Profound review
the most complete picture, at the highest operating cost
By Jānis Plūme, Founder, Outbound Pros · 10 min read · 2026-08-06
Quick answer
Profound is an AI search visibility platform that does two things most of its competitors do not: it reports on AI crawler and agent traffic hitting your own domain, and it works with real conversation data rather than only with prompts its own product invented. That combination makes it the most complete instrument sold in this category today, and it also makes it the one most likely to sit unused, because the amount of judgement required to read it correctly is closer to an analyst job than to a dashboard subscription.
What Profound is and who builds it
Profound is a New York company selling an AI search visibility platform to the enterprise end of the market. The product is a set of related surfaces rather than one report. Visibility tracking runs prompt sets against the major assistants and records whether you were mentioned, how you were described, which competitors appeared alongside you and which URLs the model cited. A conversation analysis surface works with real user query data rather than with a prompt list somebody on your team wrote, which is a genuinely different input from what most of the category uses. And an agent analytics surface reads traffic on your own domain and tells you which AI crawlers arrived, what they fetched and how often.
That last surface is the one worth paying attention to, and it is the reason Profound leads this list rather than sitting in the middle of it. Everything else sold in this category observes the output side: what the models said when asked. Agent analytics observes the input side: whether the machines that build those answers ever reached you. Those are different questions with different failure modes, and a team that only ever sees the output side will spend months rewriting content on a site that GPTBot has been served an empty container by the whole time.
Profound has also been unusually willing to publish method and research rather than only publishing conclusions. In a category where most vendor claims are unfalsifiable by construction, that is a real differentiator and it should be said plainly by a competitor.
Who Profound genuinely suits
Companies where AI search visibility belongs to one named person rather than to everyone in general, and where there is enough surface area that the answer changes by region, product line or buyer segment. If your brand is described differently in three markets and you need to know which one, this is the tool that can show you.
- Organisations with a large content estate where the question is not whether to invest but where the investment is already working
- Teams that need per market or per product line visibility rather than one brand level number
- Anyone who has to defend a budget internally and needs source level detail rather than a score that moved
- Sites where the engineering team will actually act on crawler traffic data once somebody puts it in front of them
- Brands whose description in assistant answers is materially wrong and who need evidence of it before anyone will fund a fix
The strongest use we have seen is not visibility tracking at all. It is entity correction. When an assistant describes what you sell incorrectly, the visibility number is beside the point and the accuracy field is the emergency. Profound surfaces that clearly enough that a marketing team can take it to a leadership meeting without translating it first, which matters more than it sounds like it should.
Where Profound is weak or the wrong choice
Prompt volume estimates are modelled, and they should be read the way you read keyword volume estimates in a classic SEO tool: directionally useful, precise looking, and not a measurement. No vendor has access to the query logs of the assistants. Anyone presenting an estimated monthly volume for a prompt is inferring it, and the honest way to use those figures is for ranking prompts against each other rather than for forecasting anything.
The agent analytics surface sees traffic that reaches your origin. If a crawler is being blocked at the edge, by a bot management rule you did not know was on, the traffic that never arrived is the finding, and reading an empty chart as an absence of interest is the wrong conclusion. Cloudflare has blocked AI crawlers by default on new domains since 1 July 2025 and nobody tells the marketing team. Check your edge configuration before you interpret a low number, and be aware that this failure mode looks identical to nobody caring about you.
Nothing in the product measures whether a given page returns your body copy in the initial HTTP response. That is the single most common technical failure we find on B2B sites and it is invisible to every product in this category, Profound included. It is also cheap to check yourself, which is why we built a free thing that does it.
The operating burden is the real objection. This is a product that rewards a person who opens it weekly, changes something because of what they saw, and comes back to check. Bought as reassurance rather than as an instrument, it becomes an expensive tab nobody has open. If your organisation does not have that person yet, hire or assign that person first and buy the tool second. That ordering advice costs us nothing and it is the most useful sentence in this review.
| Dimension | Rating | What that means in practice |
|---|---|---|
| Data provenance | Strong | Mixes genuinely observed data, agent traffic against your own origin and real conversation data, with prompt sampling. The mix is the point. Prompt volume estimates remain modelled and should be read as such. |
| Engine coverage | Strong | The major assistants plus the Google AI surfaces. Coverage breadth is not the differentiator here. What it does with each engine is. |
| Crawler access and rendering, gate one | Partial | Sees which agents reached your origin and what they fetched. Sees nothing about crawlers blocked upstream at the edge, and nothing about whether the returned HTML contained your copy. |
| Source and corroboration discovery, gate three | Strong | Cited URL detail is deep enough to build an off domain plan from, which is where most of the actual work lives once your own pages are readable. |
| Method transparency | Good | More published research and method than is normal in this category. Modelled figures could still be labelled as modelled inside the interface rather than in a support article. |
| Operating burden | Heavy | Assumes an owner. High ceiling, high floor. Bought without a named operator it produces charts and no decisions. |
Disclosure: we are not a neutral party
We run managed outbound as Outbound Pros, and this site is part of that group. When a company solves its pipeline problem with an inbound visibility platform, that is a company we did not sell to. We take no affiliate commission from any tool on this site, so the bias runs in exactly one direction and it runs against Profound rather than for it. Which is the argument for reading this particular review: a competitor calling a product the most complete instrument in its category is doing so against interest. Check the criticisms instead of the compliments. Every one of them is specific, and every one of them is something the vendor could answer without contradicting a fact stated above.
Profound questions we get asked
Is Profound worth it for a company with one marketer?
Usually not, and not because the product is bad. It is because the product assumes weekly attention from somebody who can act on what it shows. One marketer covering the whole funnel will open it in month one, admire it, and stop. Start with the free first party reports from Microsoft and Google, add a frozen panel of twenty to thirty prompts you run manually each month, and buy a platform when the manual version is genuinely the bottleneck rather than when it is merely tedious.
Does Profound measure ChatGPT visibility accurately?
It measures what the assistants answered when its prompts were run, which is real data about model output and is not the same thing as your visibility. OpenAI publishes no first party citation reporting, so nobody, including us, can offer you a measured ChatGPT visibility figure. What every product in this category does there is sampling. Profound samples more carefully than most and supplements it with conversation data, which narrows the gap without closing it.
Will Profound tell me if AI crawlers can read my site?
Partly. Its agent analytics show which crawlers reached your server and what they requested, which is more than any competitor offers. It will not tell you whether the HTML you returned contained your body copy or an empty root element, and it cannot see a crawler that a firewall or CDN rule rejected before it arrived. Those two checks take about ten minutes with curl and they decide whether anything else in the stack matters.
Profound or a cheaper prompt tracker?
Ask what decision you are trying to make. If the decision is which pages and which off domain sources to invest in next quarter, source level depth earns its cost. If the decision is whether anyone is mentioning you at all, a light tracker answers that for a fraction of the operating effort and you can upgrade later without losing anything except a history you had not learned to read yet.
Free, no signup. It tests the two things no visibility platform can see: whether a non rendering crawler gets a 200, and whether what came back contained your actual words.
Last updated: 2026-08-06
See what a crawler sees
on your own site
Paste a URL and get the extractability read: what an assistant can actually retrieve, and what it cannot.
Free. No signup, no email capture.