How it works

How we get AI to recommend you.

Not a dashboard and a to-do list. A monthly loop: we ask AI what your customers ask, find out why it picks someone else, fix it, and check what moved.

The screens below are real, from a client's AI Visibility Command Center in November 2026, with names, places and phone numbers changed.

Get your AI Visibility Score Talk to PiXENDA®

01 · Ask

We ask AI what your customers ask.

Twenty questions a buyer near you would actually ask, from who to call for a termite inspection to affordable pest control nearby. We put each one to ChatGPT, Gemini, Perplexity and Claude, with live web search, and save every answer word for word.

You can read every one of them. Nothing is summarized away.

One of 80 answers from a monthly run. ChatGPT names the client first: "My first call." Swipe or tap to see the full screen.

02 · Grade

We grade who AI recommends.

Being named isn't enough. Every answer is graded for every company in it: championed, recommended, mentioned, mixed or discouraged. You see where you stand on each engine, against the competitors AI names instead, and whether it cites your own site as its source.

Named in 49 of 80 answers, more than any competitor, and recommended or championed in 46. Swipe or tap to see the full screen.

03 · Find why

We find out why.

Every page on your site is checked for what changes AI answers: facts that disagree from page to page, claims you can't back up, template text left behind, missing structured data. Each issue gets a severity, an owner and the fix that closes it.

Site Watch found 10 issues across 41 pages, from an unfinished reviews page to years in business counted differently on 11 pages. Swipe or tap to see the full screen.

04 · Plan

We put every fix in order.

One plan, rebuilt every month. The facts get settled first, because every page and listing repeats them. Then the site, then the pages that need your approval, then your listings, then new content.

Each task says who owns it, what it waits on, which questions it should move and how we'll know it's done.

38 tasks in six phases. PiXENDA® could start 10 of them the same day. Swipe or tap to see the full screen.

05 · Fix

We do the work, word for word.

Every task becomes a work order: the page, the exact text there now, the replacement, and the check that proves it's done. Your approval is the only thing we wait on. Then we make the change ourselves.

One of the work orders from the plan above. Swipe or tap to see the full screen.

06 · Measure

We check what moved.

Next month, the same questions go to the same engines. You see every measure side by side with the month before, engine by engine.

With one answer per question per engine, small swings are normal, and we tell you so. The trend over two or three months is what counts.

The same 20 questions, a week apart. Changes this small are normal variation; the trend shows over months. Swipe or tap to see the full screen.

How we measure

The method, in plain words.

Each month we ask ChatGPT, Gemini, Perplexity and Claude the 20 questions a client's customers ask, with live web search, about the client's own towns. We save every answer, grade how strongly each one recommends the client and its competitors, and record which sites it cites. Clients can read every answer.

We also check where the client shows in Google's map pack for the same questions, and scan the site for anything that changes what AI says.

What we don't count yet: calls and bookings from AI. That needs visit and call tracking on the client's site, and we add it to the report as soon as it's running.

See where you stand.

Start with the free score: it shows whether AI can read your site at all. Then ask us what AI says about you, and who it recommends instead.

Get your AI Visibility Score