SEO Consulting

Otterly.ai vs Scrunch AI for AI Visibility Tracking in 2026

Otterly.ai and Scrunch compared on engines, credits, citations, and the November 2026 prices each vendor still prints.

Otterly.ai is the better fit when you choose the engines, at least three of them, from a menu of seven that includes Claude and Gemini on every self-serve plan, and the thing you are buying is a credit budget rather than a fixed engine list. Scrunch is the better fit when one published dollar price has to name eight engines on the Core card, including Gemini, Grok, and Meta AI, plus a response cap, citation targets, and a small allowance of pages to optimize and pages to generate. Neither one is a citation guarantee. On 19 November 2026 the Otterly cards I opened were priced in euros: Lite at €29 a month, Standard at €189, and Premium at €489. Scrunch's brand card showed Core at $250 a month. Enterprise was custom on both.

Last updated - November 2026

I am Austin Heaton, an AEO consultant. The visibility logs I am asked to look at are mostly for fintech. The rest are B2B SaaS and a few other industries when the buyer question is the same. A tracker is useful when I can open the prompt, the engine, and the URL. It is not useful when a team exports a score and calls it a pipeline, or when two pages on the same site still describe two different products.

Otterly's euro cards, the save lines, the "4 of 7" prompt counts, and the comparison table are from Otterly's pricing page, read on 19 November 2026. The same URL's meta description still describes an older prompt count in dollars. I am not treating that description as the card. Credit math, the engine menu, and the rule that you do not buy engines as add-ons are from three help articles dated 10 November 2026: plans and pricing, how to add engines, and how many prompts you can track. Citation gaps are the citations article. The audit is the GEO audit. Coverage versus citations is the brand-report insights. A launch post that uses dollars and rounder prompt counts is New OtterlyAI Pricing. I did not create an account. I did not run a prompt.

Scrunch's Core card, the eight engine names, the 15,000 response line, and the comparison table are from Scrunch's pricing page, read the same day. The older Core description, including 125 prompts and four platforms, is still on the pricing FAQ and on Scrunch's own tools page. Platform choice is the platforms FAQ. The trial and the credit-card rule are the trial FAQ and the shorter trial FAQ. Definitions, with a banner that the guide may be stale, are the brand-presence how-to. Observe, optimize, and deliver are the products FAQ. Citation targets are the citation optimization post. I did not start the trial. I did not export a report.

The pages already on this site are not being rewritten as a second invoice. Otterly on 4 November, when the card was four fixed engines plus paid add-ons and Lite was 15 prompts, is the Otterly.ai review. I am not pasting that add-on price list forward. The help article dated 10 November says you do not purchase additional engines. Scrunch on 10 November, when the brand card I can cite from that review showed 125 prompts, 5,000 responses, and four engines, is the Scrunch AI review. The card I opened on 19 November prints 15,000 responses and eight engine names. I am not carrying 5,000 forward. Peec beside Otterly is Otterly.ai vs Peec AI. Peec beside Scrunch is Peec AI vs Scrunch. Scrunch beside Profound is Scrunch vs Profound.

Key Takeaways

  • On 19 November 2026 Otterly's pricing cards displayed Lite at €29 a month, Standard at €189, and Premium at €489, with save lines of €52, €340, and €880. The compare table repeats those three monthly figures. A script on the page can rewrite a credit-pack line between a euro sign and a dollar sign. The meta description on the same URL still says Lite starts at $29 with 15 prompts, Standard is 100 prompts at $189, and Premium is 400 prompts at $489. The help article dated 10 November says prices are in dollars or euros depending on location, that the plan prices did not change that day, and that what changed was the prompt count and the engine choice. I am not converting €29 into $29. I did not flip a currency control.
  • The cards I could see were built around four engines: 16 prompts and 2,000 credits on Lite, 120 prompts and 15,000 credits on Standard, 403 prompts and 50,000 credits on Premium. The help table, same date as the "prices did not change" answer, prints 21, 161, and 537 prompts if you pick three engines and leave Claude off. Every plan can select from ChatGPT, Google AI Overviews, Perplexity, Microsoft Copilot, Google AI Mode, Gemini, and Claude. The minimum is three engines. Claude costs 3 credits a check. The others cost 1. DeepSeek, Grok, Mistral, Meta, and Qwen are marked coming soon on the 10 November engine article. They are not on the menu I could use.
  • Scrunch's brand card showed Core at $250 a month, with a 7-day free trial on the card, billed monthly. The meters on that card are 15,000 responses a month, 10 citation targets, 10 pages optimized, 10 pages generated, and 1 brand with 5 seats. The eight names are ChatGPT, Perplexity, Gemini, Google AI Overviews, Google AI Mode, Microsoft Copilot, Grok, and Meta AI. Enterprise is custom, billed yearly on the table, and the ninth name is Claude as an add-on. The pricing FAQ and the tools page still describe Core as 125 unique prompts across four platforms: ChatGPT, Perplexity, Google AI Overviews, and Copilot. I am not deleting either description.
  • The two products do not sell the same unit. Otterly prints credits, prompts, and a formula: one credit is one prompt checked once on one engine, and a daily run counts as 31 checks in a month. Scrunch's card prints responses and does not print a prompt cap. The pricing page asks "What's an AI response?" The answer did not render in the HTML I pulled. Scrunch's platforms FAQ says you choose one platform, a subset, or all of them when you create a prompt. That is a choice. It is not the missing formula.
  • Otterly's entry plan does not include the API, MCP, agent analytics, or the Looker connector. Those start on Standard, with 2,000 API requests, 2,000 MCP requests, and 200,000 agent events. Countries, languages, brands, competitors, and seats are unlimited on every Otterly column, and the prompt article says the same question in two countries counts as two prompts. Scrunch Core says countries, languages, and personas are "Any," caps competitors at 5, caps brands at 1, and puts Agent Pages, API keys, and the Data Studio connector on Enterprise. No plan I read, on either site, guarantees a mention.

