SEO Action Plan for Small Business (The 30-60-90 Day Method)

Days 1-30: Diagnose the site and fix the clearest friction

Most small business owners skip diagnosis and start writing blog posts. The first 30 days should answer a more useful question: where does the site already show a relationship between a page, a query and a business need? Search Console data is evidence about the selected period, not a forecast of when a page will rank.

Export the last useful Search Console window and inspect pages, queries, clicks, impressions, CTR and position together. Prioritize pages that are relevant to the offer and already receive impressions, pages with a clear intent mismatch, and pages whose traffic matters to a real conversion path. Keep the baseline so a later comparison can distinguish a change in visibility from a change in demand or seasonality.

• Repair broken internal links and important 404 journeys, checking that the replacement destination matches the user's intent. A crawl fix improves access and clarity; it is not a guaranteed ranking lift.
• Review titles, headings and descriptions where the visible promise does not match the query or the page. Do not rewrite every high-impression page by default, and do not treat a description as a ranking control.
• Use structured data only when the page is eligible, the markup matches visible content and the result is supported by current search documentation. Validate it, then measure appearance separately from clicks and rankings.
• Inspect the sitemap, robots rules, canonical signals and index coverage. Submit or refresh the sitemap when it is missing, stale or useful for discovery, not because submission guarantees inclusion.
• Define one business event and one search baseline: for example, qualified signup from an organic landing page, plus clicks, impressions, CTR and queries for the same URL set.

Days 31-60: Build depth only around a proven information gap

By day 30, you should have a short list of pages and unresolved questions, not a mandatory article quota. Pick the page whose improvement would matter to the reader and the business. Add a supporting article only when query data, customer conversations or a genuine comparison shows that the existing page does not answer an important sub-question.

A useful cluster is a navigation and explanation system, not a fixed number of URLs. Start with one clear intent, map adjacent intents, and link pages where the next step is genuinely helpful. A longer article is not automatically deeper, and an internal link is not automatically authority. The reader should be able to understand why each page exists.

• Give each page one dominant intent, then state what it covers and what it deliberately leaves to another page. Use Search Console queries, support questions and sales objections as inputs, not as an excuse to force keywords.
• Link from one page to another with descriptive, natural wording when the link helps the reader continue. Check orphan pages and misleading links as part of the review.
• Add an FAQ section when it resolves real questions. Add FAQ structured data only when the page and format are eligible and the markup reproduces visible content. It does not guarantee a rich result or an AI citation.
• Choose length from the intent, evidence, examples and objections the reader needs. Edit until the page is complete and usable; do not use a word-count target as a definition of quality.
• Choose a publishing rhythm the team can sustain and review. A slower sequence of useful pages is preferable to a burst of thin variations, but consistency alone does not make weak content valuable.

Days 61-90: Test distribution, trust and AI visibility

By day 60, you have a set of pages whose usefulness can be reviewed. The final 30 days are for testing distribution and trust, including how the content appears in AI answers. Treat GEO as an observation problem, not a second magic checklist: systems can change, citations are selective, and visibility is not the same as conversion.

Earned references can help discovery and trust when they come from relevant, independent sources. Look for a useful collaboration, a data point worth citing, a community answer or a product integration. Do not buy a quota of links or confuse a mention with proof that a page deserves to rank.

AI search systems may use different retrieval and display mechanisms from a conventional results page. Make claims attributable, separate facts from interpretation, show the method behind original data and keep important context close to the claim. FAQ markup can clarify eligible content, but it does not make a page citable.

• Audit a small sample for claim-level sources, clear headings, original examples, dated assumptions and an obvious answer to the target question.
• Run a repeatable AI observation set with the same language, prompts and date. Record whether the brand is mentioned, which sources are cited, whether the answer is accurate and whether the recommendation is relevant. Keep mention, citation, accuracy and conversion as separate measures.
• Pursue a small number of relevant references only where you can offer something useful. Record the relationship and the reason for the link; do not make a backlink count the success metric.
• Refresh, merge or retire pages when the evidence says they no longer serve a distinct intent. A substantive update should improve the reader's answer, not just change the date.
• Review the baseline monthly or at a cadence that fits the site's traffic. Compare like-for-like periods and annotate releases, seasonality, promotions and major search changes before attributing causality.

What this plan does NOT cover (intentionally)

This plan is opinionated about what to skip in the first 90 days. Three things you do not need yet:

• A large audit before you know the business question. Start with indexability, important journeys, templates and performance evidence that can affect access or usability. Escalate when the data shows a technical cause.
• A large paid-tool stack before you know what decision it will improve. Search Console, analytics and a lightweight research source can be enough to start; add a tool when its data answers a specific unresolved question.
• Tactics that promise instant rankings, copy competitors mechanically or produce pages without a reader need. Structured data spam, doorway pages and scaled low-value content create policy and quality risk; the remedy is not a different publishing quota.

How to actually execute this plan

A 90-day plan only works if it survives contact with a busy founder schedule. Choose a recurring review slot that is realistic for the team, define who owns each decision and keep a short change log. The useful habit is returning to evidence, not hitting an arbitrary number of hours.

At each review, record the hypothesis, the page or query affected, the change made, the observation window and the next decision. That makes it easier to stop a weak idea instead of defending it because time has already been spent.

They Will Know Me can turn connected Search Console and GA4 data into a prioritized action queue. Treat the output as a starting hypothesis: verify the page, the intent and the business context before publishing or changing a template.