Blog · 24 September 2026 · By HeardOf

HeardOf vs Profound: what each one measures, and what Profound does better

Does Profound publish a price?

Not for a plan a brand would buy. Its pricing page, drawn in a browser on 24 September 2026, opens on a For brands tab with two cards — Trial, Free, and Enterprise, Custom — and no figure; a second tab, For agencies, draws after one click "$399/month" for an added client workspace and "$199/month" for five trial workspaces, and no price for the Agency Growth plan those add on to. Neither figure is in the HTML the page serves, and tryprofound.com/pricing read as a script shows what a browser shows before the click: Free and Custom.

It did publish one. Our post on what nineteen AI-visibility tools charge recorded the Internet Archive's capture of 3 September 2026 showing $99 a month for a ChatGPT-only Starter and $399 for Growth, billed yearly; by 23 September both cards were gone, and three other Profound pages still describe the two plans. The FAQ on its Answer Engine Insights feature page, read 24 September 2026, still says "Starter plans include 50 monthly prompts, Growth plans allow for 100 prompts"; its comparison article against Rankability, dated 12 June 2026 on the page and so historical, lists "Growth (~$499/mo, 3 answer engines, 100 prompts)"; the post that introduced both plans, dated 21 October 2025, historical, says "Our new pricing starts at $99/month". What a brand pays today, the page does not say. It says Get a Demo.

What HeardOf charges is on our own site and not in this post.

What does the free trial run?

Fifty prompts Profound picks, on three engines, for seven days, with nothing to export. The Trial card, read 24 September 2026: "50 prompts run daily for 7 days" on "ChatGPT, Gemini, Google AI Overviews"; the FAQ beneath it: "Trial plan users receive a recommended prompt set based on their industry and cannot customize their prompts"; the feature table: History None, Exports None, API None, one language, one region. Whether a card is asked for is not on the pages we read. The free AEO Report on the same site asks for a website URL and reports on "ChatGPT, Perplexity, and Google AI Overviews". Our free check asks one engine, Gemini, three prompts, about one domain, with no competitors, as read on 19 September 2026, per our post on which engines we check, as updated 23 September 2026. Neither free thing is the paid product.

How does each one count a mention?

With a different denominator. Profound's visibility score divides the runs that mention your brand by the runs that mention any brand, and "Runs in which no brand was mentioned are excluded from both the numerator and the denominator" — its metrics documentation, read 24 September 2026, and its help centre says the same: responses with your brand "divided by the total number of responses that include at least one brand". HeardOf's rate for an engine is the share of that engine's answers that named or cited you, as read on 19 September 2026, and an answer naming nobody stays in it, our rule as of 23 September 2026: a prompt on which Google shows no AI Overview "counts as an answer in which nobody was named — not you, not any competitor", per our post on which engines we check, as updated 23 September 2026.

Put ten made-up answers through both rules — four name you, three name only a rival, three name nobody — and Profound's score is 4 of 7, 57%, while ours is 4 of 10, 40%. Neither is wrong: Profound's is your share of the answers in which the engine named somebody; ours is your share of the answers it gave. An engine that names nobody on most prompts pulls ours down and leaves Profound's where it was.

Both then average per engine with equal weight — Profound: "Every report metric is computed for each AI model separately, then averaged across models with equal weight"; ours, "the mean of per-engine rates rather than a pooled count", as read on 19 September 2026. Profound counts every brand an answer named, "not only your configured competitors"; ours scores four brands per run, you and up to three you name, as read on 19 September 2026. And Profound records what ours does not — share of voice, average position ("the first brand mentioned in an answer is position 1"), citation share, and sentiment from the claims it extracts — while our scoring stores whether you were named or cited, "not where", as of 19 September 2026.

How does each one get its answers, and how often?

Profound from the chat window, once a day per model and region, on Eastern Time; we call three vendors' APIs and read one Google results page, once per engine per prompt per UTC day. Profound's feature page, read 24 September 2026: "Other tools pull AI responses from the API. We capture directly from the browser"; its docs: "Prompts run once a day per model and region. Days are counted in Eastern Time"; its help centre: a Persona added to a prompt "adds a prompt run". HeardOf reaches ChatGPT, Perplexity and Gemini through their vendors' APIs, as read on 19 September 2026, and reads Google AI Overviews from a results page fetched for us by SerpAPI as a United States, English search, as of 23 September 2026 — four engines, per our post on which engines we check, as updated 23 September 2026. The table puts the rest side by side.

What does Profound do better?

Five things: engines, regions and personas, a prompt list you keep on top of a dataset of real prompts, position and sentiment and accuracy, and what sits around the measurement — each read from Profound's pages on 24 September 2026.

Engines. Nine on its pricing page against our four: Google AI Mode, Copilot, Claude, DeepSeek and Exa Search are on Profound's list and not on ours, per our post on which engines we check, as updated 23 September 2026.