What Job Are These Two Tools Doing?

Both products answer a version of "did an answer engine mention us, and which URL did it use." They do not answer it with the same unit, and they do not agree with their own older pages about what the entry plan includes. Otterly's 10 November help article calls the product a credit budget you split across engines you choose. The pricing cards I opened show that budget as euros, prompts, and a "4 of 7" label. Scrunch's pricing card calls Core the plan for a team that wants to evaluate the product, set a benchmark, and prove impact before it scales. The same company's FAQ still describes a smaller, four-engine Core. If you buy from the card and report from the FAQ, you will brief two different tools.

Want AI engines to recommend your company? Book a free 30-minute AI visibility review with Austin.

Book a Call

If the missing piece is "Claude or Gemini, on a price I can read, and I will decide which three engines this workspace uses," Otterly's self-serve menu is built for that choice. Grok and Meta are not on it. The engine article lists both as coming soon. If the missing piece is "Grok, Meta, Gemini, and Google AI Mode on one $250 card, and I can live with a response cap whose prompt math is not printed," Scrunch's 19 November card is the one that names those engines before a demo. Claude is the name that flips. It is on every Otterly plan, at a triple credit weight. On Scrunch it is an Enterprise add-on, not one of the eight.

I care about four objects. The first is the engine: a name on the card you can buy, not a name in a FAQ that describes last month's packaging, and not a name in a meta description the cards have outgrown. The second is the unit: a credit, a prompt, and a response are not interchangeable, and a slide that adds them together is a different report. The third is the URL: a mention of the brand and a citation of the brand's own page are different facts. The fourth is the next step: a recommendation, a citation target, a generated page, or a note that the login does not place the article. A tracker that nobody ties to a page is a spreadsheet. A generated page that nobody checks against the fee schedule is a different risk.

There is a packaging change on both sides, and I am dating it so a deck does not mix the months. Otterly's help article, dated 10 November 2026, answers "Did prices change on 10 November?" with no. It says Lite, Standard, and Premium cost the same, and that the change is how many prompts you get and which engines you choose. The 4 November review on this site recorded four included engines and three paid add-ons, and it recorded 15 prompts on Lite. The cards and the help table I used on 19 November do not describe that add-on. I am using the later pages for the menu. Scrunch's 10 November review on this site recorded a Core card with 125 prompts, 5,000 responses, four engines, and one country. The brand card I opened on 19 November prints 15,000 responses, eight engines, and "Any" for countries, languages, and personas. The FAQ has not caught up. A buyer who prints the FAQ into a security review will describe a product the pricing table no longer shows.

How Do Otterly.ai and Scrunch Compare?

The table uses the pages I opened on 19 November 2026. Otterly's figures are the euro cards, because that is what the pricing HTML displayed. The dollar figures in the meta description are marked as the meta description. Scrunch's 125-prompt FAQ is in the table as a conflict, not as a second column I am pretending is current. A cell that says the pages disagree means both sentences were on the site and I am not picking a winner without a login.

