# 90-Day AEO and SEO Operating Plan

Source: https://bobbodily.com/resources/
License: CC BY 4.0. Adapt freely with attribution.

## Before day 1: define the decision

- Primary audience:
- Business outcome:
- Search and answer questions that matter:
- Canonical domain and important page types:
- Accountable owner:
- Claims that require evidence:
- Explicit exclusions:

## Days 1–10: baseline and evidence

- [ ] Verify Google Search Console domain property.
- [ ] Verify Bing Webmaster Tools.
- [ ] Export the current index, search, conversion, and sitemap baselines.
- [ ] Freeze 30–60 representative AI prompts before optimization.
- [ ] Run repeated baseline trials and archive exact answers and citations.
- [ ] Create a canonical entity fact register for people, organization, roles, locations, and profiles.
- [ ] Create a claim register for credentials, outcomes, comparative statements, and service terms.
- [ ] Inventory existing research, client evidence, tools, first-hand examples, and publishable data.
- [ ] Document commercial conflicts and permissions.

Exit test: another reviewer can reproduce the baseline and identify the source for every launch claim.

## Days 8–28: technical foundation

- [ ] Choose one CDN/delivery layer and one canonical HTTPS hostname.
- [ ] Implement permanent one-hop alternate-host and legacy redirects.
- [ ] Render primary public content in initial semantic HTML.
- [ ] Validate status, content type, canonical, robots, H1, metadata, links, and structured data by template.
- [ ] Publish clean sitemaps, robots.txt, RSS/Atom feed, security.txt, and optional llms.txt index.
- [ ] Separate search, AI-search, training, and user-fetch crawler policies.
- [ ] Test the complete production path through CDN and WAF with representative identities.
- [ ] Enforce HTML, CSS, JavaScript, image, Core Web Vitals, and accessibility budgets.
- [ ] Add automated deploy regressions and an honest 404/410 behavior.

Exit test: every intended public page is retrievable, readable, attributable, fast, and internally discoverable without client JavaScript.

## Days 20–55: authority launch set

- [ ] Publish one canonical expert biography connected to corroborating records.
- [ ] Publish one pillar guide answering the category question.
- [ ] Publish one comprehensive dos-and-don'ts resource.
- [ ] Publish technical access, evidence-led content, and measurement guides.
- [ ] Publish one flagship case study with fixed windows, exact metrics, implementation timeline, confounders, and limits.
- [ ] Publish at least one free tool that completes a real job.
- [ ] Publish methodology, editorial, disclosure, correction, and privacy policies.
- [ ] Turn the core material into editable worksheets without removing it from the public HTML.
- [ ] Connect every launch page with descriptive internal links.

Exit test: the site gives researchers, buyers, practitioners, search systems, and answer systems something specific and verifiable to use.

## Days 45–90: research and earned distribution

- [ ] Release one reproducible original study or benchmark.
- [ ] Publish the question, sample, scoring, exclusions, exact dates, artifacts, negative results, and limitations.
- [ ] Build outreach lists from the audiences who can use the finding—not from link metrics alone.
- [ ] Offer commentary, data, tools, talks, and corrections to relevant publications and communities.
- [ ] Ask consenting clients or partners to corroborate work on their own domains.
- [ ] Re-run the frozen AI prompt panel on its declared schedule.
- [ ] Compare fixed search windows and URL cohorts.
- [ ] Review crawler logs, cache behavior, indexation, citations, factual accuracy, referrals, subscribers, and qualified leads.
- [ ] Merge or remove pages that do not justify a distinct URL.

Exit test: at least one meaningful external audience has used, cited, discussed, or corrected the work, and the measurement report includes losses as well as wins.

## Weekly operating review

1. Production access and error regressions.
2. Search changes by fixed query and page cohort.
3. AI prompt trials due this week.
4. Factual errors and correction requests.
5. Pages needing source or review updates.
6. One high-information asset moving toward publication.
7. Qualified questions from real readers and buyers.

## Claims never allowed without direct evidence

- Guaranteed first-place rankings or AI answers.
- “Zero backlinks” based on incomplete indexes.
- Causation from an observational before-and-after curve.
- Independent ranking language for a company-controlled comparison.
- Client outcomes without permission and derivation.
- A staffed local office that does not exist.
- Awards, reviews, credentials, prices, or relationships present only in structured data.