Comparisons

Profound review: the deepest tool in the category, sold one engine at a time

By Arnav Mukherjee, founder of TofuBofu · September 7, 2026

We compete with Profound, so read this with that in mind. I'm writing it anyway because the honest version is more useful to you than the flattering one, and because the last time I looked closely at their pricing I got a detail wrong and had to correct it publicly.

Here's what happened. Profound's pricing grid puts the tier names across the top and the features down the side, and if you read the page as linear text the engine coverage lines up against the wrong columns. Our first pass concluded the mapping couldn't be pinned at all and we wrote a rule telling ourselves not to claim it. That rule was worse than the missing fact, because it made the next person hedge a sentence we could have verified. Someone re-measured by the geometry of the grid, matching each value to its column by pixel position, and the mapping came out clean. So the numbers below are pinned that way, and I re-read the whole page again on 7 September 2026 before publishing this.

The short version: Profound builds the two best things in this category, and it sells engine coverage as the thing you climb the price ladder for. Whether that's a good deal depends entirely on which problem you have.

What it is, and what it costs

Profound is an AI visibility platform: it tracks how your brand shows up when people ask AI engines questions, and it's built for marketing teams rather than for a founder poking at it on a Sunday. Read at their own pricing page on 7 September 2026, in the brands view with the billed-yearly toggles off:

• Starter, $99 a month. "ChatGPT tracking only", 50 prompts tracked, 100 Agents credits a month, email support.
• Growth, $399 a month. "3 Answer Engines tracked", 100 prompts tracked, 400 Agents credits a month, email support.
• Enterprise, custom. "Up to 9 Answer Engines tracked", multiple companies tracked, a tailored prompt plan, dedicated Slack support, SSO/SAML and SOC2 compliance.

Switching the yearly toggle on advertises two months free, so the figures above are the more conservative month-to-month pair. One point of fairness worth stating plainly: Profound was quote-only for most of its life, so plenty of articles still describe it as demo-only pricing. Those were true when they were written. They aren't now, and at the time of writing our own pricing page is one of the pages still carrying the old framing, which we're fixing.

Agent Analytics is the best thing in the category

This is the part I'd pay for, and I'd rather say so than bury it. Every tool in this category, ours included, is good at telling you whether an engine mentioned you. Almost none of them can tell you whether that mention sent a human to your website. AI referral traffic is hard to see: the referrer is often stripped, the visit lands as direct, and your analytics file the most valuable traffic you have under nothing.

Profound's answer is to go under the analytics layer and read the infrastructure. Agent Analytics lists integrations with Akamai, AWS, Cloudflare, Fastly, Google Analytics, Google Cloud Platform, Netlify, Vercel and WordPress, across unlimited domains. If your site already sits behind one of those, you have the raw evidence sitting in logs and Profound turns it into attribution. That's the right architecture for the problem, it's expensive to build, and it's better than what we ship. We tag AI referrals without cookies and count conversions, which works and is a smaller idea.

The second thing worth the money is Prompt Volumes, which they describe as seeing what users are prompting AI answer engines. Demand data for prompts barely exists. Keyword tools tell you what people type into Google, and there's no equivalent for what people ask ChatGPT, so most of this category, again including us, infers buying questions rather than observing them. Anyone selling real observation of that is selling something scarce. Their Agents product, autonomous workers for content generation and optimisation, plus Sheets for running those at scale, is the third leg, and it's where the credit allowances on each tier get spent.

Where it falls short: you buy engines one at a time

The constraint isn't hidden and it isn't a trick. It's the shape of the product. At $99 the label on their own card reads "ChatGPT tracking only". Your buyer doesn't know that's the tier you bought. They ask whichever assistant is open, and on a Starter plan every engine except one is a blind spot you're paying to maintain.

Going to $399 buys "3 Answer Engines tracked", which their grid maps to ChatGPT, Perplexity and Google AI Overviews. Worth knowing what that third one is: AI Overviews and AI Mode are different Google surfaces, and they behave differently on commercial vendor questions. We swapped AI Overviews out of our own product for AI Mode last month precisely because of how the two differ on buying queries, so treat "Google" in any vendor's engine list as a question rather than an answer, ours included. And Claude sits in the Enterprise column alone. Whatever you make of that, it's a notable place to draw the line.

None of this makes Profound a bad product. It makes it a product priced for a buyer who has already decided AI visibility is a budget line with an owner, and who's scaling depth on the engines that matter most to them. If that's you, the ladder is rational. If you're a five-person firm trying to find out whether AI recommends you at all, you're being asked to pay $399 to see three of the places your buyers ask.

Who should buy it

Buy Profound if you have a marketing team with someone who owns AI visibility as a job rather than a curiosity, and your site already sits behind Cloudflare, Akamai, Fastly or a similar layer. Agent Analytics turns infrastructure you're already paying for into attribution nobody else in the category can produce, and that alone can justify the line item. Buy it too if prompt demand data would change what you publish, because Prompt Volumes is close to unique. And buy it if you need SOC2, SSO and a dedicated Slack channel, which is an enterprise procurement reality rather than a nice-to-have.

Don't buy Profound if what you need is breadth across engines at a small-company price, or if nobody on your team is going to operate a platform. The tools that reward you are the deep ones, and deep tools need an operator. A firm with no marketing hire will buy $399 of capability and use the dashboard twice.

Where we fit, and where we don't

We built TofuBofu for the other buyer, and the axis is coverage and execution rather than analytical depth. We query six engines, ChatGPT, Claude, Perplexity, Gemini, Google AI Mode and Bing Copilot, on every plan including the free one, because engine count isn't what we meter. And we don't stop at the measurement: the product drafts the content the gap calls for, publishes it to your CMS, and re-scans to show whether the answer moved. Measurement is close to commoditised in this category now. Doing the work is not.

Where we don't compete: if you want CDN-level AI traffic attribution or observed prompt demand, buy Profound, because we don't have those and saying otherwise would be a lie you'd discover in a week. Pick on which problem is yours. If the question is "am I invisible, and what do I do about it", start with a scan across all six engines. If the question is "AI sent us traffic last quarter and I need to prove it to a CFO", that's the other tool. For the fuller side-by-side we keep TofuBofu vs Profound and the alternatives list updated as prices move.

Frequently asked questions

Sources and further reading

• Profound pricing, brands view, read at source and screenshot-verified 7 September 2026: the $99 and $399 month-to-month prices, the tier labels quoted above, and the feature comparison grid carrying the Agent Analytics integrations. The per-tier engine mapping was pinned by element geometry on 28 August 2026 rather than by reading order
• Microsoft: New AI Visibility Insights in Bing Webmaster Tools, 16 June 2026: the free first-party citation reporting worth running alongside any paid tool, on Microsoft surfaces
• Google Search Console Help: Generative AI performance report, read at source 29 August 2026: Google's own free counterpart, with dimensions of Pages, Countries, Dates and Devices