Profound vs VeritasLinks
The category leader, measured with our own instrument — and the two things a single number cannot tell you.
Profound is the largest company in this category: roughly a billion in valuation, a $96M Series C in February 2026, more than 700 enterprise customers including Target, Walmart and U.S. Bank, and coverage across more than ten AI engines at a prompt volume far beyond ours. Any honest comparison starts there. We ran Profound through our own instrument on 22 July 2026. The composite came back at 521 on a 300–870 scale, which is a good score. The interesting part was not the number. It was the 37 points between the five models that produced it, and a gap further down the report that counting mentions cannot reach.
Real dossier
Five models, five different opinions
VeritasLinks analysis of Profound · generated 22 July 2026
37-point spread · overall 521
| Model | VeritasScore |
|---|---|
| DeepSeek | 532 |
| Claude | 522 |
| Perplexity | 512 |
| Grok | 512 |
| ChatGPT | 495 |
| Overall | 521 — Good |
Context
Where 37 points sits
- 37 — Profound
- ~100 — Median across measured companies
- 232 — Widest recorded
The finding
Known by everyone, chosen by almost no one
Profound's visibility component came back at 130, near the top of the range we see. It was mentioned in every one of the ten test prompts, held 83.74% share of voice, and sentiment ran 98% positive with nothing negative at all. By every metric a visibility tool exists to sell, that is close to a perfect result.
Then the simulated buyer panel scored brand preference at 5.
The objections were specific rather than vague: no proof of ROI, black-box algorithms, and no control over the output. EEAT came in at 65, with thin presence on G2, Capterra and TrustRadius, and no analyst coverage standing behind the claims.
Counting mentions cannot surface that gap, because being talked about and being chosen are two different states, and the distance between them is where deals are lost.
brand visibility 130 · focus-group brand preference 5
Method
The difference is not how many prompts. It is what happens to each one.
Profound's headline volume — millions of prompts a day across more than ten engines — is the total across its entire customer base, not the depth applied to any one brand. Per brand, every monitor in this category works from a focused prompt set, and so do we.
What differs is what gets reported. A language model samples: ask it the same question twice and you get two different answers. A monitor asks, records what came back, and puts that figure on a dashboard. Ask again tomorrow and you get another figure, plotted as a trend — but a trend assembled from single draws cannot separate a real movement from the model's own variance. That is not a marginal effect: in our own testing, one model moved 232 points across fifteen identical runs with nothing changing between them.
VeritasLinks reports the distribution instead of the point. A single analysis puts a focused prompt set through five production models and repeats it, with follow-up probes where the pipeline supports them — between 360 and 700 scored model answers in the AI-visibility layer alone, before any score is calculated. The spread between models, the disagreement between them, and a flag when the run itself was low quality are published alongside the figure. The spread is the product.
Research
What happened when we tested the models themselves
Reliability study: 15 repeated runs of one identical prompt against one company. This is our published research into model stochasticity. It is not what a customer report contains.
15 repeated runs · stable shared range 44–79 · Perplexity (sonar) 232 · σ 76.3
This is the evidence behind the argument above, and the reason we publish a range rather than a figure on its own. It is a study we ran against our own domain to test the instruments — not a description of what you receive when you run an analysis.
At a glance
| Feature | Profound | VeritasLinks |
|---|---|---|
| Starting price | $99/mo Starter (ChatGPT only); $399/mo Growth; Enterprise $2,000–5,000+ quoted | $49/mo, first analysis free |
| Free first analysis | No — plan or demo required | Yes |
| Self-serve, no sales call | Starter and Growth self-serve; Enterprise sales-led | Yes |
| Engine coverage | 10+ engines, very high prompt volume across its customer base | 5 production models, a full scored dossier per model |
| Depth per brand | Focused prompt set, reported as a figure | Focused prompt set, repeated across 5 models — 360–700+ scored model answers per analysis |
| What is reported per result | The figure | The distribution: spread, cross-model disagreement, run-quality flag |
| Cross-model disagreement measured | No | Yes — Split-Perception |
| Perception Affect Field | No | Yes |
| AI-simulated focus groups | No | Yes |
| Investor-style scoring (VeritasScore 300–870) | No | Yes |
| Pitch-deck compression check | No | Yes |
| Remediation content generated | No | Yes |
| Best for | Enterprises that need maximum engine coverage and a team to operate it | Founders and lean teams who need to understand and fix how models describe them |
Where Profound wins
Profound is the most established product in this category and the coverage leader. If your requirement is breadth — the widest set of engines, very high prompt volume, enterprise reporting, and a team to operate it — Profound is a coherent choice and we are not a substitute for it. It is built for organisations with an AI strategy function and a procurement process. VeritasLinks is not a clone of that, and pretending otherwise would not survive your first demo.
Where VeritasLinks wins
VeritasLinks wins when you need to know whether a number means anything. Five scored models with the spread attached, the disagreement between them measured directly rather than averaged away, a simulated buyer panel that surfaces the objections behind a score, an investor-style read on how you would land in a diligence conversation, and the remediation content to change the next answer. Self-serve at $49 a month with the first analysis free, because we would rather you check the numbers than take our word for them.
Which should you pick?
Pick Profound if breadth of coverage is the requirement and you have the budget and the team to operate an enterprise platform. Pick VeritasLinks if you want a statistically honest read — the spread published alongside the score — plus the layers underneath it: why models describe you the way they do, what a buyer panel objects to, and what to publish next. If you are undecided, run the free analysis on your own domain and compare it against whatever your current tool is showing you.
To go deeper, run a full AI brand perception audit, read what GEO actually means, or compare plans.
FAQ
Common questions
Is VeritasLinks a cheaper Profound alternative?
- On price, yes — $49 a month with a free first analysis, against $99 for Profound's ChatGPT-only Starter tier and $399 for Growth. It is not a like-for-like replacement, though. Profound is built for coverage at enterprise scale. VeritasLinks is built to measure the distribution and to explain and fix perception.
What does VeritasLinks do that Profound doesn't?
- It publishes the spread between five models rather than a single figure, measures cross-model disagreement directly, runs simulated buyer panels that surface objections, scores you the way an investor researching you would, and generates the content to close the gaps it finds.
When is Profound the better choice?
- When breadth is the requirement — more than ten engines, very high prompt volume, enterprise reporting — and you have a team to operate it and a budget to match.
You scored Profound at 521. Isn't that just one snapshot?
- It is, and that is exactly the point. It is why the figure is published with the 37-point spread beside it rather than on its own. A single number, ours included, is one draw. The spread is what tells you how much to trust it.
How many questions does an analysis actually ask?
- A focused set of around ten unique prompts, put through five production models and repeated, with follow-up probes where the pipeline supports them. That produces between 360 and 700+ scored model answers in the AI-visibility layer alone. The prompt set is deliberately narrow; the depth comes from repetition across models, which is what makes a spread measurable.
Do I have to talk to sales?
- No. Run your first analysis free, then subscribe at $49 a month if it earns it.
See what five AI models really say about your brand — free.
First analysis free
First analysis free · Self-serve · $49/mo