Pricing
Priced on the questions you track
Every plan is the same product. What changes is how many questions we ask, how many assistants we ask them of, and how often. What an assistant gets wrong about your store is never one of the things a plan decides.
The free check — no account
Six discovery questions, one assistant, about twenty seconds. It tells you whether you are named at all, and it reports the interval honestly: at six answers, zero mentions is not proof that AI has never heard of you, and it will say so rather than print a scary 0%.
The first full report — no card
Connect a domain and the first end-to-end run is on us: rank, share of voice, who is named instead of you, the rivals you never listed, and every wrong fact. It is yours to keep and to share, on a permanent link, whether or not you ever pay us.
Starter
One store, one assistant, the questions that decide a sale.
Billed monthly as 180 AED. Cancel any time.
ChatGPT only
- 30 prompts
- Re-checked every other day
- 100 agent credits a month
- Rank, share of voice and wrong facts
- Other assistants need Growth
Then $49/month
Growth
Every assistant a shopper might ask, side by side.
Billed monthly as 915 AED. Cancel any time.
All four assistants
- 100 prompts
- Re-checked every other day
- 400 agent credits a month
- Per-engine breakdown
- AI traffic and revenue from your analytics
Then $249/month
Enterprise
Several brands, several markets, or your own schedule.
Invoiced to suit your finance team.
All four, plus any we add
- Unlimited prompts
- Daily, or your own schedule
- Credits to agreement
- Several brands or regions
- API access and a named contact
15 minutes to size it
14 days free on Starter and Growth. Card at sign-up, first charge on day 15, cancel before then and you are not charged at all.
What you are actually buying
Three numbers decide a plan, and each one is a real constraint on what we can run rather than a lever to make an upgrade look attractive.
A question, asked and recorded
One prompt asked of one assistant once is one answer, and we keep the answer — the full text, the citations, the date. Every rate you see is counted from those answers, so any number on your dashboard can be traced back to the paragraphs it came from.
Growth collects roughly 6,000 answers a month. That sample size is the reason a confidence interval narrows as you keep tracking, and it is most of what you are paying for.
A cadence, not a snapshot
One day tells you where you stand. A cadence tells you whether you are moving, and it is the only thing that can tell you the day you fell out of an answer you used to win.
Every other day is deliberate. Repeated measurements of the same question are correlated, so our intervals already discount them — daily collection would roughly double the cost of a plan while barely sharpening the answer. Where daily genuinely matters, it is priced as what it costs.
A credit, which is a page
Measuring tells you that an assistant recommends someone else. A credit is what turns that into a change you can publish: one page or one block of copy, drafted in your store's voice, in English or Arabic.
Each draft carries the passage from your own site it was grounded in, and says plainly where it had no source rather than filling the gap. You get a document you can hand to whoever writes your copy.
Line by line
Every row says what it means as well as what it is. A limit you discover after paying reads as a trick, so they are all here.
- What the card is charged
- Prices are shown in dollars and settled in dirhams, at a pegged rate. This is the figure on the statement — never a conversion done at the moment you look.
- Questions tracked
- The shopping questions we ask on your behalf — “best gym leggings in Riyadh”, not a keyword. You write them, or you edit the ones we infer from your site.
- Assistants asked
- Each engine answers differently. You can lead on Perplexity and be invisible on Claude, and a blended score would hide exactly that.
- How often
- A single day is a snapshot. A cadence is what lets you tell a real move from noise, and tells you the day you drop out of an answer you used to win.
- Answers a month
- Questions × assistants × runs. This is the sample every rate on your dashboard is computed from, and it is why the intervals narrow over time.
- Agent credits
- One credit drafts one page or one block of copy — the exact paragraph to change, in your store's voice, with the passage it was grounded in.
- Full report re-runs
- A fresh end-to-end audit you can send to someone. Separate from tracking, because a report is a document and tracking is a line on a chart.
- Documents you can export
- Every agent run is filed as a document in your workspace and exports to Word or markdown, with its citations and sample sizes intact. Not a panel you have to screenshot.
- Wrong facts about your store
- Every claim an assistant makes about your price, stock, shipping or returns, checked against your own site. Shown in full on every plan — including the free report.
- AI traffic and revenue
- Read from your own Google Analytics, read-only, in your property's own currency. Nothing is modelled and nothing is installed on your site.
- English and Arabic
- An Arabic question retrieves the Arabic web, so it is a different measurement rather than a translation. The two are reported separately, never averaged.
- Several brands or markets
- One account holds one workspace. Multiple brands, or the same brand measured separately per market, is an Enterprise arrangement.
- API access
- Pull your own numbers into your own reporting. Not self-serve — it is scoped on the call.
What a plan never decides
If an assistant is telling shoppers your shipping is free when it is 60 EGP, or quoting a price you raised in March, that is costing you money right now. You will be shown it in full — the claim, the engine that said it, the prompt that produced it and the sentence on your own site that contradicts it — on the free report, on Starter, on a frozen workspace, forever.
The fix is not gated either. Charging someone to find out what a machine is lying about them is a toll booth, and we would rather be the mechanic. What you pay for is the asking, again and again, and the work of moving the answer.
Every rate we publish carries the number of answers behind it and a 95% interval — on the site, in the product, in the emails and in the exports. If a number cannot carry one, we do not publish it.
Trials, cards and stopping
14 days free
Full access from the first minute — the same collection, the same dashboard, the same alerts. The card is taken at sign-up and first charged on day 15.
Cancel any time
From inside the app, in two clicks, with no call and no retention flow. Cancelling before the trial ends means you are never charged.
7 days of grace
A card that fails does not switch anything off. Tracking runs for 7 more days while we email you, because an expired card is an admin problem, not a decision.
Frozen, never deleted
When collection stops, everything already measured stays readable — permanently. Come back later and the series resumes, with the gap shown rather than smoothed over.
Find out where you stand first
The check is free and takes about twenty seconds. Decide what a plan is worth once you have seen your own number.