Three stages. Thirty days. Measurable AI citations.
Every Answer Intent engagement follows the same three-stage process. Stage 1 audits where you are - 30 AI prompts, three engines, a Day 0 baseline. Stage 2 builds the deliverables - schema, LLMS.txt, content - and hands them to your developer with an implementation guide. Stage 3 measures - a 30-day retest compares your new AI citation rate against the baseline. The same process scales from a R4,900 audit to a Tier 4 multi-page build.
Audit
5 business days
Day 0 to Day 5
- 30-prompt AI query battery
- Share of Model Response score
- Schema and LLMS.txt audit
- Day 0 baseline document
Build
10 business days
Day 5 to Day 15
- JSON-LD schema (per page template)
- LLMS.txt file (2026 standard)
- AEO content drafts
- Implementation guide
Measure
30 days after deployment
Day 45 to Day 60
- 30-prompt retest (same protocol)
- Day 0 vs Day 30 delta table
- Day 60 forecast
- Guarantee trigger if no citations
Audit: 5 business days
The audit confirms exactly where your business is invisible in AI search and why. We run 30 carefully constructed prompts across ChatGPT, Perplexity, and Google AI Mode (90 total queries). We capture which competitors are cited instead of you, what schema gaps exist on your site, and what content is failing to extract. The output is a Day 0 baseline document and a tier recommendation.
What we do
Prompts are constructed in five blocks: direct brand queries (does AI cite you by name), category-discovery queries (who does X in your location), problem-led queries (I need help with Y), comparison queries (X vs Y service providers), and partner or donor pathway queries (businesses that serve Z). Each block runs across all three engines in a private window with no logged-in accounts, so results reflect the AI's base citation behaviour.
The technical audit runs in parallel: schema presence and validity across every routable URL, LLMS.txt presence and structure, robots.txt AI bot permissions, sitemap coverage, and content extractability scoring for your top 5 pages. Schema errors and missing types are catalogued with recommended fixes.
What you receive
The Day 0 baseline document is a structured report with verbatim AI responses for each of the 30 prompts, your Share of Model Response score by category and by engine, and a citation-rate comparison against your top 5 competitors in each prompt block.
It also includes a schema audit table (type, page, status, priority), a content extractability assessment (which pages AI engines can parse and which they cannot, with the specific failure mode for each), and a prioritised tier recommendation with the scope and pricing for the build stage. This document becomes your evidence file for all future measurement.
What's required from you
A completed intake form (approximately 10 minutes) covering your business category, geographic service area, top 5 services or products, and 2 to 3 reference competitors if you have them in mind. URL access to your live site - no logins, no CMS access.
A 30-minute kickoff call is available but not required. If you can fill out the intake form thoroughly, the audit runs without a call. Most clients find the intake form alone is sufficient.
Timeline
Build: 10 business days
The build produces the three deliverables: schema, LLMS.txt, and AEO content. Schema is generated per page template, validated against the schema.org specification and Google's Rich Results Test, and packaged with deployment instructions. LLMS.txt is built to the 2026 standard. Content is rewritten or written fresh to be extractable - definition layers, answer-first paragraphs, FAQ clusters.
What we build
Schema types covered as a minimum on Foundation Sprint and above: Organization (legal name, address, contact, knowsAbout), Person (for named founders with jobTitle, knowsAbout, sameAs to LinkedIn), Service (one block per core offering), FAQPage (per page carrying FAQ content), BreadcrumbList (for all routable pages), and WebSite. HowTo schema is added where process content exists. Article schema for editorial pages.
Tier 4 adds LocalBusiness where applicable, Product schema for e-commerce or configurable service products, and any additional types the site structure requires. Every schema block uses @graph with @id cross-references so AI engines see a connected entity map, not isolated fragments. LLMS.txt includes title block, description, key pages with per-URL summaries, FAQ links, and contact section.
What you receive
A delivery folder containing: one schema file per page template (HTML code blocks, paste-ready), the LLMS.txt file (upload-ready), content drafts in markdown format (one per deliverable), and the implementation guide.
