How the program actually runs

How the program runs

Every engagement runs as a four-phase loop: audit, eligibility, answer engineering, corroboration. Each phase ends with a measurable exit condition, and no phase is skipped because the previous one looked healthy.

Work is tracked in a shared queue with weekly engineering handoffs. You always know what is shipping this week, what is blocked, and which prompt cluster it is meant to move.

Onboarding in two weeks

Onboarding is deliberately front-loaded so the first fix ships inside the first fortnight.

• Day 1-2: access, analytics and Search Console connection, prompt discovery workshops.
• Day 3-5: technical crawl, render audit and AI crawler policy review.
• Week 2: baseline prompt panel, entity model draft, first engineering queue.
• Week 3 onward: weekly shipping cadence and monthly share-of-answer reporting.

Reporting and metrics

The monthly report leads with share of answer per prompt cluster, then AI-referred sessions and conversions, then the work that moved them. Vanity metrics are excluded by policy, and every chart carries the underlying prompt list so results are auditable.

• Share of answer: mentioned, cited, recommended or absent.
• AI referrals and assisted conversions from analytics.
• Eligibility health: crawl, index, schema validation and rich results.
• Corroboration: retrievable third-party sources mentioning the entity.

Security and access

We work read-only on your analytics and Search Console wherever possible, use scoped credentials, and never store production secrets in reports or repositories. Access is revoked at offboarding and documented in the engagement record.

AEO glossary

Short definitions for the terms used across this site and in reporting.

• AEO: answer engine optimization — earning citations inside generated answers.
• Share of answer: the proportion of tracked prompts where your brand is included.
• Prompt panel: the fixed set of buyer questions measured every month.
• Entity: the machine-readable identity of your brand, products and people.
• Retrieval: the step where a model selects sources before generating an answer.
• Corroboration: independent sources repeating the same fact.