Scrunch reviews compared: no rating anywhere, eleven case studies with a byline, and the date hidden in the URL

Six different people have put their own names on Scrunch's published proof, which is more human accountability than anything else in this category offers, and it comes with one weakness that is easy to miss: the date is usually in the web address rather than on the page. Fetched on 20 October 2026, its case studies index listed eleven stories by named company with six named authors between them, exactly one card showed a last updated line, and no review score and no review count appeared on the home page, the pricing page, the about page or that index, under both a Chrome on Windows and a Safari on macOS request, all HTTP 200. The old scrunchai.com address now resolves to scrunch.com.

We sell a tracker in this category and we publish no rating either, so run the criterion below against us as well as against them.

The criterion, applied to all six specialists and to us

One question, settled from a vendor's own pages: what evidence about this product can a buyer read today, who produced it, and does the claim carry a count and a date? Of the six specialist vendors our own audits named, read on 20 October 2026, three published a rating figure and one published the number of reviews behind it. Three published no third party rating at all: Scrunch, Profound and Rankscale. We publish none on our home page, our pricing page or our proof page, read the same day.

The directories that would carry the missing count refused the request. g2.com, capterra.com and trustradius.com each answered HTTP 403 to both agents, so there is no reading from any of them on this page. A refused request tells you nothing about what is on the other side of it.

So on the review line, Scrunch and this site are in the same empty row, and the comparison has to move to the evidence each company does publish. Scrunch publishes considerably more of it than we do.

Six bylines, which is the strongest thing on that index

The case studies index lists eleven stories by named company: Akamai, Proper Propaganda, Strapi, Stratabeat, Tinybird, AlchemyLeads, Big Leap, Runpod, Clapping Dog Media and BairesDev. Six different people are credited as authors across them, including Eric Wendt and Kevin White, and Kevin White is named as Head of Marketing on the company's own about page.

Treat that last fact as information rather than as a catch. An in house byline is a disclosed byline: the reader knows who wrote it and can see on the same site what their job is. The alternative, a case study with no author at all, is the same document with the accountability removed. Several of these stories also quote a named person with a named employer and job title, such as the VP of Marketing at Strapi and the Head of Growth at Tinybird, which is a reference you could follow up rather than an anonymous endorsement.

The about page supports that: founders Chris Andrew and Robert MacCloy are named with their roles, five more executives are named, and the company dates itself from 2023. For a subscription you may renew three times, knowing who runs the company is worth more than a score.

The one field the index mostly leaves out

Of the eleven cards, exactly one shows a last updated line, reading 04.02.2026. Six of the eleven carry a year and a month inside the URL itself, in the form 2025-04, 2025-07, 2026-01 and so on, and the remaining ones carry no date signal anywhere a reader would look.

That matters more here than it would in most categories, because every number in these stories is a measurement of machine behaviour, and machine behaviour moves. A citation growth figure from April 2025 was produced before several of the engines in question changed how they assemble answers. A reader who cannot see the month cannot weight the claim, and a model summarising the page will not invent the date for them.

The practical move for a buyer is simple: open the story, look at the web address, and read the year and month out of it before you read the percentage. Where the address carries no date, ask for it.

What the published entry tier actually contains, which is how to settle the worth question

Read 20 October 2026, Scrunch's pricing page publishes one buyable tier and one custom tier. The entry tier, Core, states a flat ceiling rather than a meter: 125 unique prompts, 5,000 responses a month, 5 site audits a month, 1 brand workspace, 5 user licences and 4 supported engines, named as ChatGPT, Perplexity, Google AI Overviews and Copilot, with a seven day free trial on the button. The Enterprise tier is custom and carries nine engines, API access, single sign on, expanded model coverage and its Agent Experience Platform.

Two things follow from the shape of that. First, the product sells itself on more than monitoring: its own product menu names agent pages, agent traffic and site optimisation alongside monitoring and citations, so part of what you are buying is changing what an agent reads on your own site. Second, a flat ceiling is easy to plan against and the one number to check is the prompt count, because 125 unique prompts is the line your buyer question list either fits inside or does not.

What the seven day trial can settle	What it cannot
Whether 125 prompts covers your real buyer questions	Whether a figure from an eleven month old case study still holds
Whether four engines match where your buyers ask	What the nine engine coverage looks like, since that is the custom tier
Whether the site audit finds anything your team did not know	How the data leaves the tool, since API access is on Enterprise

How often this name appeared in our own readings, counted twice and never merged

The first count is about sources. In the 597 reading run of 15 October 2026, putting buying questions to ChatGPT and Gemini across the United States, the United Kingdom and India, the Scrunch domain appeared in the source list of 11 readings.

The second count is about recommendations and has its own denominator. In a wider run on 16 October 2026, 1,260 readings of the same two engines, counting a vendor once per reading where its name appears in the saved answer text, Scrunch AI was named in 21 readings. Nothing here adds those numbers, averages them or presents one as a version of the other. Both runs covered two engines and say nothing about Perplexity, Claude or Google AI Overviews.

Both counts are small for a company publishing this much, and that is the honest shape of this market: the vendors with the most published proof are not automatically the vendors the engines hand to a buyer. Those are two different competitions.

Questions buyers ask about Scrunch before signing up

Is Scrunch worth it?

Decide it inside the seven days, since there is no rating to decide it for you. Write out every buyer question your customers would actually ask, keep your own brand name out of all of them, and see whether the list fits inside 125 unique prompts. Then run the site audit side of the product, because that half is the part a rival monitoring tool will not do for you, and it either turns up work your team had not spotted or it does not. Those two answers are the purchase.

Why can I not find Scrunch on G2 or Capterra?

We have no reading to offer, because both sites returned HTTP 403 to both of the agents we tried on 20 October 2026. Open them in your own browser and read the review count before the score.

Has the company changed its domain?

Requests to the older scrunchai.com address resolved to scrunch.com on 20 October 2026. Any link or screenshot you hold from earlier in the year may point at the old form, which is worth knowing before you compare an old quote with a new page.

Are vendor written case studies worth reading?

Yes, and these are better than most, because they carry an author and in several cases a named customer with a job title. Read them with two questions in hand: what month is this from, which here usually means reading the web address, and what was the number before the work started. A story that answers both is evidence. A story that answers neither is a brochure with a graph in it.

How many engines does the buyable tier cover?

Four, named on the page: ChatGPT, Perplexity, Google AI Overviews and Copilot. Nine sits on the custom tier. If your buyers use Claude or Gemini, settle that before the trial starts rather than after.

Read the web address before you read the percentage

This is a company that publishes who wrote its proof and who runs it, and then hides the one field that tells you whether the proof is current. Both halves are worth copying in opposite directions: put a human name on anything you publish about your own results, and put the month in the visible text and not only in the slug, because the engine that summarises your page for a buyer reads the words and not the address bar. Then do the thing no case study can do for you: ask your own buyer questions blind, on the engines your buyers use, and write down today's answer with today's date on it. The comparison that sets this vendor against the one metered most differently from it is at Scrunch AI vs Rankscale, one of 915 pages on this site, 46.7 per cent of 1,960, that nothing else here links to.