The implementation guide is specific to your platform. It covers Webflow, Framer, WordPress (via the Yoast or Rank Math schema field or a custom code plugin), Shopify (via the theme.liquid head section), Squarespace (via the code injection field), and custom-coded sites. Each platform section specifies exactly where to paste and how to validate. Estimated developer time is 2 to 4 hours total across all three deliverable types.
What's required from your developer
Schema deployment: paste each schema block inside the head element of the relevant page template. For sites with templated pages (services pages, blog posts), one schema block covers all instances of that template. The guide specifies which pages need unique schema and which share a template.
LLMS.txt deployment: upload the plain text file to the root directory of the hosting environment so it is accessible at yourdomain.com/llms.txt. This is the same process as uploading robots.txt.
Content deployment: paste markdown content into the CMS editor. The guide covers heading-level mapping for each platform. After deployment, run the validation links in the guide to confirm schema is live and parseable.
Timeline
Measure: 30 days after deployment
Measurement is what makes Answer Intent accountable. Thirty days after your developer deploys, we re-run the same 30 prompts across the same three engines. The Day 30 retest compares directly to your Day 0 baseline. New citations are counted. Improvements in Share of Model Response are quantified. The 60-day boundary triggers the AEO Proof Guarantee if no citations appear.
How we measure
The retest protocol is identical to the Day 0 audit: same 30 prompts, same three engines, same private-window protocol (no logged-in accounts, fresh session per prompt), same capture template. This consistency is what makes the comparison valid.
The comparison table is built prompt-by-prompt: Day 0 result (who was cited), Day 30 result (who is cited now), and a delta column (gained, lost, unchanged). Share of Model Response is recalculated by category and by engine. If you gained share, the specific prompts where you appear are noted. If you have not moved on certain prompts, the hypothesised reason is documented and a recommendation is added.
What you receive
The Day 30 progress report is a structured document with three sections. First, the delta table showing every prompt's Day 0 result vs Day 30 result side by side. Second, a citation analysis covering which queries you now appear in, which competitors lost share to you, and the hypothesised attribution (which deliverable drove which citation).
Third, a Day 60 forecast: if trajectory continues, which additional prompts are likely to show citations by Day 60, and which remain unlikely without further work. The forecast includes specific recommendations for any deployment fixes or additional content work that would improve the Day 60 result.
If citations have not moved
The first diagnostic is verifying the deployment is live and parseable. Schema must be in the correct location and return a 200 status from the schema.org validator. LLMS.txt must be accessible at the root. Content must be live on the indexed pages.
The second diagnostic is robots.txt - it must allow the AI bots (GPTBot, ClaudeBot, PerplexityBot, and Googlebot) rather than blocking them. The third is validating that schema parses without errors in Google's Rich Results Test. If all three pass and citations have not moved by Day 30, we produce additional schema or content work and hold for the Day 60 measurement. If Day 60 passes with no citations, the AEO Proof Guarantee activates and we continue building at no charge.
What happens at Day 60
Day 60 is the final guarantee measurement. We re-run the same protocol and document the result. If you are cited in at least one AI-generated answer across ChatGPT, Perplexity, or Google AI Mode, the guarantee condition is satisfied and the engagement closes successfully.
If not, we keep working - no additional charge, no expiry date on the commitment. We produce more deliverables, test new prompt categories, and re-measure at agreed intervals until the citation appears. This has not yet been required on any completed engagement, but the commitment exists in writing before the engagement begins.