CriterionOtterly.aiScrunch, brand plans
Best fit I can defend from the 19 November pagesA published card where you choose at least three engines from seven, including Claude and Gemini, and the prompt count falls as you add engines or add Claude. Seats and countries are unlimited. The login meters credits, not a fixed response total on the card.A published $250 card that names eight engines, including Gemini, Grok, and Meta, with 15,000 responses, 10 citation targets, and 10 pages optimized and 10 pages generated. Claude is not on that list. The FAQ still describes a four-engine Core.
Price I could readCards: €29, €189, and €489 a month on Lite, Standard, and Premium. Save tags: €52, €340, and €880. The annual control is labeled 15 percent off. Standard adds 5,000 credits for €65 a month. Premium adds 10,000 for €99 a month. Enterprise is custom. The meta description still says $29 for 15 prompts, $189 for 100, and $489 for 400. Help says the currency depends on location. I did not switch it.$250 a month for Core, with a 7-day free trial on the card, billed monthly. Enterprise is custom, and the table's billing cell says yearly. The pricing FAQ still says Core is $250 and then describes the older prompt pack. It also says Agency Core is $500 a month and Agency Enterprise is custom. I did not get an agency allowance table out of the pricing HTML.
Engines on a self-serve cardChoose at least 3 from ChatGPT, Google AI Overviews, Perplexity, Microsoft Copilot, Google AI Mode, Google Gemini, and Claude. All seven are marked included on Lite, Standard, and Premium. Coming soon, on the 10 November help article: DeepSeek, Grok, Mistral, Meta, and Qwen. Claude is 3 credits. Every other listed engine is 1.Core names eight: ChatGPT, Perplexity, Gemini, Google AI Overviews, Google AI Mode, Microsoft Copilot, Grok, and Meta AI. Enterprise says nine, with Claude as an add-on. The platforms FAQ says nine names including Claude, and says you pick one, a subset, or all when you create a prompt. The pricing FAQ and the tools page still say four: ChatGPT, Perplexity, Google AI Overviews, and Copilot.
Prompts, credits, responsesCards, at 4 of 7 engines: 16 prompts and 2,000 credits, 120 and 15,000, 403 and 50,000. Help, 3 engines and no Claude: 21, 161, and 537. Daily counts as 31 checks. Weekly counts as 10 in the credit article, and the pricing FAQ instead says weekly runs about four times a month. One prompt per country. Credits do not roll over and do not pay for the API.The card prints 15,000 responses a month and does not print a prompt cap. Collection is daily on Core, and daily or every 3 days on Enterprise. The pricing FAQ still prints 125 unique prompts. The trial FAQ still says the trial includes 125 prompts. The answer to "What's an AI response?" did not render. I am not inventing the conversion.
Mention, citation, and the next stepBrand visibility, sentiment, share of voice, brand profiling, and a Brand Visibility Index are marked on every plan. Domain ranking and citation source tracking are on every plan. Recommendations are 3 a week on Lite and unlimited above that. The citations article separates a gap, where competitors are cited and you are not, from the URL itself.The how-to defines brand presence as how often the brand appears, position as top, middle, or bottom, and citations as how often your site is linked. Sentiment is positive, neutral, or negative in the body, and positive, mixed, or negative in that page's FAQ. Citation targets are 10 a month on Core and unlimited on Enterprise. Fact-checked pages are 100 on Core.
What you can do after the reportA GEO audit, with a crawlability check and a content check, is included on every plan in the compare table, and the pricing help article says GEO audits have no limit. The product does not print an article allowance. Prompt research, keyword-to-prompt, query fan-out, and bulk import are on every column.Core prints 10 pages optimized and 10 pages generated a month. Agent Pages, the delivery row, is marked off on Core and custom on Enterprise. The tools page says Scrunch does not currently offer AI-generated content. The products FAQ describes creating content and serving it to agents. I am not merging those three sentences.
People, places, and accessUnlimited brands, competitors, seats, countries, and languages on every column, including Lite. Workspaces: 1, then unlimited. API and MCP start on Standard at 2,000 requests each, then 5,000. Agent analytics: none, then 200,000 events, then 1,000,000. Looker starts on Standard. SSO is Enterprise.Core is 1 brand, 5 seats, 5 competitors, 5 sub-brands, and 5 competitor sub-brands. Countries, languages, and personas say Any. Agent traffic is 1 domain, then multiple. MCP is on both columns. API keys and Data Studio are Enterprise. Google sign-in is on both. SAML and OIDC are Enterprise. Support is email on Core, and email, chat, and a shared Slack channel on Enterprise.

If the seven-engine menu and the eight-engine card are about to be two purchases, Book a Call and bring one prompt the buyer actually asks. The prompt shows which card can even track it before anyone argues about a currency toggle.

The number I would circle on Otterly is not €29. It is the engine count inside the credit budget. The card's 16 prompts are the four-engine daily figure, matching the help table's "4, no Claude" row for Lite. Drop to three engines and the same 2,000 credits are 21 prompts. Add Claude to a four-engine set and the help table drops Lite to 10 prompts, Standard to 80, and Premium to 268. Track all seven and the table prints 7, 53, and 179. The price on the card does not change when you do that. The number of questions you can watch does. A founder who hears "16 prompts" and then ticks Claude has bought a smaller set than the screenshot.

The number I would circle on Scrunch is the response cap, and then the fact that the prompt cap is no longer on the card. Fifteen thousand responses is three times the 5,000 the 10 November review on this site recorded, and the engine list is twice as long. Those two changes do not automatically mean you can track twice as much. If a response is one engine's answer to one prompt, eight engines every day is a much hungrier meter than four. The card says collection is daily. It does not say every prompt runs on every engine every day. The platforms FAQ says the user picks the platforms per prompt. Until the missing "what is a response" answer shows up in the account, I would not put 15,000 and eight engines and "daily" into one sentence that sounds like a guarantee.

