Answers to the questions people actually type
Why ChatGPT gives you a different answer than your customer
Both answers are real. The variance is a property of the thing, not a fault in your check — and it is the reason a single look at your own screen tells you almost nothing.
Read itThe pages AI reads before recommending anyone
And it was probably not your website. The cited sources are the most actionable half of any reading, because they turn "be more visible" into a list of specific pages.
Read itHow often AI answers change, and what that means for tracking
The useful question is not whether AI answers change — they do, constantly — but how much of that movement is signal. Almost everyone reports the noise as progress.
Read itHow to get your business mentioned by AI assistants
Assistants compose answers from what they can find written about a category. That makes this a coverage and description problem, not an advertising one — which is worse news for your budget and better news for your odds.
Read itChatGPT, Claude, Gemini and Perplexity recommend differently
They disagree with each other considerably more than they disagree with themselves — which means a reading from one of them is a reading about one of them, and your buyers are not all in the same place.
Read itYour traffic fell and your rankings did not move
If the summary above the results answered the question, the click never happened — and nothing in a rank tracker will ever explain it to you.
Read itWhat AI Overviews did to click through rates
Every number published for this is somebody’s sample of somebody’s queries. The mechanism is general; the size is yours, and it is measurable on your own set.
Read itHow to get into the answer box
The summary is written from a handful of pages. Being one of them is a different job from ranking below it, and the page shape that wins is unusually specific.
Read itRanking first and being invisible at the same time
These are not contradictory readings. They are two surfaces reporting different things about the same question, and the contradiction is usually where the actual problem is.
Read itWhich questions trigger an AI answer and which do not
The set is unstable and nobody outside the engines knows the rule. What you can know is which of your own questions trigger one today, which is the only version of this that is actionable.
Read itSEO for a small business with no budget and no agency
Most of what an agency would charge you for in the first three months is work you can do in an afternoon. The expensive part is what comes after, and you may not need it.
Read itWhat an SEO tool cannot tell you in 2026
The gap is not accuracy. The established platforms are good at what they do. The gap is that the question a buyer asks now gets answered somewhere none of them look.
Read itWhen to hire an agency and when not to
Most businesses that hire early are paying somebody to do the afternoon of work they could have done, and then paying monthly for reports about it.
Read itThe parts of SEO that still work, and the parts that stopped
The answer is not "SEO is dead" and it is not "nothing has changed". It is that ranking became necessary and stopped being sufficient, and a few specific tactics went from weak to worthless.
Read itHow to measure share of voice in AI answers
Every decision in this method makes the resulting number smaller. That is not modesty — each one exists because the alternative produces a figure that falls apart the first time somebody checks it.
Read itWhy appearance counts are the wrong number
Counting is the first thing everybody does and the last thing that should be reported alone, because position inside an answer carries most of the value.
Read itHow to tell a real change from normal drift
Two readings differ reportably only when their confidence intervals do not overlap. Everything else is the same reading, taken twice, with a story attached.
Read itHow to increase your share of voice
Share of voice responds to coverage, position and question value. It also responds, fraudulently, to changing what you count — which is why the method has to be fixed before the campaign starts.
Read itWhat a competitor you have never heard of tells you
That is not a gap in the data. It is the data: your competitive set as a model sees it, which is frequently not the set on your slide.
Read itWinning in a red ocean, when everyone is already there
Saturation is measured on one surface and assumed on the others. That assumption is frequently wrong, and it is the cheapest strategic advantage available right now.
Read itFinding a blue ocean before it fills
The old problem with blue ocean thinking was that you could not tell whether the water was empty or just unmeasured. On this surface you can look.
Read itHow to spot an opportunity in AI answers before competitors do
Composed answers move faster than rankings and much faster than revenue. That makes them a leading indicator, if you watch the right part of them.
Read itThe questions nobody in your category is answering
A conventional content gap is a list of things competitors already did. The more valuable gap is the set of questions where nobody has said anything worth quoting.
Read itWhen to compete on a question and when to concede it
Conceding deliberately is a strategy. Conceding by spending three years failing to win is the same outcome with an invoice attached.
Read itCategory creation versus category entry
Both are legitimate. On this surface they have unusually clear and different signatures, which makes the choice measurable rather than philosophical.
Read itWhy Reddit shows up in AI answers about your category
Our own first full run found reddit.com, youtube.com and g2 among the most-read sources. That is not a quirk of one measurement — it is what happens when a machine looks for people describing a category rather than a company describing itself.
Read itPositioning a small brand so an assistant can describe it
This is the cheapest advantage available to a small business and the one most consistently thrown away, because vague positioning is comfortable and specific positioning excludes people.
Read itWhat to publish so an AI can quote you
Answer engines retrieve passages, not documents. Six sharp self-contained answers outperform four hundred more words of prose, and the difference is structural rather than stylistic.
Read itGetting into the directories and lists assistants read
This is the least glamorous work in the whole category and reliably the highest-yield, because the set of pages deciding your answers is much smaller than it feels.
Read itFounder visibility and why assistants notice it
Not because assistants care about personalities, but because a named human making a specific claim produces exactly the shape of source these systems retrieve.
Read itWhat to do when AI recommends your competitor
It is information, not a verdict. It means the assistant found more to say about them than about you, in the places it looked — and that is a different problem from being worse, with a different fix.
Read itHow to check your AI visibility for free
You can do this by hand in an afternoon, with a spreadsheet in a week, or in under ten minutes with no account. All three are listed here, including the two that do not involve us.
Read itAI search is not SEO, and the difference costs money
You can sit in the top five and be absent from the part most people read. The two surfaces share an input — what is published about you — and reward almost opposite things with it.
Read itunbound geo pricing, explained
The ladder is rows and refresh rate, not access. From the entry plan upward every plan carries every capability, because a locked feature teaches a customer the product is smaller than it is at the exact moment they are exploring.
Read itDoes ChatGPT recommend your business
You can check in under ten minutes. What most people do instead is ask it themselves, and that gives an answer no customer of theirs will ever see.
Read itIs unbound geo legit
A fair question to ask about any tool you found through a search. Here is who runs it, what it costs, what it actually measures, and the four things it cannot do.
Read itPublished in batches by complete subject rather than all at once, because switching a large number of pages live in one go is the pattern quality classifiers punish. The commissioning list runs to 38 pieces; these are the ones that are written.