Why we built Referenced

I run a small agency. Most of what we do is unglamorous: keep client sites fast, keep their rankings from sliding, report back every month with numbers that show the work is working. It's a good business, and until recently I understood the whole shape of it — where traffic came from, what moved it, what to say when a client asked "is this working?"

That last part is getting harder.

Watching the clicks change shape

Over the past couple of years I've watched AI answers start doing a piece of the job search used to do. Someone asks ChatGPT which accountant to use, or what the best tool is for something, and gets a name, sometimes two, and no further click. No results page, no scrolling past the sponsored links to find the honest option — just an answer, and whichever brands the model decided to say out loud.

For some of the queries that used to send a client traffic, that traffic isn't coming the way it used to. It isn't gone for every business — but the direction is clear enough that ignoring it isn't a serious option for anyone doing this work.

The question I can't honestly answer yet

Clients ask good questions. "Where do we rank for X" has a straightforward answer, because ranking is something you can check: run the search, read the position, done. Increasingly I'm getting a different question — "what does ChatGPT say about us?" — and I don't have the same kind of answer for it.

I can open ChatGPT myself and ask, and I'll get a response. But one response, from one session, on one day, tells you almost nothing about what a client's actual customers see when they ask the same thing next week. AI answers vary from one run to the next, and differ again across ChatGPT, Google's AI Overviews, and Perplexity. Given how many clients are going to start asking this question, "let me check and get back to you" won't hold up as an answer for long.

Built for enterprises, priced like it

The tools that measure this exist. I looked. What I found were platforms built for enterprise marketing teams, priced for enterprise budgets — serious money every month before you know if it matters for your size of business. That makes sense for a company with a marketing budget line for it. It makes no sense for the businesses I actually work with, or for an agency like mine that needs to check this across a dozen or more client sites, not one.

So there was a gap: real measurement, at a price a small business or a small agency could actually justify.

What we're building

Referenced is measurement first: ask the questions your customers actually ask, across the engines that matter, on a schedule, and turn the results into a score you can watch move. Not a magic fix for AI visibility — nobody's selling one of those honestly — but the same kind of grounded reporting I already give clients for search, extended to cover the thing now sitting next to it.

It's early. Pricing is built for the businesses I actually work with, not enterprise budgets, and I'd rather ship something useful now than wait for it to be perfect. If you're an agency or a small business asking the same question I was, I'd like to hear whether this is useful to you.

— Ben