Otterly.AI reviews compared: five pieces of recognition, one rating figure, and the source of that figure sits inside an image
Read on 20 October 2026, the Otterly.AI home page prints 4.8 out of 5 beside a G2 logo that is served as an image with empty alt text, so a reader sees the source and a machine sees a bare number.
Published by AI Knows Us · Updated 20 October 2026
Otterly.AI publishes more third party recognition than any other specialist in this category and no review count at all. Fetched on 20 October 2026, its home page hero prints the text 4.8/5 next to two images, a G2 wordmark and a G2 star graphic, and the figure carries no number of reviews beside it. The same hero names a 2025 Gartner Cool Vendor recognition for AI in Marketing. Its about page names being made a High Performer for Answer Engine Optimization Tools by G2 in 2026, and the home page adds Tekpon Top SEO Software Q4 2025 and two OMR placements. That is a strong set of signals, and the one piece a buyer most wants, how many people the 4.8 is an average of, is on none of those pages.
We sell a tracker in this category and we publish no rating of any kind, so our own row on the criterion below is empty. Check the documents rather than taking our word for it: every response we read was saved on 20 October 2026.
The request shape that decides whether you can read this site at all
Before any of the above, a practical note that cost this programme an hour once already. otterly.ai answered HTTP 403 to a Chrome on Windows request and HTTP 200 to a Safari on macOS request, minutes apart, on both the home page and the pricing page. The response bodies were 75KB of challenge for the first and 222KB and 283KB of real document for the second. The user agent string was the whole difference. If you have ever concluded that this vendor blocks you, try another browser before you write that down. A 403 from a host records a refused request, and it is not a reading about what the page contains.
The criterion this page uses, applied to all six specialists and to us
One question, answerable from a vendor's own pages: 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 and pricing pages of the six specialist vendors our own audits named, plus one more page each where one existed, all on 20 October 2026.
| Read 20 October 2026 | Rating figure published | Count of reviews published | Source named in text |
|---|---|---|---|
| Otterly.AI | Yes, one | No | In an image, not in text |
| Peec AI | Yes, two different ones in one footer | No | Yes |
| Depra AI | Yes, one | Yes | Yes, with the date read |
| Profound | No | No | Not applicable |
| Scrunch | No | No | Not applicable |
| Rankscale | No | No | Not applicable |
| AI Knows Us (ours) | No | No | Not applicable |
On the first column Otterly beats us outright, and on the awards column it beats everybody here including us by a distance. We hold no analyst recognition and no directory badge. That is worth saying plainly before the next section, which is a criticism of one detail inside an otherwise strong page.
The detail worth fixing, and it matters most to a company selling this
Strip the markup out of that hero the way a text extractor does, and the three elements read as: an image with alt text of nothing, a second image with alt text of nothing, and the characters 4.8/5. The G2 wordmark and the star graphic are both <img> elements carrying alt="". The Gartner logo beside the Cool Vendor line is the same. So a person looking at the page sees a sourced, attributed rating, and anything reading the text alone sees a floating 4.8 out of 5 with no issuer and no sample size.
This is an ordinary web build decision and almost every site in this market makes it somewhere. It is pointed out here because of what the company sells. Otterly.AI sells brand visibility on answer engines, which is a product about whether machines can read and attribute what you publish. Its own strongest piece of social proof is the one claim on the page a machine cannot attribute. Putting the issuer and the review count into the text, in words, would cost one sentence and would make the figure quotable by the systems it sells coverage of. The same fix is the one its own product recommends to its customers.
The about page does it correctly for the awards: both the Gartner recognition and the G2 High Performer placement are written out in words with their years. That is the right pattern, and the rating is the one element that has not caught up with it.
Where the review count would have come from, and why it is not the whole answer
We tried to read the denominator ourselves. g2.com, capterra.com and trustradius.com each returned HTTP 403 to both user agents on 20 October 2026, so there is no count to report from any of them and we are not going to estimate one.
