You are underwriting these numbers. We built the product around that.
When you put your logo on a measurement, your client's trust in the figure is really trust in you. Every structural decision here — the bands, the stored answers, the removable branding, the non-removable methodology footer — follows from taking that seriously.
Your identity, everywhere it should be.
- Your accent colour across every highlighted figure and control, in the dashboard and the report.
- Your logo and name in the sidebar, the report header and the export.
- Your domain — the client visits you, not us.
- A live dashboard, not a monthly PDF. The client-facing report is a page they can open, and the export is that page printed — one component, so screen and PDF cannot drift apart.
And three things that stay fixed
- The engine palette. An engine’s hue is how a reader identifies it across every chart. If Perplexity is blue in one agency’s report and green in another’s, the palette has stopped being information.
- The semantic colours. Positive, negative, warning. Same reasoning.
- The methodology footer. Not a watermark and not an advertisement — it is what makes the figures above it defensible when the client’s internal team asks where they came from. A report without it is a slide.
Accent colours are validated as plain hex before they reach the page. That is a security control rather than fussiness: a CSS custom property is substituted wholesale wherever it is read.
Built for the twentieth client, not the first.
God-view
Every brand you run, on one screen: movement since the last period, coverage for the cycle, and whether anything failed. The screen you open on a Monday before anyone asks you anything.
Setup in minutes
Brand, owned-domain set, aliases, competitors, prompt set, cadence. The market carries engines, geography and languages, so there is no per-client configuration archaeology.
Alerts that stay quiet
Movement alerts fire when two bands stop overlapping, not on week-to-week wobble. An alert system that cries wolf gets muted, and a muted alert system is worse than none.
Roles
Owner, admin and member. Account managers read and report; configuration — the things that change a denominator — is held by people who can be asked why.
Proposal mode
A short-lived brand with a small prompt set, for a pitch. Real measurement on a prospect’s category, without burning a client slot on a deal you have not won yet.
Export where you already report
Branded PDF, CSV, and a tenant-scoped read API that carries the same bands and run counts the interface shows. Looker connector on the roadmap.
Where the margin actually is.
AI-visibility work has become a live line in agency retainers — a substantial share of agencies that raised prices in the last year cite GEO services as a reason. The measurement layer is what makes that line defensible, and what makes it renewable.
- It creates the reporting rhythm. A fortnightly band with a stability breakdown is a standing agenda item, which is the difference between a project and a retainer.
- It finds the work. Contested prompts and missing third-party sources are a concrete, addressable list — with URLs — rather than a strategy conversation.
- It survives scrutiny. When the client’s internal team asks where the number came from, the answer is a click, not a call.
- It prices in rand. USD billing against a volatile currency is a margin risk you carry silently until the month it stops being silent.
What one client brand costs to serve
| Configuration | Data cost / brand / month |
|---|---|
| 50 prompts, 6 engines, n=3, weekly | ≈ $130 |
| 20 prompts, 6 engines, n=3, weekly | ≈ $53 |
| 50 prompts, fortnightly | ≈ $66 |
| 50 prompts, daily | ≈ 7× the weekly figure |
Measured, not estimated — from 210 live runs after batch submission and a mid-tier analysis model cut cost per run by 79%. We publish this because “all engines included” is only credible if the economics behind it are, and because you are going to price on top of it.
The commitments that matter to a reseller.
We do not sell to your clients
No direct outreach to brands you have introduced, no “powered by” badge teaching your client our name, and no referral directory that routes work around you. The agency is the customer, not a channel we tolerate until direct sales work.
We do not sell the intervention
No content generation, no automated fixes, no “done-for-you” tier. The optimisation work is yours, permanently. A vendor that both scores your client and sells them the fix is a competitor with a head start.
We do not charge per engine
Six engines in every tier. Your client asking about Gemini should not become a change order, and coverage should not be a commercial negotiation halfway through a contract.
We do not hide the instrument
The methodology is public, versioned, and printed on every report. If we change how a number is computed, it appears in the changelog with a date — so you can always tell your client whether the world moved or we did.
What a pilot looks like.
| Week | What happens | What you get |
|---|---|---|
| 0 | A working demo on one of your real client categories — your prompts, live engines, raw answers opened in front of you | A view of whether the instrument holds up on your category |
| 1 | Tenancy provisioned, branding applied, first two brands configured together | Your dashboard, your colours, your domain |
| 2 | First cycles complete across all six engines | First bands, first citation tables, first stability classes |
| 4 | Second cycles complete | The first honest fortnight-over-fortnight comparison — and the first client-ready report |
| 6+ | Remaining brands onboarded, alerting tuned, reporting rhythm set | A standing deliverable you can charge for |
Week 4 is the first point at which we will let you show a client a trend, because it is the first point at which one exists. Anything earlier would be two readings taken too close together, and we will not print that with your logo on it.
Bring one client category and three competitors
The demo is a real cycle on your prompts, in your market, with the raw answers opened in front of you. It takes about forty minutes and it is the fastest way to find out whether we are wrong.