Profound reviews compared with what it publishes instead: twenty one named customer outcomes and no star rating

Twenty one named companies have let Profound publish a number about them, and not one of those numbers is a review. Fetched on 20 October 2026, its customers page lists 21 story links by company name, most of them headlined with a percentage or a multiple, while the home page, the pricing page and the customers page carried no rating figure and no review count at all, under a Chrome on Windows request and a Safari on macOS request, both HTTP 200 with identical bytes. So the useful review question here is not how many stars Profound has. It is whether anybody can tell you how one of those percentages was produced.

We sell a tracker in this category, so read our ranking criterion rather than our opinion, and check it yourself. Everything below came from documents we requested and saved on 20 October 2026.

The criterion, stated so you can apply it to anybody including us

One line, answerable from a vendor's own pages in ten minutes: what evidence about this product can a buyer read today, who produced it, and does the number carry a count and a date? We read the home page and the pricing page of all six specialist vendors our own audits named, plus one further page each where one existed, on 20 October 2026. Three of the six published a rating figure. One of the six published the number of reviews behind its figure. Three published no third party rating of any kind, and Profound is one of those three.

The directories where that count would normally live refused the request. g2.com, capterra.com and trustradius.com each returned HTTP 403 to both user agents we tried, so we could not read a denominator from them and we are not going to guess at one. A 403 records a refused request, not an absent figure, and this programme has already caught itself treating one as the other.

Our own position on this criterion is the weakest in the set. aiknowsus.com publishes no third party rating, no review count and no customer case study on its home page, its pricing page or its proof page, read the same day. Profound publishes 21 named customer stories and we publish none. That is a plain gap and it is ours.

Twenty one stories, and the field none of the headlines carries

The customers page lists 21 story links by named company. The headline outcomes include Plaid at a 300 per cent rise in answer engine referral traffic, MongoDB at a 50 per cent rise in AI search visibility, Arizona College of Nursing at 51 per cent on AI referred traffic conversions, Aleph at 82 per cent on attributed traffic, Jordan Digital Marketing at 34 per cent on revenue, Hone at 800 per cent on visibility, Ramp and Alchemy each at seven times, Lake.com at five times, CRS credit API at twenty times, and OpusClip at 45 per cent brand visibility with what the page calls the top citation share in thirty days. Above them sit aggregate claims: more than 25,000 marketers in over 90 countries, 235,000 unique agents built and 120,000 hours saved.

That is a far larger body of named evidence than a short review listing, and it is worth more. It is also missing one field, consistently, in every headline on the index: the window and the baseline. A 300 per cent rise is a ratio between two readings, and a ratio starts being useful when you know what the first reading was, when each was taken, which engines were asked and how many times. Profound's own trial table names exactly those fields for its own product, which makes them fair to ask for.

Three testimonials sit on the same page. Each names a person and a job title, and none names an employer. A quote attributed to an Organic Growth Manager cannot be followed up, while a named person at a named company is a reference call you can book, so that is the field to ask about when a case study is read out to you.

What its free trial publishes, which is the part a review cannot replace

Profound's pricing page publishes two routes, and the free one is specified tightly enough to plan around. Read 20 October 2026, the trial states 50 prompts run daily for seven days across ChatGPT, Gemini and Google AI Overviews, in one language and one region, with unlimited seats. It also states what the trial does not include, in the same table: history none, exports none, API none, support none. Above it, the enterprise tier is marked custom with a demo, and carries up to nine answer engines, single sign on and SOC 2.

Published negatives are rarer than published features in this category, and four of them in one column is a vendor telling you where the trial stops. For a buyer that is more useful than a rating, because it tells you what question the seven days can settle.

What you want to know	Can the seven day trial answer it
Does the tool see my buyer questions at all	Yes. 50 prompts daily is enough for a real prompt set.
Is the answer stable across days	Yes, and this is the best use of seven consecutive daily runs.
Can I keep the evidence	No. The trial column reads exports none and history none, so capture it yourself as you go.
Will it cover the engines my buyers use	Only for the three named. Claude, Perplexity and the rest sit on the enterprise tier.

What our own readings recorded about this name, as two separate counts

The first count is about sources. On 15 October 2026 we put 102 buying questions per market to ChatGPT and Gemini across the United States, the United Kingdom and India, took 597 usable readings and recorded 10,652 citations. tryprofound.com appeared in the source list of 61 of those readings, the highest figure of any specialist vendor domain in that run.

The second count is a different measurement of a different thing, and nothing here adds it to the first or treats one as a version of the other. On 16 October 2026 we put 210 questions per market to the same two engines, which is 1,260 readings, and saved the answer text. Counting a vendor once per reading when its name appears in the text, Profound was named in 219 of those 1,260. Being read to build an answer and being handed to a buyer as the answer are separate events with separate numbers.

Both runs covered two engines. The products in this category, ours included, watch five surfaces, so neither number says anything about Perplexity, Claude or Google AI Overviews.

The four questions to send before the demo

• Pick one case study and ask for the prompt set behind the percentage, including whether the brand name appears in any prompt. A prompt that names the brand inflates the score.
• Ask for the baseline reading and its date, and the second reading and its date. Two dates and two numbers is the whole claim.
• Ask how many times each prompt was run. A rate over runs is comparable and a rate over prompts is not.
• Ask for a named reference in your own industry who will take a call.

Questions buyers ask before paying for Profound

Is Profound worth it?

Decide it on the seven days rather than on a rating, because the rating does not exist and the trial does. Write 20 buyer questions your customers would actually type, keep your company name out of every one, run them daily for the full week, and export nothing because you cannot: paste each day's answers into your own sheet with the date. At the end you have a dated baseline you own, and the purchase decision is whether the dashboard is worth paying for on top of the sheet you just built.

Why can I not find Profound reviews on G2 or Capterra?

Both sites refused our request on 20 October 2026, each returning HTTP 403 to a desktop Chrome agent and to a desktop Safari agent, so we have no reading to report from either. Open them in your own browser, where a normal session usually gets through, and read the count before you read the score. A score with 12 reviews behind it and a score with 200 are two different kinds of information.

Which engines does the trial actually cover?

Three, named on the page: ChatGPT, Gemini and Google AI Overviews, in one language and one region. The nine engine list belongs to the enterprise tier, so if your buyers ask Perplexity or Claude, settle that in writing before the call.

Is a vendor case study worth anything at all?

Yes, and more than a short review listing, once you have the method behind it. A named company, a named outcome and a published window is a stronger claim than an anonymous five star rating, because every part of it can be checked by somebody who was there. The weakness is never that the vendor wrote it. It is a missing denominator.

The reading that settles a percentage

Twenty one companies have let Profound publish a number about them, which is a real signal and more than most of this market can show. What none of those numbers carries on the index page is the pair of dates it sits between, and that is the one thing you can produce for yourself in a week with no contract: ask your own buyer questions blind, on the engines your buyers use, write down who gets named and which pages the answer was built from, and date the page. Do it once now and once after the work lands, and you own both ends of your own percentage. If you want to see the same criterion turned on us, including the fields where Profound is ahead, our side by side is at AI Knows Us vs Profound, which is one of the 915 pages on this site, 46.7 per cent of 1,960, that nothing else here links to.