Position · Tooling
The best-funded platform in the category raised $96 million on exactly this problem. Their own announcement explains why.
In short
AI visibility platforms measure mentions, citations, and share of voice across engines. They do not fix crawler access, restructure pages, or build third-party citation supply. Buying a subscription before running a manual test is buying a number you cannot yet interpret, and a recurring cost attached to no decision.
The position
There is a version of this argument that sounds self-serving coming from an agency, so here is the honest structure of it.
Monitoring platforms are genuinely useful. They automate a manual process, store history you would otherwise lose, and catch changes faster than a person running prompts by hand once a month. If you are going to measure consistently for a year, one will save you time.
What they do not do is close a gap.
A dashboard reporting that you appear in 11 percent of category answers has told you a fact and assigned you no work.
You do not have to take that from an agency. The category's own leader has said it, in the announcement explaining why they raised nearly a hundred million dollars.
The evidence
Profound raised a $96 million Series C in February 2026 at a billion-dollar valuation, led by Lightspeed Venture Partners with Sequoia Capital and Kleiner Perkins participating. That brought total funding to roughly $155 million across four rounds in about eighteen months, and made it the first unicorn in the category.
Now read what they wrote in the announcement, describing why they raised it.
“As our customers operationalized that insight, they often had to export Profound data into separate orchestration and automation tools, learning and using multiple systems just to execute.”
Profound, Series C announcement, February 2026
That is the whole argument, made by the company with the most to lose from it. Customers bought excellent measurement, then had to leave the platform to do anything with it. The company's response was to build execution into the product, and investors valued that at a billion dollars.
Measurement is becoming a feature of something larger, and the something larger is execution. That is not a criticism of monitoring tools. It is what the best-funded monitoring tool in the market concluded about its own product.
Steelman it
Three situations, stated plainly, where buying software is the right call and hiring anyone is not.
You already have execution capacity. An in-house technical SEO team, a content team, and a PR function. You have people who can act on a finding. In that case the tool fills the only real gap, which is measurement, and it is money well spent.
You are in a settled category and defending. If you already own your category, ownership persists month over month 90.4 percent of the time according to Semrush's 2026 study of 1,094 categories. Defending mostly requires knowing quickly when something slips. That is monitoring's best use case.
You need the history more than the analysis. Twelve months of consistent data has value a manual quarterly test cannot reproduce, and rebuilding it retroactively is impossible.
If none of those describe you, the tool is measuring a problem you have no capacity to fix.
Translate it
| The work | Tool | Requires a person |
|---|---|---|
| Detect that you are absent from category answers | Yes | |
| Track share of voice month over month | Yes | |
| Alert when a competitor gains ground | Yes | |
| Find that GPTBot is disallowed in robots.txt | Some | Fixing it, always |
| Restructure a page so a passage can be lifted | Yes | |
| Decide which twenty pages get restructured first | Yes | |
| Earn a mention on a third-party site | Yes | |
| Publish original data other people cite | Yes | |
| Fix inconsistent business information across nine listings | Some detect it | Yes |
| Decide what any of it means for next quarter's budget | Yes |
Roughly 82 percent of AI citations come from earned media, per Muck Rack's December 2025 analysis, and around 77 percent of sources cited about a brand are off-page, per Omniscient Digital's review of more than 23,000 citations.
Look at where those two numbers land in the table. The largest lever in the discipline sits entirely in the right column.
The sequence
Run the manual test. Thirty minutes, four engines, a frozen prompt set. You now know the shape of the problem.
Fix access. Free. robots.txt, indexation, listing consistency.
Fix structure. Editing, not rebuilding. The highest-return hour on most sites.
Buy monitoring. Now the number means something, because you have changed inputs and want to see the effect.
Fund citation supply. The largest and slowest lever, and the one worth a real budget.
Most brands run this list in reverse, starting at step four because it is the easiest thing to purchase and the easiest thing to show a boss.
One question worth sitting with. If a dashboard told you tomorrow that your share of voice is 9 percent and your nearest competitor is at 31, what would happen next inside your company? If the honest answer is a meeting and no assigned work, the tool is not the missing piece.
Common questions
It depends on whether you have execution capacity. A tool measures and reports. If you have people who can restructure pages, fix technical access, and earn third-party coverage, a tool fills your only gap. If you do not, the tool reports a problem nobody is assigned to fix.
They run prompt sets across AI engines automatically, track brand mentions and citations over time, calculate share of voice, and alert on competitor movement. They measure. Most do not restructure content, repair crawler access, or generate third-party citations, which is where most of the work sits, though the leading platforms are now building execution features for exactly that reason.
After. A manual test across four engines takes about thirty minutes and tells you the shape of the problem, which is what makes a tooling decision informed. Buying first produces a number you cannot yet interpret and a subscription attached to no decision.
Partially. Most will show you which prompts and engines you are missing from, and some detect technical access issues. Diagnosing whether the cause is structure, entity consistency, or third-party corroboration, and deciding which to fund first, still requires a person.
Because AI answers are genuinely volatile. Profound's own research found up to 90 percent of cited sources in AI answers can change over time, and that different models rely on largely distinct source sets. AirOps found only 30 percent of brands remain visible across consecutive responses. A tool that reports a single number without a run count is reporting some of that volatility as signal.
Before you buy anything
You arrived here thinking about step four. Plenty of brands are still on step two and do not know it. The AI visibility audit from Conspicuouz Creative Group runs the prompt set across ChatGPT, Claude, Gemini, and Perplexity, identifies which of the three structural causes is producing your result, and tells you whether a subscription would be measuring something you can act on. Free, and you keep the findings either way.
Or run the thirty-minute version yourself first. That is the honest order.