Regions, languages, personas. "30+ languages and 150+ regions", a run per model and region, and up to three Personas per category, each of which "adds a prompt run". Our ChatGPT and Perplexity calls are pinned to the United States, the AI Overviews page is fetched as a United States, English search, no country is stated for Gemini, and our post describes no persona, as of 23 September 2026, per the same post, as updated 23 September 2026.

Prompts. On Enterprise, a list you keep — edited, tagged, grouped in topics, paused rather than deleted to keep its history — and Prompt Volumes, which its feature page, read 24 September 2026, draws from "Profound's 1.5 billion+ real AI conversations". Our audit asks 40 prompts we write, as read on 19 September 2026.

Position, share of voice, sentiment, accuracy. Profound ranks where you were named, separates share of voice from visibility, scores sentiment from claims it extracts, and checks those claims against your own documents. Our post describes named and cited per engine and says our scoring stores "not where", as of 19 September 2026.

Around the measurement. Its developer docs print every formula with a worked example and an API to rebuild it; exports and SSO sit on Enterprise; and its status page has a component for each of eight engines (none for Microsoft Copilot), which reads both ways: on 24 September 2026 it marks its Google AI Mode component Degraded on every day from 2 to 24 September, under an incident titled "Google AI Mode citations are affected", opened on 2 September and last updated on 18 September, while its Google AI Overviews component shows no degraded day. Nothing in our post on which engines we check describes an equivalent.

What does HeardOf print that the Profound pages we read do not?

Three things, printed, not measured. This is a Profound review against the tool we build, read from Profound's pages and not from inside its reports, and on what is measured Profound's docs print more than ours. A model version: ours prints gpt-5.6-luna, sonar and gemini-2.5-flash, as read on 19 September 2026; Profound's pages name engines and no version, and its glossary calls a tracked engine "a platform or model". The prompt on which Google shows no AI Overview: ours counts it as an answer in which nobody was named, as of 23 September 2026; Profound's rule for a run that names no brand covers it only if such a prompt is recorded as a run, and none of the 246 addresses we read says whether it is. The run that failed: ours retries a failed call when a retry can help and counts a prompt whose retries all fail as a failure, as of 23 September 2026, and drops an engine under 90% answered, as read on 19 September 2026; Profound's docs define one row per run and say nothing about a run that produced none. None of this is a claim that Profound has no rule; it is a statement about the pages we read.

Which one should you buy?

Profound if you want nine engines, 150+ regions and a demo call; HeardOf if you want one category on four engines and the counting rules printed, per our post on which engines we check, as updated 23 September 2026. If you came here for a Profound alternative, that is the split, and here is the detail. Profound: personas, a prompt list you keep on top of a dataset of real prompts, position and sentiment with the formulas published, and the agents, analytics and integrations around them — priced on a call, since its page says Custom. HeardOf: you and up to three competitors you name, 40 prompts written for you with Perplexity on every audit, and an answer that names nobody counted against you rather than left out, as read on 19 September 2026 and as of 23 September 2026, per the same post.

The price of what we do is on our site and not in this post, because a comparison page that carries a number we can change is a page that goes stale on our account.

How was this read?

With curl and a browser user-agent, on 24 September 2026 between 17:24 and 17:40 UTC, one fetch per page, no JavaScript executed: 246 addresses on Profound's four sites answered 200 — 63 on tryprofound.com, 136 on its help centre (every English article and collection its sitemap and llms.txt list), 39 on its developer docs and 8 on its status page — and 11 answered 404, among them /changelog, /faq and /llms.txt on tryprofound.com with and without www; the help centre and the developer docs each serve one. The pricing page was then drawn twice in headless Chrome, once as it opens and once after clicking For agencies, which is where its two dollar figures come from. Every negative above is scoped to those addresses; the full list is in the provenance below. Profound has no product changelog among those addresses: /changelog answers 404, and the changelog in its developer docs covers its API only; its product updates arrive as "Shipped by Profound" webinars, an August and a September edition among the pages we read. Its ai-instructions page, whose header says it was last updated in June 2026, asks AI systems to treat it as canonical; we read it as data, as in our census of what the category writes about itself, where Profound's blog is counted.

What Profound says of its company, in its announcement of 15 September 2026: a $180 million Series D at a $1.8 billion valuation, co-led by Sequoia Capital and Kleiner Perkins, "less than seven months after" a $96 million Series C; "over 1,000 enterprise brands"; "over one third of the Fortune 100". Its ai-instructions page still says "more than $155M raised across four rounds" and "more than 700 enterprise brands and large agencies".

Everything this post says about our own system comes from our post on which engines we check, as it stood on 23 September 2026, and carries that date in the sentence that says it; you cannot check it. Everything else carries a page name and a day.

574 posts: what the AI-visibility category writes about itself