What it does
You see it today. Not in next month's report.
Four weeks of paying for a campaign that broke three weeks ago: that is the real price of looking too late. Stevin sees it the same day, tells you why research says it matters, and records what you did about it.
Three things. Today already.
No roadmap, no coming soon. This runs, and you can check it.
Signals on your own data
Checked daily for drift in your ads, your budgets and your measurement. You hear about it with the reason attached, not in next month's report.
See who advertises in your market
From the public ad registers of Google and Meta we track who advertises in your market and with what. Public data, so this works before you are a client.
A logbook that stays
Every change with a name, a moment and a reason. Including the decisions that turned out wrong, with what we learned. Whoever starts here next year reads up.
Your own screen
Just ask. On your own numbers.
This is the client portal. Not a report you have to read, but a question you ask. The answer comes from your campaigns, measurement included, and it changes nothing. Click a question.
Stevin Assistant
An AI assistant, not your consultant.
Hello!
Ask about your calls, enquiries and campaigns in plain language.
Example of the client portal on app.stevin.ai, with made-up figures from an installation company. The buttons under an answer are illustration here, not function.
And this it deliberately does not do.
Because a system allowed to do everything by itself is exactly the system you can no longer check.
It changes nothing in your campaigns
Not a single instruction runs from our system to Google or Meta. It flags and prepares a proposal; a person decides and executes. That is a choice, not a phase.
It does not connect to everything up front
Ad platforms, analytics and your shop we read today. Your point of sale, your ERP or your DMS we connect when your question calls for it, not because it looks good on a list. If there is demand, we connect it.
What has to be right before AI can do anything for you.
Not out of tidiness, but because each step makes the next one possible. Skip one and everything above it is guesswork with more compute.
Grip first
01
Ownership
The accounts are in your company name and you decide who gets access. Without this you cannot switch executor tomorrow, however good the rest is.
02
Measurement
A conversion is a real enquiry, not a button click. If your tracking fires at the wrong moment, every number above it is unusable.
03
One source
One place where the numbers meet, instead of four dashboards contradicting each other. Otherwise you argue about who is right instead of what to do.
04
Memory
What was tried, why, and what came of it. Without it every new party starts from zero and you pay twice for the same lesson.
Only then this
05
Automation
Only here. What happens the same way four times can do itself. Before this you are guessing which work to automate.
06
AI
On top, not underneath. With the four steps above in place this is a flywheel. Without them it is an expensive way to go the wrong way faster.
We always start at the bottom. Not because it is the fun part, but because the rest rests on it.
Ask why. There is a source underneath.
Underneath sits a knowledge layer of 105 principles, built from published research. They weigh in on every analysis. Three of them below, as they sit in the system.
Growth comes mostly from new buyers, not from making existing ones more loyal.
Brands of similar size have near identical loyalty. Those that grow, grow through acquisition.
Ehrenberg-Bass
Does not mean retention is irrelevant, and it does not hold for brands already serving nearly their whole category.
Brand building and activation belong together, around 60/40.
Isolated short term campaigns do not add up to long term growth.
IPA Databank, Binet and Field
Does not mean 60/40 is a law. It is a calibration point that shifts with category, growth stage, budget and purchase frequency. And not that every small business budget belongs in expensive brand channels.
Check your measurement before judging anything else.
An event firing at the wrong moment invalidates any diagnosis. Classics: a purchase event firing on page load, or UTM tags lost along the way.
Core principle, always weighed
Does not mean you cannot advertise without perfect measurement. It means you do not read your numbers blindly.
We do not decide which model you use. Work with STDC, with See Think Do Care or with 5A, and the advice follows your phases and their measures. The model is yours; what we add is what demonstrably works per phase.
Look first, decide after.
The diagnosis runs on your own numbers and puts in black and white where you stand. A proposal comes after that.
Start the diagnosis