Most GEO comparisons rank on capability. That ranking is useless in week one, when the real constraint is different: can you get a number out of this thing before your next planning meeting, or does it require a discovery call, a security review and a quote?
Access does not track price. Six of the nine platforms on every 2026 shortlist will show you something today for nothing. One publishes a price but still routes you through a demo. One publishes no price at all. Those last two get collapsed into “sales-gated” constantly, and they are different problems.
Every figure below is marked: [vendor page] read on the vendor’s own page · [vendor code] from its shipped price module · [archived page] an Internet Archive snapshot · [secondary] a reputable third party · [not published] no vendor figure exists · [our arithmetic] a number I calculated, not a vendor statement.
The access ladder
Tool | Fastest path to real data | Entry price the vendor publishes | Provenance |
|---|---|---|---|
Profound | Free AEO Report — three engines, no signup | $99/month Starter, month-to-month | [vendor] |
Athena (AthenaHQ) | Essential tier — free, “Includes 300 credits”, 5 engines | Free; Starter $295/month | [vendor] + [vendor JS] |
Semrush (an Adobe company) | Free AI Visibility Checker, logged out | AI Visibility Base $99/mo per domain; Semrush One Starter $199 | [vendor] + [vendor JS] |
Ahrefs | Free AI Visibility Checker; Custom Prompts allowance in every paid plan | Lite $129/mo monthly, $108 annual | [vendor] |
Scrunch (a Sitecore company) | 7-day Starter trial, “no credit card required” | $250 per month (billed annually) or $300 month-to-month | [vendor, one summarised read] |
Otterly.AI | Free trial — the vendor states both 7 days and 14 days | $ 29 /month Lite | [vendor] |
Peec AI | None published — self-serve purchase only | €85 /mo in Europe, $95 /mo elsewhere | [vendor] + [vendor JS] |
Evertune | None — “Request a demo” is the only route in | $800 / month Pro | [vendor] |
Brandlight | None | Not published — /pricing returns HTTP 404 | [not published] |
Checked 1–2 September 2026. Scrunch is one of the few here that publishes both billing terms in a single string, with no toggle to operate — on one summarised read.
Sales-gated and unpriced are two different problems
Evertune publishes $800 / month for Pro: no billing toggle, no billing-term qualifier, no annual discount — and no self-serve checkout, since the only route in is Request a demo. You can size the line item before you speak to anyone; you just cannot buy it alone.
Brandlight is the other case. Its /pricing URL returns a 404 and its homepage carries no figures. The ~$199/month entry price circulating in comparison tables traces to third-party aggregators and to sat.brandlight.ai, a vendor-owned but programmatically generated subdomain rather than product documentation.
The honest entry is not published — and the same goes for engine coverage, where the homepage names six engines (ChatGPT, Gemini, Perplexity, Copilot, Grok, Google AI) and states no total. With Evertune you can decide whether the call is worth taking. With Brandlight, the call is the price discovery.
Profound’s self-serve tier is month-to-month
$99/month is Profound’s default-displayed Starter price. “Billed yearly” is a per-card toggle that ships in the off state; switch it on and the card reads $82.50/month. Both terms are published, and Profound publishes no yearly total on either setting. Multiply if you want one, but that is your arithmetic, not Profound’s price — which is exactly how the $1,188 commitment attributed to this tier in comparison tables came about.
So the thing to notice about $99 is not a commitment. It is the contents. Growth sits at $399/month, or $332.50/month with the toggle on, and what the two tiers buy is this.
Starter vs Growth | Starter | Growth |
|---|---|---|
Engines | One — ChatGPT | Three — ChatGPT, Perplexity, Google AI Overviews |
Unique prompts | 50 | 100 |
Responses | 1,500 | 9,000 |
Seats | One | Three |
Starter also caps you at one language and region.
Engines, API, SSO, exports, multi-language and multi-region all sit above Starter. A $99 baseline is a one-engine baseline. That, not the billing term, is the constraint to price — see what these tools actually measure.
Read the toggle before you read the number
Most pricing confusion here is a rendering default, not a trick. Profound ships its annual toggle off, so the first number you see is month-to-month. The others vary.
Peec ships its toggle on, so the first number a European visitor sees is the annual one. [vendor page] + [vendor code]
Peec | Month-to-month | Toggle on |
|---|---|---|
Europe | €85 | €70 |
Elsewhere | $95 | $80 |
Tables labelling €85 “billed yearly” have it backwards.
Otterly shows Lite at $ 29 /month monthly and $25 /month annual, with no “billed annually” string anywhere on the page. Its “15% off” badge holds for one tier of three.
Otterly tier | Discount the annual price gives |
|---|---|
Lite | 13.8% |
Standard | 15.3% |
Premium | 13.7% |
The badge implies about $415.65 against a page price of $422. A 23 June 2026 snapshot carried those same annual prices with no badge at all. [archived page]
Semrush’s standalone AI Visibility Base is $99/mo per domain monthly and annually: the annual discount on that tier is zero.
None of it is deceit; it is structure. But a screenshot of a pricing page is not a price.
