Annual is 30% off, billed once a year.
Local
$19/mo
$13.25/mo
We ask ten questions a customer would type. You do not write them: they come from a shared library for your category and area, refreshed every month. This is the tier where sharing does the most work, because two businesses on the same street are being asked about in almost exactly the same words.
Measured provider cost, nothing cached: $0.80 a run
Starter
$49/mo
$34.25/mo
Twenty questions instead of ten, still written by us from the shared library, and refreshed every week rather than every month. Same single assistant, checked more often.
Measured provider cost, nothing cached: $1.60 a run
Growth
$149/mo
$104.25/mo
Twenty five questions written for your business, put to all four assistants every week, on both surfaces: the model behind the API and the product a customer actually opens. Every question is sampled ten times on the consumer surface, so a confidence band arrives with the first run rather than in week three, and you get the disagreement between the two surfaces that no other tool reports. A report costs the same as one month of this and tells you where you stand once. This tells you every week, and tells you when it moves.
Measured provider cost, nothing cached: $8.65 a run
Scale
$399/mo
$279.25/mo
Thirty questions written for your business, put to all four assistants, run whenever you ask rather than on a fixed day, and across more than one town or country.
Measured provider cost, nothing cached: $10.38 a run
The engine count is the decision that matters. Whether an engine costs $0.03 or $0.10 a call, the number of engines you check is what multiplies that cost, and it is also what determines whether you know your standing on one buyer surface or four. Prompt count moves the bill by a little; engines move it by a lot.
You are pitching next month, or the board asked a question, or you want to know where you stand before deciding whether any of this is worth doing monthly. Buy the report, read it, and owe us nothing.
The report
$149 once
Every question put to all four assistants on both measurement surfaces, ten times each on the consumer surface and three on the provider API. The repeats are the point: identical runs of the same question minutes apart agree on only 36% of the brands they name, so a single answer is one draw, not a measurement. Sampling is what makes the figures defensible on day one rather than in week three.
The deep report
$249 once
The same question set, sampled twice as hard: twenty times on the consumer surface and five on the provider API. The interval around every figure narrows, which is what you need when two competitors sit close enough together that the ordinary report will honestly tell you it cannot separate them.
it also buys you a month
Buy a report and you get a month free on any plan, for 30 days after the purchase. On Local, Starter, Growth the month is covered outright and nothing carries over. On Scale a month costs more than the report, so the $149 comes off your first invoice and you pay the difference. The credit never exceeds what you actually paid, because a rule that did would make buying a report the cheapest way onto the most expensive plan.
what it cannot tell you
Whether anything is changing. A movement needs at least three snapshots separated in time, and a report is one moment however deeply it is measured. If the question is "where do we stand", buy the report. If it is "is what we are doing working", that is a subscription, and no amount of depth bought on one day substitutes for it.
Three runs is not a marketing number. Our own study put the same prompt to the same surface twice and the two runs agreed on the sources cited only 23 percent of the time. A report built on one run of each question, ours or anyone else's, is mostly measuring that variance. This is why the replicate count is the thing separating the two reports above rather than the prompt count.
| feature | free | local | starter | growth | scale |
|---|---|---|---|---|---|
| Price | $0 | $19/mo or $159/yr | $49/mo or $411/yr | $149/mo or $1251/yr | $399/mo or $3351/yr |
| Prompts | 5 | 10 | 20 | 25 | 30 |
| Prompt source | Shared library | Shared library | Shared library | Your own | Your own |
| Engines | 1 | 1 | 1 | 4, all of them | 4, all of them |
| Refresh cadence | One time | Monthly | Weekly | Weekly | On demand |
| Markets | 1 | 1 | 1 | 3 | 5 |
| On-demand runs a month | None | 2 | 4 | 12 | 30 |
| MCP calls a month | None | 200 | 1,000 | 5,000 | 20,000 |
| SEO layer (rank tracking, keywords, backlinks, site audit) | Not included | Included | Included | Included | Included |
| Data export | Shareable report link | Full export, one click | Full export, one click | Full export, one click | Full export, one click |
| Measurement surfaces | API only | API only | API only | Both, and where they disagree | Both, and where they disagree |
| Confidence bands | Not enough runs yet | From the 3rd run | From the 3rd run | From the 3rd run | From the 3rd run |
Exact on-demand run and MCP call quotas are set per tier in the plan configuration and shown once you are signed in, since they are a data value in the plans table rather than a marketing number. What is fixed here is that they are hard limits on every paid tier, never soft targets.