Users are the procurement point that actually differs in print. Otterly says unlimited team members on Lite. A five-person team is not a reason to leave the €29 card. Scrunch Core includes 5 seats and 1 brand. The sixth person, or the second brand, is an Enterprise conversation on the table I read, unless a seat can be added without a plan change. The pricing page asks "Can I add an additional user to my plan?" The answer did not render. I am not filling it in. Projects run the other way for an agency. Otterly Standard is the first plan with unlimited workspaces. Scrunch's agency price, $500 for Agency Core, is still only on the FAQ, and that FAQ does not print the new response cap next to it. I would not brief Agency Core as "Core, but $500, with 15,000 responses" until an agency card says so.

What Otterly Includes at the Price on the Card

I read the cards the HTML displayed, which were in euros. Lite at €29 a month includes 16 prompts when the label says 4 of 7 engines, 2,000 credits, daily tracking, 1 workspace, 3 recommendations a week, brand visibility and sentiment, citation source tracking, brand profiling, ChatGPT ads tracking, and AI search optimization. API, MCP, agent analytics, and the Data Studio connector are marked not included. The save tag is €52. Twelve times €29 is €348. Fifteen percent of that is €52.20. The save line matches a 15 percent year, rounded to the euro, which is also what the annual control's label says. I did not flip the control, so I did not see the annual charge replace the monthly one. The help article says to read the pricing page or the billing tab for the exact figure, because the currency depends on location. That is the instruction I am following by refusing to translate the card.

Standard at €189 is the card the page marks as the popular step. It keeps the seven-engine menu and moves the four-engine prompt count to 120, on 15,000 credits. Workspaces and recommendations become unlimited. API and MCP are 2,000 requests a month each. Agent analytics is 200,000 events. The Looker connector is on. A pack of 5,000 extra credits is €65 a month, and the compare cell's title says the same thing. Premium at €489 is 403 prompts and 50,000 credits at the four-engine label, daily or weekly, 5,000 API requests, 5,000 MCP requests, 1,000,000 agent events, a personal onboarding session, and 10,000 extra credits for €99 a month. The save tags are €340 and €880. The same 15 percent check lands on those two lines as well. I am not turning the save tag into a second monthly price. It is a yearly discount described on the page.

Enterprise is custom. The hero card says unlimited prompts and unlimited credits, and "starting from 1,000 search prompts." The compare table prints 1,000-plus prompts and custom credits. The 10 November help table says "from 1,000" on the four-engine row and "custom" on the three-engine row. Three wordings, one plan. I would ask which of them the order form uses before anyone writes "unlimited" into a brief. SSO, enterprise compliance, custom payment terms, a customer success manager, and a set of onboarding sessions are on that column and not on the three priced cards. I did not book the demo.

The credit article is the page I would put next to the card, because the card's 16, 120, and 403 are one row of a larger table. One credit is one prompt checked once on one engine. Claude counts as three. Daily tracking counts as 31 checks a month. Weekly counts as 10. A prompt on four ordinary engines, every day, is 4 times 31, which is 124 credits. Standard's 15,000 divided by 124 is 120.97, and the table prints 120. Lite's 2,000 divided by 93, which is three ordinary engines times 31, is 21.5, and the table prints 21. Premium's 50,000 divided by 124 is 403.2, and the card prints 403. The arithmetic fits the table they published. It does not fit the meta description's 15, 100, and 400, and it does not fit the launch post's 20, 160, and 535 for three engines or its 400 for Premium at four engines. The launch post also prints the prices as $29, $189, and $489. I am using the help table dated 10 November, and the euro cards, for the counts. I am using the launch post as a page that has not been brought into line with them.

Weekly tracking is the row the help center does not settle. The credit article and the plans article say a weekly check counts as 10 in the month, against 31 for daily, and that weekly is a Premium and Enterprise switch. The pricing page's own FAQ says weekly checks run about four times a month instead of about 31, so the same credits cover roughly seven times as many prompts. Ten checks is not four checks. Thirty-one divided by ten is about three times the prompts, not seven. Thirty-one divided by four is close to the seven-times sentence, and it is not what the credit article uses. The launch post says Premium at four engines is 400 prompts daily or 1,250 weekly. Four hundred times 31, divided by 10, is 1,240, which is the neighborhood of 1,250 if you start from the launch post's 400 rather than the card's 403. I am not going to "correct" the FAQ into the credit article. Both are on the site. A buyer who is counting on a seven-fold jump should read the billing screen after the weekly switch, not this paragraph.

Countries are free in the compare table and expensive in the prompt article, and both statements can be true. Every plan says unlimited countries and languages. The prompt article says one prompt per country, so the same question in the United States and the United Kingdom counts as two prompts. You are not paying an extra country fee. You are spending the credit budget twice. Brand reports do not multiply the prompt. One prompt in three reports is still one prompt. Credits are shared across workspaces, and each workspace can use a different engine set, so the same 15,000 credits buy more prompts in a three-engine workspace than in a seven-engine one. Lite has only one workspace, so that split starts at Standard.