What “free” actually gets you
Free path | What it is | Cost of entry |
|---|---|---|
Profound AEO Report | One-off report across three engines | No signup |
Semrush AI Visibility Checker | Logged-out checker | No signup |
Ahrefs AI Visibility Checker | Logged-out checker | No signup |
Athena Essential | The product, metered — 5 engines, “Includes 300 credits” | Signup |
Scrunch Starter trial | 7 days of Starter | “No credit card required” |
Otterly trial | 50 prompts — over three times the 15 in its paid $29 Lite tier | Signup; card requirement not captured |
Ahrefs’ layer is bundled rather than free: every paid plan carries a Custom Prompts allowance, starting on Lite at 5 tracked prompts and 150 checks, where a check is one prompt × one platform × one location. Standalone packages start at $50/mo. It is that $50 package, not the AI Visibility Index, that carries the allowances comparison tables routinely attach to the Index instead.
Standalone package | Figure |
|---|---|
Prompts | 83 per day |
Checks | 2,500 per month |
Overage | $0.020 per check |
The Index is a separate add-on, $199/mo for select platforms and $699/mo for all, and its $199 card carries no feature list at all. Ahrefs runs no trial and states that it never discounts — on a page that prices Lite at $129/mo monthly and $108 annual.
Semrush has a related catch: extra seats are $45, $80 and $100, and none of them include AI visibility, which the page says must be bought separately per user — with no number attached. [not published]
Athena’s free tier, and the two things it does not say
Essential is listed as Free, with $25 free credit and a tooltip reading Includes 300 credits. It is the most generous genuinely free thing here, with two gaps to carry into any plan built on it.
Recurrence is not stated. Starter’s line reads “$300/month free credit”. Essential’s carries no “/month”, and the comparison table’s “Credits /month” row omits the free tier entirely. Whether the 300 credits arrive monthly or once, Athena does not say — so do not assume they refill.
Credits are not a uniform unit. The rate depends on the engine.
Cost per response | Engines |
|---|---|
One credit (six engines) | ChatGPT, AI Overviews, AI Mode, Perplexity, Gemini, Copilot |
Five credits (five engines) | Grok, Claude, DeepSeek, Meta AI, Mistral |
The vendor says it both ways on its own pages: the /plans table states flatly 1 credit = 1 AI response, the credit calculator says the heavier five cost 5 credits per response. For the free tier the multiplier is moot, which is itself the limit — Essential’s five engines are all one-credit engines. The four gated behind Starter (Grok, AI Mode, Claude, DeepSeek) carry the tooltip Available on the Starter plan; Meta AI and Mistral are costable in the calculator but unlisted on the plans page. A free baseline is 300 responses and can never include Claude or Grok.
Starter is $295/month. The annual figure is $245/month, billed annually — a string that exists only in the site’s JavaScript bundle, never in the served HTML. Athena publishes no annual total, and its own code carries a “2 months free” badge that does not reconcile against twelve months at $245. [vendor code] + [our arithmetic]
The page has also moved under itself. Starter was $95/mo under a promo running to 8 June 2026, and $295/mo by 29 July, on the same credit allowance [archived page] — so third-party tables may be describing the old page.
A five-day evaluation plan
Day 1 — Write the prompt list before you open a tool. Twenty to forty prompts from what your sales team actually hears. Include the unflattering ones: “is [brand] worth the price,” “[brand] vs [competitor],” “why do people leave [brand].” The list outlives the platform: it is the asset, the tool is the meter.
Day 2 — Get a free baseline. Profound’s AEO Report, the Semrush and Ahrefs checkers, then Athena’s free tier with your list. Record the raw answers, not just the scores.
Day 3 — Read the answers as a human. Read twenty engine answers end to end, mark every factual assertion — prices, plan limits, integrations, certifications, policies — and check each against what is currently true. Note also whether anything you are reading is a paid placement: OpenAI began showing labelled sponsored placements to logged-in US Free and Go users on 9 February 2026 and opened a self-serve ChatGPT Ads Manager on 5 May 2026.
Day 4 — Check your own server logs. Which AI crawlers fetched which URLs this month? No vendor required, and it is the only observation here that is not a sample.
Day 5 — Decide what you are buying. By now you know whether your problem is absence, misrepresentation or instability. Those have different buyers on this list, and the third is not solved by tracking — only by more sampling.
What a trial cannot tell you
Tool | Samples per prompt per cycle | Provenance |
|---|---|---|
Evertune | 100 per model — the same homepage also says “up to 100x per model” | [vendor], internally inconsistent |
Peec | 1 — “executes each prompt once every 24 hours on every AI model you’ve selected” | [vendor] |
Athena | 1 credit = 1 AI response; ×4 with fan-out on | [vendor] |
Profound | 1 per prompt per engine per day | [my arithmetic] on published caps |
Ahrefs | Custom Prompts: 1 check = 1 prompt × 1 platform × 1 location. Index: not published | [vendor] |
Otterly | Not published; a binary daily reading implies one | [vendor docs] |
Scrunch · Semrush · Brandlight | Not published | [not published] |
Eight of nine either take one draw per cycle or decline to say. Set that against the IAB’s August 2026 position — that any metric derived from one response per query reflects a sample of one — and the comparison stops being a feature checklist and becomes a methodology audit. The instability is not a vendor failing: run-to-run variation is driven mainly by batch non-invariance in the inference stack rather than by temperature, and nobody here controls that stack.
