you average the vectors of your 3 best clients, you get a reference point. X marks the spot.
Yup, you get a reference point (X) that a similarity search can use as a reference to measure how far each signup is from that mark.
So, if you avg. vectors get a point that captures what those clients have in common, it'll be making your life WAY easier to compare new signups against it.
The 3-table build
To build that, you'll need a matching system, which is the thing that'll pick up what you should pay attention to. And you can build it with a Postgres database and just 3 simple tables.
- Signals table: new signal arrives > an agent finds, classifies & adds whatever context it can ++ checks if it fits before you can see it.
- Evals table: It'll evaluate and store what your team sees based on your new client's behavior. It'll track w/e your clients are doing and score that accordingly.
- Enrichment table: It combines your client's product usage, internal & external data to create one profile, then it'll map & show you how close that guy is from your "X" mark.
Connect those pieces & tam-dam > you'll be able to spot the right fit and builder, understand what they really need, and help them get unstuck wherever they're in your funnel.
Aaand, ofc, it won't overlap with current clients, or anyone who asked you not to reach out, those'll get skipped & this way you'll not be spamming the wrong people, but actually talking to the right folks at the right time.
TL:DR
In short, the pills that solve your pain are most likely a Postgres database with 3 tables.
- signals & tracking
- accounts & evals
- people & self-improvement
So, when an account gets a signal, an agent finds their builder, enriches them, verifies, and checks the timing. Then a similarity search ranks each signup by how close they sit to your 3 best clients, aka X marks the spot.+++ a cron job (or grok bot) that fills signals every now & that's it, folks.
That's the whole thing. I'd highly recommend y'all try to build something like this if you want a shortlist of names to look into & to stop spamming/losing time with folks who are long gone.