Resources
How answer engines pick the brands they name.
Written for B2B SaaS teams and the agencies that run search for them. Some pieces work through a method you can use this week. Others explain something that has just changed, and what it means for a brand trying to get recommended.
Start here
Listening tools tell you volume. They do not tell you what people said.
A mention count can be technically correct and completely useless at the same time. Four ways that happens, found by reading every mention behind a single number.
9 minute readReddit lost 86% of its ChatGPT citations in four days. Nobody at Reddit changed anything.
The most-cited source on ChatGPT Search became one of the least-cited in under a week, without publishing anything differently.
6 minute readAI search
How answer engines decide which brands to name, how to measure it honestly, and what actually moves it.
Before you can measure AI visibility, you have to pick the questions
Every brand that starts tracking AI visibility hits the same wall in week one, and it is not a technical one. Somebody has to decide what to ask.
8 minute readWill AI-written content get you penalized?
It is the first question every marketing team asks before it lets a tool near the website, and the answer people repeat to each other is usually wrong in both directions. Some believe any AI involvement is fatal. Others believe nothing is. Neither survives reading what the search engines actually published.
7 minute readOne question, three gates, and the fix at each one
A buyer asks one thing. Between that question and your name appearing in the answer there are three separate checks, they happen in order, and each one fails differently. Most reporting tells you only that you are missing, which is the easy half.
8 minute readAnswer engines do not rank you. They quote you.
Optimizing for a ranking gets you a position in a list. Optimizing to be quoted gets you named inside somebody else's sentence, which is a different job.
8 minute readWhy AI-referred visitors convert four times better
A channel producing a fraction of the volume and several times the conversion rate deserves more attention than it gets. Most brands are not even measuring it.
5 minute readYour buyer decided before they ever reached you
For twenty years the first thing a buyer did was arrive. Now the first thing they do is ask, and get an answer that may not include you.
6 minute readYou asked ChatGPT once and it named you. That is not a measurement.
Everyone starts the same way. Type the category question into ChatGPT, see whether the brand comes back, screenshot it if it does. The result feels solid. It is closer to a coin toss.
7 minute readYour structured data was written for search engines. Answer engines want something else.
Most sites already carry structured data. Nearly all of it was added to win a star rating or a recipe card, which is a different job from being quoted correctly in somebody else's answer.
8 minute readYou are not in the answer because you are not in the sources
Brands rewrite the page they control and wonder why nothing shifts. The answer was assembled somewhere else, out of sources they have no presence in.
7 minute readThe same question has a different answer in every country
A brand selling into three countries usually tracks one number. That number averages three different realities, and it hides whichever one is failing.
6 minute readYour brand, and your customers
Staying recognisable when many people, and now software, write for you. And what your own orders already tell you.
Why a broad question rarely names a challenger
A challenger runs its first AI visibility check, types the broadest question in its category, and finds the market leader named and itself missing. That result is real. It is also the least useful question a challenger can start with.
6 minute readYour brand voice does not survive the handoff
Every brand book is a description. What you need is a specification, and the difference is whether a stranger can apply it without asking you a question.
7 minute readThe brand questionnaire nobody fills in
Every engagement starts with a document that assumes the client has three uninterrupted hours. They do not, and the work starts anyway.
6 minute readConsistency is not sameness
The most common brand failure is not inconsistency. It is a brand so tightly policed that nothing memorable survives review.
5 minute readThe market around you
Competitors, the conversation, and timing, without drowning in dashboards.
Running the work
Where growth comes from when paid gets expensive, and how to tell whether any of it worked.
Paid acquisition reprices every year. The other routes to being found do not.
When the cost of the same customer keeps climbing, the options are to accept worse margins, spend less, or stop relying so heavily on the channel. Most brands pick the first by default.
7 minute readMarketing does not have a data problem
Every quarter another dashboard gets commissioned to answer a question the last dashboard already answered.
5 minute readOver-tooled and under-decided
Somewhere between the fifth and the ninth tool, a marketing team stops operating and starts administering.
5 minute readGet your brand named
BrandVelocity shows B2B SaaS challengers how AI assistants answer their buyers’ questions, then fixes what those answers get wrong.
Ask us your own category question