The arithmetic is worth doing once — plain binomial maths throughout. [our arithmetic]
Answers sampled | What it buys you |
|---|---|
72 | A visibility rate near 25%, 95% interval roughly ±10 points |
about 300 | The same rate, tightened to roughly ±5 |
roughly 294 per period | Detects a move from 20% to 30% at standard power |
A 50-prompt daily reading does not get there inside a trial window.
So a trial tells you whether the platform is pleasant, whether the prompt coverage fits your category, and whether the answers about you are wrong. It cannot tell you whether the visibility line moved. Anyone showing you a five-day trend line in a demo is showing you noise with a slope on it — testing that properly is a separate protocol.
Questions for the two that need a call
Evertune and Brandlight will not show you data first. That is a legitimate model; it just means you arrive without leverage unless you bring questions. The first four suit every vendor on the list.
Will you run my prompt list, not your demo list, in the call? A demo on prompts the vendor chose tells you nothing about your category.
How many times is each prompt sampled, per engine, per cycle? One is the most common honest answer, and it sets the noise floor under every number you will ever see.
Do you separate sponsored placements from organic mentions, and do you sample logged-in or logged-out? Since ChatGPT started carrying ads, that decides what a “mention” is. Nobody in this set publishes an answer.
What do you check my brand’s claims against? See the accuracy gap — usually a document set you maintain yourself.
What is the shortest contract, and who owns the roadmap at renewal? Two of the nine changed hands this year. Adobe completed its acquisition of Semrush on 28 April 2026 and launched Adobe Brand Visibility on 17 June, merging Adobe LLM Optimizer with Semrush’s AI Optimization. Sitecore announced its acquisition of Scrunch on 3 June 2026, and neither party disclosed terms: the ~$225M in circulation is Bloomberg’s, from people familiar with the matter, with both companies declining to comment. [secondary, unconfirmed]
The honest summary
If you want to know whether you have a GEO problem, you can find out this week for $0: Profound’s AEO Report, the Semrush and Ahrefs checkers, Athena’s free tier, a Scrunch trial that, on its own pricing page, asks for no card, and your own server logs.
If you want to know whether a platform will fix it, no free tier will tell you and neither will five days — that takes a protocol and about a month. What the free layer is good for is arriving at the call already knowing what the engines say about you. For what the tiers above it cost, see the pricing breakdown.
How this was verified
Re-verified against primary sources on 1–2 September 2026. Several claims in the 1 September version did not survive.
What changed. Profound’s Starter was described as $99/month billed yearly, implying $1,188 up front. That was wrong in both directions. $99 is the month-to-month price, and the yearly toggle ships off. The annual price is $82.50/month, and Profound publishes no yearly total — so the $1,188 was multiplication presented as a vendor figure. Peec’s “7-day trial” came from secondary sources; Peec publishes no trial length. Scrunch was listed as sales-gated; it publishes both billing terms and a 7-day trial with no card required. Brandlight’s widely cited ~$199/month was removed: /pricing is a 404 and no vendor page carries a figure. Evertune moved out of the unpriced group, since it publishes $800/month and is demo-only. The free-tier layer was rebuilt on Profound’s AEO Report, the two logged-out checkers and Ahrefs’ bundled allowances; the HubSpot grader detail was dropped rather than repeated unverified.
Contradictions are reported, not resolved: Athena’s flat “1 credit = 1 AI response” against its own 5-credit engines; Evertune’s “100 times” against “up to 100x”; Otterly’s two trial lengths and its 15% badge; Scrunch’s stale Core/Enterprise FAQ against its current Starter/Growth page.
Known limits. Scrunch’s tier contents and trial terms rest on one read that went through a summarising fetch, so they are hedged. Otterly’s card requirement was not captured. The recurrence of Athena’s 300 free credits, Evertune’s prompt unit and every Enterprise price except Otterly’s floor are unpublished. The batch non-invariance point is carried without a citable URL, so it is stated qualitatively. Every annual total here would be my multiplication, and none is printed. The annual per-month prices ($82.50, $332.50, $245, $25, €70, $80, $108) are vendor-published. These nine recur on 2026 shortlists; the IAB counts more than 20 providers, so read the set as a shortlist, not a census. Knowledge Company publishes those limits rather than smoothing them over, on the same basis as the claim register at knowledgecompany.ai/claims.
Sources: Profound pricing · Semrush AI search pricing · Adobe completes Semrush acquisition · Ahrefs Custom Prompts · Ahrefs AI Visibility Index · Peec pricing · Otterly pricing · Athena plans · Athena credit calculator · Scrunch pricing · Brandlight · Evertune · OpenAI: testing ads in ChatGPT · Bloomberg on the Sitecore–Scrunch deal