Key Takeaways
- The tool I bought — Started with Otterly at $129/mo for the per-URL granularity and the friendly UI. Upgraded to Profound Pro at $499/mo at week six when the query set hit 150 and the daily refresh became material.
- The procurement step that mattered most — The paid two-week pilot. I ran one with Otterly and one with Peec AI on identical query sets. Free trials would have given me the vendor demo; paid pilots forced their teams to support my actual workload.
- The mistake — I optimised for monthly price first and query depth second. The right ordering is query inventory first, then everything else.
- The single test that picked the winner — Which tool surfaced span-level citation attribution that mapped back to a specific paragraph on my page. Otterly and Profound passed. Several others returned only domain-level attribution, which is not actionable for editorial decisions.
Why this procurement conversation looks different
LLM-visibility tracking is two years old as a market category. The 2026 vendor field has roughly ten serious entrants. The pricing ranges from a $49/mo free-tier upgrade to enterprise contracts in the high-six-figures. Almost everything written publicly about the category is vendor marketing. The amount of first-person buyer-side material is close to zero.
I ran a procurement exercise for my own site over March and April 2026. I shortlisted three vendors, ran two paid pilots, made a purchase, upgraded the purchase at week six, and learned one specific procurement lesson at the end. This piece is the documented version of all of that. It pairs with the selection-grade comparison at wetheflywheel.com/en/guides/best-llm-visibility-tools-2026/, which scores the same ten vendors on twelve axes.
What I needed the tool to do
Three things. First, track citations of prommer.net pages across ChatGPT, Perplexity, Gemini, and Claude on a fixed query set, weekly at minimum. Second, attribute citations at the span level (which paragraph), not just the domain level, so the data could drive editorial decisions. Third, expose an API I could use to feed citation metrics into the GA4 dashboard I use to track the rest of the site's referral programme.
Three things I did not need. A content-optimisation engine giving suggestions. Multi-domain coverage. A team plan with seats.
The category is fast-moving enough that the "must-have" and "do not need" lists are different from what they would have been six months earlier and different again from where they will be in six months. The shape of the requirements is more durable than the specific items inside them.
The shortlist of three
The 2026 field has ten serious vendors: Profound, Peec AI, AthenaHQ, Otterly, Scrunch, Evertune, Rankscale, Bluefish, Semji, and Goodie AI. I shortlisted three for a paid pilot.
- Otterly, the low-cost entry. Per-URL citation granularity, friendly UI, $129/mo. The pick if my query set stayed below 50 and I did not need daily refresh.
- Peec AI, the mid-market default. Solid engine coverage, content-optimisation suggestions, $249/mo. The pick if the content-optimisation feature turned out to be a workflow upgrade rather than a feature dump.
- Profound, the enterprise upgrade path. Span-level attribution, daily refresh on Pro, native API, $499/mo. The pick if the programme grew faster than I expected.
I left out the other seven for specific reasons. Bluefish was beta-only at the time. Goodie AI was newer with narrower coverage than I needed. Rankscale was the cheapest option but had partial Gemini coverage and no Claude. AthenaHQ was strong but I wanted to compare per-domain pricing (Otterly, Peec) with per-query pricing (Profound) and the procurement signal would have been muddier with both per-query options in the same evaluation. Scrunch and Evertune are enterprise-only with custom pricing and were out of band for a single-domain solo evaluation. Semji is content-optimisation-first and I did not need that.
The two paid pilots that decided the purchase
I ran two pilots in parallel: Otterly and Peec AI, two weeks each, on an identical 30-query test set covering my most-cited-worthy pages. Paid, not free, because free trials run on vendor demo schedules and paid pilots run on the buyer's actual workload.
Two weeks was the right window. The vendors had time to set up the workspace properly, the dashboards populated with three refresh cycles of data, and I had time to push a question to each vendor's support team and see how they responded. The Otterly support response was thorough and on-product. The Peec AI response was equally thorough but redirected toward upselling the content-optimisation tier, which gave me data I would not have otherwise had.
Free pilots run on the vendor's schedule. Paid pilots run on yours.
Why I picked Otterly first
For weeks one through six, Otterly won on cost-to-capability. The per-URL attribution worked. The dashboard surfaced the signal I needed without surfacing noise. The engine coverage on ChatGPT and Perplexity was strong; coverage on Gemini was good enough for the query set; Claude coverage was partial but my audience does not skew Claude.
The honest answer for why I picked Otterly is that the $129/mo price made the procurement conversation with myself trivially short, and the per-URL granularity meant I was not giving up the actionable data layer for the cost saving. The category-default move would have been Peec AI at $249/mo, and that would have been a defensible call. For a single-domain solo programme starting with a small query set, Otterly's value-to-cost ratio was clearly better.
Why I upgraded to Profound at week six
Two things broke the original procurement assumption.
First, the query set grew. Thirty prompts became fifty became 150 inside six weeks as I added query classes for each of the major topic clusters on the site. The Otterly plan I was on supported the larger query set in principle but the dashboards were not designed for that depth; I started losing the trend lines under the prompt density.