No competitor in this category publishes its per-call cost. We do, because the entire pitch of this product is that a number without its receipt is not a finding. Measured against the live provider API on 30 July 2026, one prompt sent to one engine costs between $0.0256 and $0.0964, a 3.8x spread, depending entirely on which model answers and how long its answer runs.
| engine | model | average | measured range |
|---|---|---|---|
| Perplexity | Search-native | $0.0256 | $0.0116 to $0.0376 |
| Claude | Frontier | $0.0631 | $0.0555 to $0.0819 |
| Gemini | Fast | $0.0702 | $0.0104 to $0.1123 |
| ChatGPT | Frontier | $0.0964 | $0.0345 to $0.1189 |
| All four engines, one prompt | $0.2553 | ||
The provider's own margin on top of the model cost is flat at $0.0006 a call across all four engines, with no exception measured across 51 calls. Cost tracks the model and the token budget, not the request itself. Whoever picks the model picks the cost, which is exactly why the tiers are priced on engine count rather than on anything else.
Five prompts against Perplexity, the cheapest engine we measure, at $0.0256 a call: $0.128 per scan. At 100 free scans a month that is $12.80 in provider cost.
| conversion to a paid plan | new customers, of 100 scans | new monthly revenue |
|---|---|---|
| 0.5%, the pessimistic case | 0.5 | $24.50 |
| 2% | 2 | $98.00 |
| 5% | 5 | $245.00 |
Revenue is modelled against the $49 Starter tier. Even the most pessimistic conversion rate on this list, one new customer for every two hundred scans, pays for that month's entire free-scan cost by itself. That is the argument for not cutting the free scan down to something thinner than it is.
Not soft targets, not a fair-use policy enforced after the fact. A plan's quota is a hard ceiling: reaching it stops the action and says why, in the product, at the moment it happens. That is a deliberate design choice, not an oversight, and the reason is margin math we ran against the Growth tier at $149 a month.
| usage pattern, on the $149 Growth tier | monthly cost | margin |
|---|---|---|
| Nominal: 20 prompts, 2 engines, weekly, plus the SEO layer | $13.60 | 91% |
| Plus ten extra on-demand runs in the month | $38.00 | 74% |
| Running daily instead of weekly | $76.00 | 49% |
The scheduled prompts barely move the number. On-demand runs are what erode it, and an MCP-connected agent can call on-demand runs without anyone watching a screen: a loop is not a hypothetical failure mode, it is the realistic one. That is why the quota is enforced on every call, including read-only ones, and why it is part of the plan from the first release rather than a patch added after someone got a surprise invoice.
Why is this cheaper than an SEO platform charging $129 a month?
It isn't, exactly, and that comparison is the wrong one. Classical SEO data, rank tracking, keyword research, backlinks, costs two to three orders of magnitude less per call than a live answer from an AI engine. That is why the SEO layer is included free in every paid tier here: it is nearly free for us to include. What you are paying for is the AI answer measurement, which is genuinely the expensive part, and which a classical SEO tool does not do at all.
What happens if I cancel?
Cancellation runs through Stripe's own portal. Access does not end mid-cycle: it continues through the period you already paid for, and reverts to the free tier when that period ends. No further charge follows.
Do I own my data?
Yes. A one-click full export is available from the account settings screen at any time, no request needed: every brand, competitor, scan, cell, mention, metric, finding and diagnosis tied to your account, as one file. A product you cannot leave easily is not one you should trust with the measurement in the first place.
What if my usage spikes past the quota?
The action stops and the product says why, rather than letting the call through and surprising you with the bill later. That applies the same way whether the spike comes from a person clicking too fast or an agent connected over MCP calling in a loop.
Five real questions against a live engine with web search on. You get the answer text, who got named instead of you, and the sources it read. No account, and it costs us about $0.128 to run.