There is a second reason not to lean the decision on that number even once you have it. Across 1,260 readings of ChatGPT and Gemini taken on 16 October 2026, we recorded 20,143 citations across 4,947 domains, and the whole review directory class accounted for 0.7 per cent of them. An earlier run on a different question set, 597 readings on 15 October 2026, put the same class at 0.8 per cent of 10,652 citations. Two question sets, eleven days apart in design, same answer. A rating is useful for persuading a person, and on this evidence it is close to invisible to the two engines we measured.
What its own pricing document publishes, which is the checkable part
Read on 20 October 2026 under the Safari agent, the Otterly pricing page publishes four tiers with figures on three of them, a named allowance unit of search prompts at 15, 100 and 400 rising to from 1,000 on Enterprise, unlimited team members on every tier including the cheapest, daily tracking frequency throughout, and a stated annual saving. One workspace on the entry tier and unlimited workspaces above it. Four engines are included on every tier, ChatGPT, Google AI Overviews, Perplexity and Microsoft Copilot, and Claude, Gemini and Google AI Mode are paid add-ons on all four.
That last line is the one that moves a bill, and it is the real answer to whether this is worth it for you. Count the engines your buyers actually use before you compare any tier against anything.
How often this name turned up in our own readings, counted two ways
These are two separate measurements of two separate things, kept apart on purpose. On sources: in the 597 reading run of 15 October 2026, otterly.ai appeared in the source list of 58 readings. On naming: in the 1,260 reading run of 16 October 2026, counting a vendor once per reading where its name appears in the saved answer text, Otterly.AI was named in 265 readings. Neither figure is added to, averaged with or divided by the other. Both runs covered ChatGPT and Gemini, two engines, and say nothing about the other three surfaces this category watches.
Questions buyers ask about Otterly.AI before signing up
Is Otterly.AI worth it?
The entry tier is buyable off the page without a call, so settle it with your own week rather than with the 4.8. Write 15 buyer questions, which is exactly the entry allowance, keep your brand name out of all of them, and watch the daily readings for seven days. Then add up what Claude, Gemini and Google AI Mode cost as add-ons for the engines your buyers actually use, and compare that total against the tier above it, because the add-on route is sometimes the more expensive one.
How many reviews is the 4.8 out of 5 based on?
The page does not say, on the home page, the pricing page or the about page, read 20 October 2026. The issuer is shown as a logo image rather than in text. Open the directory in your own browser and read the count first, then the score.
What is a Gartner Cool Vendor, and does it mean the product is best?
It is a recognition of a vendor the analyst considers interesting and novel in a named market, in a named year, which here is AI in Marketing in 2025. It is a judgement about the company being worth noticing rather than a benchmark of the product against its rivals, and it is not an aggregate of customer experience. Read it as what it is, which is still more than most firms in this market have.
Why does the site sometimes refuse me?
We recorded HTTP 403 to a Chrome on Windows agent and HTTP 200 to a Safari on macOS agent on the same two pages within minutes, with an identical result when the order was reversed. Change the browser and try again.
Does a star rating help a business get named by an assistant?
On our own measurement it does very little. Review directories took 0.7 per cent of 20,143 citations across 1,260 readings on 16 October 2026, replicating 0.8 per cent from an earlier 597 reading run. Build the listing because buyers read it, and expect the naming to be won somewhere else.
One sentence would make the strongest line on that page quotable
There is a neat lesson sitting in this vendor's own hero, and it applies to every company that has ever been pleased with a badge. The proof you are proudest of has to survive being read by something that ignores pictures, because that is how an assistant assembles an answer about you. Write the issuer, the score and the count in words, in a sentence, next to the graphic, and the claim becomes something an engine can repeat on your behalf rather than something only a visitor can see. If you want to see that same criterion turned on us, including the columns where this vendor is plainly ahead, it is set out at AI Knows Us vs Otterly.AI, one of 915 pages here, 46.7 per cent of 1,960, that no other page on this site links to.