Second, I shipped an intervention (the canonical answer block rollout across 27 pages) where I needed daily citation data to know whether the deployment had landed properly. Otterly's weekly refresh on Starter was not enough resolution for that week. The upgrade to Profound Pro gave me daily refresh, span-level attribution, and the API I had originally said I needed and then negotiated away to save the $370 monthly cost difference.
The lesson is the one I would underline if I were giving a procurement deck on this category. Buy the right tier for the next six months, not the right tier for the next week.
The five-stage procurement playbook
The structure that should have been in my head from the start.
| Stage | What it actually means |
|---|---|
| 1. Inventory | List every query that maps to a high-value page. I ended with 150. I thought I had 30 when I started. |
| 2. Shortlist | Three vendors per lane. I picked Otterly (low entry), Peec AI (mid-market default), and Profound (enterprise upgrade path) from the 10-vendor field. |
| 3. Paid pilots | Two weeks each on identical query sets. Paid, not free, because free trials run on vendor priorities. |
| 4. Score | Coverage on the four engines that matter for my audience. Span-level attribution rate. Refresh cadence. API quality. |
| 5. Buy + plan to upgrade | Buy the right tier for the next six months. Plan the upgrade path explicitly. Otterly → Profound was always on the roadmap; I tried to skip the upgrade and paid for that. |
What I would do differently
- Inventory the queries before the first vendor call. The biggest mistake was treating the query inventory as something I would discover during the evaluation. The right move is to write it down first, even if the list ends up rougher than the ones the vendors produce.
- Start at the tier that will be right at month six. Pricing tiers in this category are not designed to be skip-able. The cost of getting the tier wrong twice exceeds the cost of starting at the right one.
- Negotiate the API access into the entry tier. Otterly's API is a Pro feature. Asking for it on Starter was a free option that I did not exercise. Most vendors in this category have a quietly-listed price point for adding the API to a lower tier; the worst they can say is no.
Related on the network
Why did you write this publicly rather than keep it internal?
Two reasons. First, the LLM-visibility category is two years old and most of the public content is vendor marketing. There is almost no first-person buyer-side material. Second, my own credibility on AI search is partly a function of being measurable inside the engines I write about, and a public case study is part of that measurement.
Why two tools instead of one?
I did not buy two simultaneously. I bought Otterly first and ran on it for six weeks. At week six the query set had grown from 30 to 150 prompts and I needed daily refresh on a launch-week intervention, so I upgraded to Profound Pro. Otterly is still in the toolchain as a sanity check on the Profound numbers; I would not recommend that two-tool setup as a default, but it works as a transition.
How much did you pay total?
About $1,800 over the first six months. Otterly at $129/mo for the first six weeks, then $499/mo for Profound Pro from week seven onward. Two paid pilots at roughly $100 each. The procurement time was meaningful too: about twelve hours of evaluation work across the two pilots and the scoring conversation.
What would you change about the procurement process?
I would do the query inventory before the first vendor call. I started with 30 queries because that is what I thought my visible AI-search surface looked like. The actual surface for my pages was closer to 150 queries. If I had known that going in, I would have started with Profound at $499/mo and skipped the Otterly stage entirely. The cost difference over six months would have been about $1,200, but I would have had cleaner data for those weeks and a less embarrassing procurement conversation with finance.
How does Profound compare to Peec AI?
For my use case, Profound won on three axes: span-level attribution, daily refresh, and the API. Peec AI won on price and on the content-optimisation suggestions, which are a genuine product differentiator if your workflow runs through them. If you are running a one-domain in-house programme with a 50-prompt query set and you want the suggestions inside the dashboard, Peec AI is the better pick. If you are running a 150+ prompt programme and need the API to feed your own pipelines, Profound is the better pick. The category-level write-up at wetheflywheel.com/en/guides/best-llm-visibility-tools-2026/ has the full 10-vendor scoring rubric.
Did you consider building this in-house?
For the first two weeks of the experiment, yes. I sketched a lightweight version against the OpenAI, Perplexity, and Gemini APIs. A weekend of work, maybe two. I would have spent three to five hours per week maintaining it. After running the math I bought Otterly instead. The vendor delta over a homegrown tracker is the engine-by-engine drift handling, the response parsing for citation extraction, and the fact that I do not have to keep up with API changes across four vendors. Worth $129 to $499 a month. Not worth four hours of my own time per week.
What is the one thing every buyer of this category gets wrong?
Under-buying on query depth. The default purchase pattern is to optimise for the lowest viable monthly cost and then discover the query set has outgrown the plan by week six. Map the query inventory before the procurement call. The cost difference between the right tier and the wrong tier is dwarfed by the cost of the wrong-tier conversation with finance two months later.
Ready to Transform Your AI Strategy?
Get personalized guidance from someone who's led AI initiatives at Adidas, Sweetgreen, and 50+ Fortune 500 projects.
