This page covers two things. How to read the report, and what to do about what it says.
Read part one once. After that you'll only need part two.
This is a monthly report, not a platform. No dashboard, no login. A PDF arrives and that's the product.
There are bigger tools in this space. Profound, Semrush, Ahrefs and plenty of others, and several of them do more than this does. Years of historical data, sentiment scoring, category-wide tracking, dashboards a whole team can log into. If you want those things, buy one of them. They're good at what they do.
I built this because I wanted it for my own business and for the clients I work with. A fixed set of questions, asked every day, with the answers kept so any number can be checked.
Two consequences worth knowing up front.
Your numbers won't match another tool's. Every tool differs on which questions get asked, whether mentions or citations count, how often it checks, and whether it reports fresh results or a stored library. Two tools can both be honest and differ by forty points. What matters is that the method here stays fixed, so the trend means something.
Nothing here is modelled. Some tools publish estimated visibility or projected impressions. Every figure in your report counts answers that actually came back. If you want to see the answers behind a number, ask. They're stored.
The percentage at the top is how often your site was cited, meaning linked to as a source, across every answer sampled that month. Beside it is the change from last month in points.
That's the single number to watch. Everything below it explains where it came from.
Two different things get counted here. The gap between them matters more than either one on its own.
Named means a platform said your business name in its answer. No link, no click, nothing in your analytics. But your name entered the buyer's shortlist, and plenty of people go on to search a name they just heard.
Cited means it linked to your site as a source. That can send a visit, and it means a page of yours was treated as evidence rather than recalled from memory. Citations respond faster to work on your site, which makes them the more useful of the two to act on.
The "vs last" columns show movement in points, per platform. A single blended number hides the case where one platform is climbing while another slides, and that's usually where the story is.
"Questions you appeared on" counts how many of your tracked questions you showed up on at all, even once. It's a breadth measure. Appearing strongly on two questions is a narrower position than appearing occasionally on eight.
Daily citation rate across the month. The light band is named, the solid line is cited, the dashed line is a seven-day average, and the grey line is last month over the same days.
Watch the dashed line. It's the only one that shows direction without daily noise.
The same citation rate, split by platform, smoothed over seven days.
Expect these lines to diverge. They're different products with different retrieval systems, and strength on one with weakness on another is a diagnosis rather than an error.
Google AI Mode runs on Google's index, so it tracks closest to normal search rankings and local results.
ChatGPT answers from training first and searches when it decides the question needs it. That makes it slower to notice site changes and heavily influenced by how widely you're discussed elsewhere. Its searches run on Bing.
Gemini has its own retrieval layer and favours established, well-linked sources. It's the most conservative about naming smaller businesses.
Perplexity searches on nearly every question and cites heavily, so scores here usually run highest. It's also quickest to reflect new content.
Strong on Perplexity but weak on ChatGPT means good pages that nobody talks about. The reverse means a known brand whose pages aren't usable as a source.
One square per day per platform. Solid means cited, faded means cited on some platforms, pale means not cited.
This section explains the phone call where someone says they asked ChatGPT and didn't see you. AI answers are generated fresh every time, so the same question can include you Tuesday and skip you Wednesday with nothing changing on your end.
One answer is one roll of the dice. The rate across a month is the signal.
Share of all brand appearances across your tracked questions, with movement since last month.
This isn't a list of competitors you gave us. It's who the platforms actually named when answering questions about your category. Sometimes that surfaces a competitor you don't think about much. Occasionally one you've never heard of.
The movement column is where to look. Someone gaining four or five points did something that's working right now.
Every tracked question, with your rate this month, your rate last month, and who else appeared.
Rows where you sit at 0% and competitors are listed are the clearest work in the whole report.
Which pages on your own site the platforms used as a source, and how often.
Usually a short list, often with one page doing most of the work. That page is worth protecting.
Every domain cited in answers about your category, with new ones flagged.
Most people scroll past this. It's the most useful section on the page. More on why in part two.
Three items generated from that month's numbers. They change as the numbers do.
These are patterns pulled out of the data rather than a strategy. They point at where to look.
Four patterns, four different problems. Find the one that matches your report.
Looks like: rows in the by-question table at 0% for you, with two or three competitors named.
Means: nothing on your site answers that question directly. The platform went looking and found theirs.
Do: build or rewrite the page that answers that exact question, in the words a customer would use. Not a service page with the answer buried in paragraph nine. A page whose job is answering it.
One distinction worth learning. Hiring questions, like "who should I call to replace my roof in Brentwood," are winnable by local businesses. Diagnostic questions, like "why is my roof leaking," tend to get answered with manufacturers, national publishers and forum threads. If you're missing from a diagnostic question, that's a harder fight and often not the right one to pick.
Looks like: named well above cited. Seventy percent and twenty percent, say.
Means: they know who you are but aren't reading your pages. Your site isn't usable as a source.
Do: add the specifics. Platforms cite pages that state things plainly:
"We proudly serve the greater metro area with quality service" gives a machine nothing to quote. "We serve Bedford, Forest, Madison Heights and Amherst, and a standard replacement runs four to six weeks out in spring" gives it something to use.
This is the fastest pattern to fix. You already have the authority. You're just not giving them anything to cite.
Looks like: named and cited both near the floor across most questions.
Means: the wider internet isn't discussing you. Not a website problem, and on-page work won't fix it quickly.
Do: go to the sources section. Those domains are where answers about your category come from. Review sites, directories, trade publications, forums and YouTube channels. Real presence on them is the most direct route to getting mentioned.
Reviews where reviews get cited. A listing where listings matter. A mention in a roundup. Genuine participation in a forum where people ask about your category. This takes months rather than weeks. It also moves the numbers more than anything else on this list.
Looks like: good numbers, with one page holding most of the citations.
Means: you've earned a narrow position.
Do: keep that page current, because platforms weight recency and a stale page loses ground quietly. Then build the neighbouring pages that answer adjacent questions, and link them together. Extending from something that already works beats starting cold.
The sources list shows where the answers in your category come from, produced by the platforms themselves rather than by guesswork.
Any new domain on it is worth a look. It's either a competitor gaining ground or a publication that started covering your category. Someone earned a place there this month, and you can find out how.
Reacting to one bad answer. Check the monthly rate before calling a meeting.
Writing pages that target the tracked questions. The question set samples how customers ask. Write for the customer and the sample takes care of itself.
Treating this as separate from SEO. The pages platforms cite most are usually the ones already ranking well. This measures a different surface, not a different discipline.
Expecting traffic to move first. AI citations produce far fewer clicks than search results do. Being in the answer often matters more than being clicked, because people search your name afterwards. Judge this by named and cited, not by a jump in sessions.
Judging a change in two weeks. Give anything four weeks before deciding whether it worked.
The report tells you which pattern you're in. Base Hit Marketing does the work that changes it.
Talk to us