The free trial is real on the help page and vague on length. The plans article says no credit card is required, and that monitoring stops when the trial ends unless you pick a paid plan. The prompt table prints a trial column: 80 prompts on three engines without Claude, 60 on four, down to 26 on all seven. I did not find a number of days in the articles I used. I am not going to borrow Scrunch's 7 days and stick it on Otterly. The 4 November review on this site is also the wrong place to copy a trial length from, because the packaging under the trial has changed. Sign up and read the expiry. I did not.

A few claims should not migrate into a client report. The pricing FAQ says that in a stability study, tracking 100 prompts daily for 30 days cut the variation in brand coverage by about 73 percent compared with tracking 10. That is their sentence. I did not see the study file. The citations article says Otterly analyzed 1,400,000 citation links in September 2025 and that the most frequently cited sources were brand websites, news and media, and Reddit. I am citing the sentence. I am not re-auditing the 1,400,000. Logos on the pricing page are not a result I measured.

What Scrunch Includes at a Published Price

The brand card on 19 November is a different Core from the one this site described on 10 November, and I am going to keep both dates visible. Core is $250 a month. The bullets are 15,000 responses, 10 citation targets, 10 pages optimized, 10 pages generated, and 1 brand with 5 seats. Eight model names sit on the card. The comparison table adds collection daily, countries and languages and personas as Any, 5 competitors, 3 AI search trend topics, funnel stages, AI referrals, agent traffic for 1 domain, signals, and 100 fact-checked pages. Data export and MCP are on. API keys, Data Studio, SAML, a security review, and Agent Pages are not. Support is email. Onboarding is self-serve. There is no customer success manager on Core and no strategy review.

Enterprise is custom. The table says the billing is yearly, the response count is custom, and collection can be daily or every 3 days. Citation targets are unlimited. Pages optimized, pages generated, and pages served through Agent Pages are custom. Brands and seats are unlimited. Competitors are unlimited. Trend topics are unlimited. Claude is an add-on rather than one of the eight. API keys, Data Studio, SAML and OIDC, a security review, guided onboarding, a dedicated manager, strategy reviews, and executive reporting are on that column. The card says nine LLMs and then lists the eight plus Claude as the add-on. I am counting Claude as the ninth name, and I am not putting it on the $250 row.

The FAQ that a buyer will find by searching "what does Scrunch cost" has not moved with the card. It still says Core includes 125 unique prompts, five site audits a month, one brand workspace, five user licenses, and four platforms. It still says Enterprise includes nine platforms and names Claude in that nine, alongside Gemini, Google AI Mode, Grok, and Meta, as if those were not already on the Core card I just read. The tools page repeats the older shape: Core at $250 for 125 prompts across ChatGPT, Perplexity, Google AI, and Copilot, and a 7-day trial of that same 125-prompt, four-model set. It also says Scrunch was founded in 2023, came out of stealth in 2024, and that a funding line on the page is $26 million. I am citing the page. I did not audit the round. The same tools page says Scrunch does not currently offer general-purpose AI-generated content. The pricing card says 10 pages generated a month. Those sentences cannot both be the product description for the same Tuesday. I would ask, on the trial, what "pages generated" produces, and I would not tell a content team they bought a writer because a bullet uses the word generated.

The products FAQ is a third description, and it is a workflow rather than a price. It says the Agent Experience Platform covers monitoring, auditing, optimization, and delivery. The example it walks is labeled as an example: a brand in 45 percent of responses and competitors in 78 percent. I am not using 45 or 78 as a benchmark. They are the illustration's numbers. The delivery step says a person can approve recommendations, pull from approved internal sources, and have Scrunch serve an optimized version to agents while the human-facing site stays unchanged. That delivery sentence is the interesting one, and the pricing table puts "Pages served via Agent Pages" on Enterprise only. Core can say it generates 10 pages. The table does not say Core serves them to agents. If a demo spends the hour on the edge delivery and the order form is Core, you have not bought the demo.

The how-to is the definition page, and it opens with a warning I would put in the footnote of any deck. Scrunch says it recently shipped major product updates, so parts of the guide may not match the platform, and it points you at the in-app assistant, the help center, or a success manager while the guide is rewritten. I am still using it, because it is the page that defines the words, and I am saying the banner is there. Brand presence is how often the brand appears in AI responses. Position is top, middle, or bottom. A later FAQ on the same page also says you can see the order the brand is listed, on average. Sentiment in the body is positive, neutral, or negative. The FAQ says positive, mixed, or negative. Neutral and mixed are not the same bucket. I will not pick one until a login shows the label. Citations are how often a page on your site is linked as a source. The how-to also says brand presence is "mentioned or cited" in one sentence. I am not going to treat that looseness as permission to add the two numbers together.