Who does what
Answer Intent produces all deliverables. Your developer deploys them. You review and approve. Nobody needs to touch each other's systems.
| Task | You | Your developer | Answer Intent |
|---|---|---|---|
| Audit access | Provide live site URL | Not involved | Runs 30-prompt query battery and technical audit |
| Intake form | Complete (10 minutes) | Not involved | Reviews and clarifies any ambiguity |
| Schema build | Review and approve | Not involved at build stage | Builds, validates, and packages per page template |
| LLMS.txt build | Review and approve | Not involved at build stage | Builds to 2026 standard |
| Content drafts | Review and approve | Not involved at build stage | Writes and revises (one round included) |
| Schema deployment | Not involved | Pastes block into page head (2-4 hours total) | Delivers code block and implementation guide; available to support |
| LLMS.txt upload | Not involved | Uploads file to root domain | Delivers file with platform-specific upload instructions |
| Content publishing | Not involved | Pastes markdown into CMS | Delivers markdown with CMS-specific paste guide |
| 30-day retest | Not involved | Not involved | Runs full retest and delivers Day 30 progress report |
| Guarantee trigger | Not involved | Not involved | Continues building at no charge if Day 60 passes with no citations |
What clients ask about the process
Seven questions from the intake form stage and the Day 30 check-in.
Most CMS platforms have built-in custom code fields where schema and LLMS.txt content can be pasted directly without developer involvement. The implementation guide covers each platform's specific paste-points for Webflow, Framer, WordPress, Shopify, and Squarespace.
The Deployment Check Call add-on (R1,800) covers the entire deployment for businesses with no technical resource - Answer Intent reviews the deployment async before measurement starts.
Two safeguards apply. First, every schema block is validated against the schema.org specification and Google's Rich Results Test before delivery - errors are caught before the file leaves Answer Intent. Second, the implementation guide includes validation links your developer runs after each paste. If the test fails, the developer can identify the issue before it affects live measurement.
The Deployment Check Call add-on (R1,800) adds an additional async review by Answer Intent after your developer has deployed but before the Day 30 clock starts.
Two signals. First, AI referral traffic in Google Analytics (look for referrers including openai.com, perplexity.ai, and bing.com) usually appears within 7 to 14 days of deployment. Second, the Day 30 retest shows direct citation movement regardless of analytics attribution.
If neither signal appears by Day 30, we investigate - starting with deployment verification, then robots.txt AI bot permissions, then schema validation. The issue is almost always in the deployment rather than the deliverables.
A structured document with three sections: a delta table showing every prompt's Day 0 result vs Day 30 result (verbatim AI responses side by side), a citation analysis covering which queries you now appear in and which competitors lost share to you, and a Day 60 forecast with specific recommendations.
Every Day 30 report references the verbatim Day 0 baseline so you can audit the comparison independently. No AI-generated summaries are used in the report - every result is a manually captured, verbatim response.
Yes, and many clients do. We provide the prompt list, the engines to test, and the protocol (private window, fresh session, no logged-in accounts) as part of the Day 0 baseline delivery. Early movement before Day 30 is common - schema and LLMS.txt are often picked up within 7 to 14 days.
The formal measurement for guarantee purposes is the Day 30 retest run by Answer Intent using the canonical protocol. Client-run checks between Day 0 and Day 30 are useful for tracking but are not the measurement of record.
AI engines do change. ChatGPT, Perplexity, and Google AI Mode each update their citation behaviour periodically - sometimes significantly. We track engine updates that affect citation behaviour and adjust monitoring accordingly.
Any major shift is flagged in the Day 30 report with context: what changed, how it affects your category, and what (if any) additional deliverable would address it. Tier 3 clients receive a monthly engine-update summary as a standing section of their Share of Model Response report.
The Tier 2 engagement closes successfully. Options from Day 60: move to the Ongoing Programme (Tier 3) for compounding AI authority - Foundation Sprint counts as Month 1 so there is no additional Month 1 charge. Take the 90-day monitoring report add-on (R4,800 per month) as a lighter ongoing measurement. Or remain on the Day 60 result with no further commitment.
Most Tier 2 clients move to Tier 3 within the first 90 days after seeing the Day 30 report confirm measurable citation movement. The momentum from Month 1 makes the case for compounding it.
Ready to see your Day 0 baseline?
Start with the Visibility Audit. 30 prompts, three engines, 5 business days. You see exactly where your business is invisible in AI search - and why. R4,900.