Services
Two products, priced in public.
Deliverables are defined as work performed on a defined cadence to a defined standard, not as outcomes achieved. We do not guarantee rankings or citations, and neither can anyone else.
Diagnostic audit
One-off, ten business days.
A ranked, costed fix list your developers can act on without us.
- Crawl and render audit, including a raw-versus-rendered diff per template. Content that only exists after JavaScript runs is invisible to most AI crawlers, and that is where the expensive problems sit
- Crawl-access matrix, covering
robots.txtplus observed CDN and WAF behaviour per named crawler, with crawler identity verified against published IP ranges - Indexation reconciliation, comparing what you serve, what Google crawled, and what it indexed, per template
- Internal link graph, with orphan pages, click depth, and a ranked "add links from these pages to these pages" action list
- Extractability review, testing whether the pages that matter answer a question in a self-contained way a machine can lift
- Structured data and entity check, covering what is present, what validates, and which of it earns anything today
- A baseline AI-answer reading, a frozen prompt set run against the major assistants, reported per engine as a sample with a stated interval, never as a single score
Findings are ranked by clicks and impressions at risk against implementation effort. An unranked list of four hundred issues is not a deliverable.
Monthly technical retainer
Ongoing, monthly.
- Weekly crawl with regression detection. Breaches are triaged and reported when found
- Fix specifications issued to your developers with the remediation pattern named, not just the symptom
- Server-log analysis of verified search and AI crawler behaviour
- IndexNow submission on publish, once integrated with your CMS in onboarding
- Bing Webmaster Tools citation reporting, where your account exposes those metrics
- The frozen prompt panel resampled monthly, reported per engine with confidence intervals
- A monthly written report, and a quarterly review call
Implementation tracked monthly as issued-versus-shipped, so that if results underperform you can see whether the work was done.
Implementation blocks
Optional. Only where you want us to ship the fix rather than specify it. Scoped and time-boxed access.
White-label / agency wholesale
Unbranded delivery, you own the client relationship. Technical scope only. What agencies hand us and what they get back →
Prices, terms and the refund condition are published in full on the pricing page. No call is required to get a number.
A fix specification, as issued
The retainer promises fix specifications with the remediation pattern named. This is one, written against a finding on this site on 8 September 2026, in the form a client's developers receive it.
| Field | Specification |
|---|---|
| Finding | The stylesheet is versioned by a query string. A request for a new version during the half-minute after a deploy is answered by the previous deployment and cached at the edge, under the new version, for a year. |
| Gate | Delivery. Below the five gates, and decisive for whether a visitor receives the current stylesheet. |
| Severity | Medium. A measurable cost with a known fix. |
| Evidence | GET /assets/style.css?v=c8977ee2 returned 29,843 bytes against a deployed file of 30,304, with cf-cache-status: HIT and age: 194. Instrument: curl from one location, 8 September 2026. |
| Cause | A query string changes the URL but not the file the edge looks up. A request during the propagation window was answered by the old deployment and cached immutable under the new URL. |
| Pattern | Content hash in the file name, not in the query string. An old version becomes a brief 404 instead of a stale hit for the whole cache lifetime. |
| Verification | After deploy, poll the HTML until it references the new file name on six consecutive samples, then fetch that file once and compare it byte for byte to source. |
| Effort | Small. One rename per release, scripted so it cannot be skipped. |
| Status | Fixed 8 September 2026. The deploy step now refuses to upload unless every page references the stamped file. The full self-audit → |
Availability
A vague answer here turns into a broken expectation later, so this is the schedule.
- Scheduled work (crawls, reports, panel waves) runs to cadence
- Correspondence and calls are handled next business day
- A detected regression that threatens indexation is reported as soon as it is triaged, and inside one business day
What we need from you
Several of the deliverables cannot be produced without access. Where it is missing, the report will say so instead of filling the gap with a guess.
| Access | Level | Why |
|---|---|---|
| Google Search Console | Full user | Reads all data, submits sitemaps, inspects URLs. Delegated owner only for a specific named task, then handed back. |
| Google Analytics | Read and analyse | Conversion and landing-page performance on target pages. |
| Server or CDN logs | Export or read | The only way to verify which crawlers reached you, by IP rather than a user-agent string anyone can spoof. |
| CMS | Editor, or scoped admin | Editor is enough to publish and wire IndexNow. Administrator only where a named task requires it, time-boxed then revoked. |
| Bing Webmaster Tools | User | First-party citation data where your account exposes those metrics. |
What we never ask for, and will decline if offered
- Your DNS or domain registrar credentials
- Hosting root or server administration
- A shared password of any kind. All access is delegated to a named identity we control, so it can be revoked without you changing your credentials
Every access grant is logged with its level, purpose and date. On termination, delegated access is revoked and your data returned or deleted at your election.
What is excluded, by design
| Excluded | Why |
|---|---|
| Ranking or citation guarantees | Not achievable, and offering one is a claim we could not substantiate. |
| Buying links, or exchanging goods or services for links | A named violation of Google's spam policies. The risk lands on your domain, not ours. |
| Bulk AI-generated content at volume | Google's scaled-content-abuse policy turns on volume, thin value and ranking intent, not on whether a human typed it. |
| A single blended AI-visibility score | One index over a noisy measurement, presented with a precision it does not have. We report per engine with intervals instead. |
llms.txt packages | No major engine documents consuming it, and Google states directly that it does not use AI text files. Cheap to publish. We will not bill you for it as a mechanism. |
| A live client dashboard | It invites reading noise as signal between reporting cycles. You get a report with the interpretation attached. |
| Content production as a retainer line | Out of scope by design. We specify targets, test whether a term is winnable, and write the extractability requirements. Your team or your agency writes the pages. |
If you need editorial volume, we will tell you that and stay out of it. Most agencies are strong on content and weak on the engineering layer (rendering, crawl access at the edge, log verification, indexation reconciliation). That gap is where the expensive failures live. We are the opposite shape, so we work alongside an incumbent rather than replacing them.
On timeframes
We will not tell you this is fast. Ahrefs' study of more than a million URLs found 1.74% of newly published pages reach the top ten within a year, and that 72.9% of pages currently in the top ten are three or more years old. Technical fixes can move indexation and impressions inside a quarter. Competitive positions on established queries take longer than any proposal admits.