Prompts, in that guide, come in two layers. A seed prompt is the question you track. A prompt variant is the platform-specific version of it. Key topics group prompts, and one prompt can sit in more than one topic. The default date view is the last 12 weeks. Filters include persona, topic, platform, funnel stage, branded versus non-branded, and country. The guide recommends starting with ChatGPT and Google AI Overviews because of reach, and it says you can track the other supported platforms from the workspace. It recommends 5 to 10 competitors. Core's table caps competitors at 5. The top of that recommendation does not fit on Core. It fits on Enterprise, where competitors are unlimited. A prompt-planning line says to multiply topic clusters by 12 to 15 questions. That is their sizing advice. It is not a cap the $250 card prints.

The response math is the part I will show as a boundary and not as a policy. The card does not say how a prompt becomes a response. If someone takes the FAQ's 125 prompts, runs them on the FAQ's four engines, every day for 30 days, that is 15,000 responses, which is exactly the new card's cap and is not the old card's 5,000. If someone takes 125 prompts and runs them on all eight engines every day for 30 days, that is 30,000, which does not fit in 15,000. If someone assumes all eight engines run every day and works backward, 15,000 divided by 8 divided by 30 is 62.5 prompts, not 125. None of those sentences is Scrunch's sampling rule. They are the edges of the numbers that are actually printed, on two pages that do not agree. The platforms FAQ says you can run one engine, a subset, or all of them on a given prompt. That sentence is how a 15,000 cap can survive an eight-engine menu: you do not have to spend every engine on every prompt. It still does not tell me whether a daily collection runs overnight on the engines you ticked, or whether "daily" is a label on a sample. I would ask that in writing.

The trial has the same split. The pricing card says start a 7-day free trial next to the $250 plan. The trial FAQ says the trial includes all Core features, including 125 unique prompts, that a credit card is required, and that the account becomes a paid subscription unless you cancel. The shorter trial FAQ says you can sign up without talking to sales, and that the trial is Core, with the four-engine examples in its body. A separate line on the pricing page, next to an Analyze action, says you can see results in 30 seconds and that no credit card is required. I am not going to merge the 30-second analyze line with the 7-day subscription trial. One of them is a look at a result. The other, on the trial FAQ, asks for a card and names the old prompt count. I did not start either one. If the trial you actually get is the new card's 15,000 responses and eight engines, the FAQ is wrong. If the trial you get is 125 prompts and four engines, the card's "Core features" sentence is doing more work than the FAQ. The account will show which one you were given. This review cannot.

Citation optimization is the off-site list, and the blog post lines up with the new card better than the pricing FAQ does. The post says the feature finds pages AI cites for the prompts you track, ranks them by an influence score and by a winnability score from 1 to 5, and can draft outreach. It says brand mentions in those sources can be checked against Knowledge Studio documents. It says suspected impersonation sites can be queued for a takedown, and that those takedowns are free and unlimited on the plans. It says the feature is included on Enterprise, and that self-serve customers get up to 10 targeted citations a month. Ten is the number on the Core card. I did not send the outreach. A draft of an email is not a placement. The influence score is theirs. I did not see the weights.

Two marketing lines stay off the slide. The pricing page says the product is trusted by 500-plus brands and agencies. I have not audited the count. The same page shows a 364 percent brand-presence line for non-branded prompts, next to a case-study link and a quote attributed to Jessie Dawson, Senior Director of Web Strategy and Operations. I did not open the case study, and I am not using the percentage as a result. The products FAQ's security list for enterprise names SOC 2 Type II, SAML, and role-based access. The pricing table's security rows are the ones I would attach to a plan. A FAQ that says enterprise customers can get a control is not a sentence that says Core includes it.

Mentions, Citations, and the Page You Do Not Own

This is the section I would make a team read before they buy either login for "AI visibility" as one number. Otterly's insights article separates the cases. Low brand coverage means the name is missing from the tracked prompts, and the article says to check the prompts before you assume the brand is invisible. A weak brand position means the name appears late when the answer lists several options. Low domain coverage means the name is in the answer and your site is not the citation. Those are three different Mondays. The citations article calls a content gap a prompt where AI cites a competitor and not you, and a citation source the third-party page the answer used. It tells you to sort by domains that get cited repeatedly, and it warns that a "top winner" ranked by percentage change can be a page that started from almost nothing. Open the count. Do not brief a percentage.

Scrunch's how-to draws a related line and then blurs it. Presence is appearance. A citation is your URL used as a source. Position is where in the answer the brand sits, in thirds, plus a ranking note in the FAQ. The products FAQ says you can track brand presence, competitive presence, position, sentiment, citations, shopping responses, search volume, bot traffic, and referral traffic. I did not find a sentence that says a mention with no URL and a URL with no mention are stored as different columns. Otterly's insights page is more explicit about that split. I am not going to pretend Scrunch's "citations" cell is the same object as Otterly's domain coverage until I have opened both rows. Similar words are how a quarterly report goes wrong. If the slide says "citations," it has to say which vendor's definition, and whether the URL was the brand's.

The fix list is gated differently, and neither gate is a published article on the domain that got cited. Otterly includes citation tracking on Lite and limits recommendations to three a week on that plan. Unlimited recommendations start on Standard. The citations article says you can export the URLs where competitors are mentioned and you are not, and hand them to content, PR, and social. The export is a list. Scrunch meters the list: 10 citation targets a month on Core, unlimited on Enterprise, with a winnability score and a drafted note. Ten targets is a sample of the sources a real category will show you. It is not a program that places the piece. The GEO audit on Otterly is the on-site half of the same problem: can a crawler fetch the URL, is the content static or injected, is there structure a model can read. The audit help article says the query fan-out view shows how an engine expands one prompt into related questions. That is a research output. It is not a page on a publisher's site.

When the cited URL is a third-party article the tracker cannot place, Offsite Posts writes and publishes industry articles about a brand on other sites, and the site says the posts, mentions, and links in a package are guaranteed while rankings, traffic, and AI mentions are not.

Ads and shopping are easy to over-assign. Otterly marks ChatGPT ads tracking as included on every plan, including Lite. I did not connect an ad account, and the pricing page does not say the same switch covers ads inside Gemini or Copilot. Scrunch's products FAQ lists shopping responses as something the workflow can track. The pricing table I read does not give shopping its own row. I would not tell a Core buyer they bought a retail report because an example workflow named shopping, and I would not tell an Otterly buyer that ads tracking is a substitute for opening the answer.

Crawl and bot data should not be merged into the prompt report. Otterly's agent analytics, which is the log of agent activity against your site, starts on Standard at 200,000 events and moves to 1,000,000 on Premium. It is a separate meter from credits. The plans article says credits do not pay for it. Scrunch includes AI agent traffic for one domain on Core and for multiple domains on Enterprise, and it includes AI referrals on both. A bot that fetched a URL is not an answer that cited it. A referral is not a mention. If the Monday slide has one number on it, it is the wrong slide.

Who Should Buy Which

Pick from the engine, the unit, and the person who reads the answer.

  • You need to choose Claude or Gemini on a self-serve plan, and you can accept a credit budget. Otterly Lite, if 2,000 credits, one workspace, and three recommendations a week can hold the month. Read the help table before you assume the card's 16 prompts survive Claude. Standard is the first card with unlimited workspaces, the API, and the connector. The Otterly review is the 4 November picture. This page is the later card.
  • You need Grok or Meta on a published price, and you do not need Claude in that price. Scrunch Core is the card that names them. Otterly's 10 November engine article still lists both as coming soon. Do not buy Otterly for an engine the help page has not switched on, and do not buy Scrunch Core for Claude. Claude is the Enterprise add-on on the table I read.
  • You need the FAQ and the pricing card to describe the same Core. They do not, on the pages I opened. Use the card for the eight names and the 15,000 responses, and treat the 125-prompt FAQ as a page that has not been updated. The Scrunch review records what the card said on 10 November. I am not asking that review to do today's job.
  • You need Perplexity and Copilot on either entry plan. Both menus include them. That tie is not a reason to ignore the unit. Otterly will spend credits. Scrunch will spend responses, under a definition the pricing page asks for and does not answer in the copy I could read.
  • You need a second brand, or a sixth seat, on the entry invoice. Otterly's unlimited seats and unlimited brands are on Lite. A second workspace is Standard. Scrunch Core is one brand and five seats. The second brand is the Enterprise column on the table I read. The Peec comparison is the wider menu if a pick-your-models card is the actual shortlist. I am not rebuilding Peec's invoice here.
  • You need the model checked against a fee schedule. Scrunch Core includes 100 fact-checked pages and 10 citation targets. Otterly's recommendations on Lite are three a week. On a live account, Austin Heaton still opens the stored answer before a citation rate goes into a deck. Neither cap does that read for you.

A tool subscription and an engagement are different bills. This page is only the tool. The four-way shortlist with Profound, Peec, and Semrush is Profound vs Peec AI vs Otterly vs Semrush. Prices on that page are not a substitute for the two pricing URLs I opened on 19 November.

What I Could Not Confirm

  • A stored answer inside either product. I read the public pages. I did not create an Otterly account, and I did not start Scrunch's trial or the 30-second analyze flow.
  • Whether Otterly's billing screen shows euros or dollars for a given account. The cards I downloaded were in euros. The meta description and the launch post use dollars at the same 29, 189, and 489. The help article says the currency depends on location. I did not flip a control.
  • Whether Otterly's weekly run counts as 10 checks or as about 4. Both sentences are on the site. The prompt jump is very different. I did not switch a workspace to weekly.
  • What Scrunch means by one response. The pricing page asks the question. The answer was not in the HTML I pulled. The 125-prompt FAQ and the 15,000-response card do not, by themselves, pick a formula.
  • Whether a new Scrunch trial provisions the eight-engine card or the four-engine FAQ. The card says a 7-day trial of the plan. The trial FAQ still says 125 prompts and four platforms, and it says a card is required. The analyze line says no card is required. I did not sign up.
  • What "10 pages generated" produces, and whether Core serves those pages to agents. The tools page says generated content of that general kind is not offered. The products FAQ describes delivery. The Agent Pages row is Enterprise.
  • Scrunch's agency allowances. The pricing FAQ prints Agency Core at $500 and Agency Enterprise as custom. It does not print a response cap, an engine list, or a seat count for those plans. The brand table is the invoice I could read.
  • The customer logos, the 500-plus line, the 364 percent case-study line, Otterly's 73 percent stability sentence, and the 1,400,000 citation count. They are on the pages. I am not using them as outcomes.

Frequently Asked Questions

Which is the better fit for AI visibility tracking?

Otterly.ai is the better fit when you choose at least three engines from a menu of seven that includes Claude and Gemini, and you can live with a credit budget that shrinks the prompt count as you add engines. Scrunch is the better fit when the Core card's eight engine names, including Gemini, Grok, and Meta, are the list you need on one published plan, and Claude can wait. Neither tool promises a citation. The prices, the credits, and the response cap are in the pricing table in this article. I did not create an account on either product.

Which engines can I track before a sales call?

Otterly's pricing table marks all seven engines on Lite, Standard, and Premium: ChatGPT, Google AI Overviews, Perplexity, Microsoft Copilot, Google AI Mode, Gemini, and Claude. A help article dated 10 November 2026 says each workspace needs at least three, and that you do not buy engines as add-ons. Grok and Meta are on that article's coming-soon list. Scrunch's Core card names eight engines, including Grok and Meta, and does not name Claude. Claude is an Enterprise add-on on that table. Scrunch's pricing FAQ still names only four platforms for Core. I am using the card for the menu, and I am not deleting the FAQ.

How do the prompt and response caps compare?

Otterly's cards print a prompt count next to a credit budget, and the help table prints a different prompt count for each engine mix. Daily and weekly do not use the same number of checks, and two Otterly pages do not describe weekly the same way. Scrunch's card prints a monthly response total and does not print a prompt cap. Scrunch's pricing FAQ still prints a prompt total from an older description of Core. The two products are not counting the same object. Those caps are in the pricing table in this article. I did not run a prompt.

Does either tool write or publish the page?

Otterly's pricing table does not include an article allowance. Its citations article says you export the URLs and hand them to a content or PR team. Scrunch's Core card includes a monthly count of pages optimized and pages generated, and it leaves Agent Pages, the delivery row, on Enterprise. Scrunch's own tools page says the product does not currently offer AI-generated content, while the products FAQ describes creating a page and serving it to agents. I did not generate a page in either login. A target list is not a published URL.

How does each one define a mention and a citation?

Otterly's insights article separates brand coverage, which is the name in the answer, from domain coverage, which is your site used as the source. The citations article calls a gap a prompt where a competitor's URL is cited and yours is not. Scrunch's how-to defines brand presence as how often the brand appears, and citations as how often your site is linked. That how-to also warns that parts of the guide may not match the product after a recent update. I did not open a stored answer in either product, so I have not checked the definitions against a live row.

Where do I read the current prices?

The figures I could verify are in the pricing table in this article, with the page each figure came from. Otterly's cards displayed one currency, the meta description on that URL still describes another, and the help article says the currency depends on location. I did not switch a control. Scrunch's Core price is on the brand card. A separate FAQ still prints an agency price without the card's response cap. If a contract is already signed, the contract is the price.

Can I add another engine or another country on the entry plan?

Otterly's entry plan already includes the seven engines. You choose at least three, and you can change them later. The help article says another engine is not a separate purchase. Countries are unlimited on the compare table, and the prompt article says the same question in a second country counts as another prompt. Scrunch's Core card says countries, languages, and personas are any, and it does not list Claude among the eight engines. Adding Claude is the Enterprise add-on on that table. The country row and the engine row are in the pricing table in this article.

Who should skip both tools?

Skip both when the job is a citation guarantee, because a prompt report is a sample of answers to questions you tracked, not a promise that an engine will mention you. Skip Otterly's entry plan when you already need Grok or Meta, because those names are still on a coming-soon list. Skip Scrunch's entry plan when you already need Claude, because the card puts Claude on the custom plan. Skip unattended publishing on any page that states a fee. A gap export is not a published URL.

Read next:

If the engine list, the credit budget, and the person who opens the cited URL are three different purchases, Book a Call and we will put the prompt, the card, and the missing URL on one page before anyone prepays a year of tracking the team will not read.

BECOME THE COMPANY AI RECOMMENDS.

Your next qualified lead starts with an answer.

Get more citations. Compete for the prompts that matter. Become the top choice in your category.

Book a Call