Choose one decision your audience already makes, then identify the page that should help them make it. Spend the first 30 days auditing that answer and establishing a baseline, the next 30 improving it, and the final 30 reviewing what changed and deciding whether to expand the work.
Keep your existing SEO program running throughout the pilot. The GEO vs SEO comparison explains which practices carry forward and how to make the business case. This plan turns that comparison into an assignment your team can scope, deliver, and evaluate.
Before you start: complete the readiness checklist
Copy this checklist into your project brief. Record the evidence and a named owner beside each open item. Resolve gaps that would prevent visitors from getting an accurate answer or prevent your team from running a comparable test.
☐ Audience decision. Name the audience and the decision they’re trying to make. Use search queries, sales calls, support tickets, or interviews to show why the question matters.
☐ Primary answer page. Record its URL, what it already answers, and the most consequential gap. Decide whether the fix belongs in a section, a supporting resource, or its own page.
☐ Technical access. Check that the intended URL and its key content are accessible, that indexing and canonical settings reflect your intent, and that platform controls match your organization’s policy. SEO foundations for AI visibility
☐ Entity and fact accuracy. Confirm the names, attributes, and relationships the answer depends on. Check consequential details such as product availability, service areas, eligibility, or integration limits against the people or records that own them. Entities and brand understanding
☐ Answer completeness. Write down what a reader needs to decide, including the direct answer, key conditions, tradeoffs, and a useful next step. Identify what’s missing from the current page. Audience intent and answer-optimized content
☐ Evidence. Identify the source for each material claim. Decide what documentation, firsthand example, or expert review the team needs before publishing.
☐ Reader path. Identify the pages that should lead readers to this answer and the supporting evidence or tools readers should reach from it. Topical authority and content architecture
☐ Measurement and capacity. Name the outcome you’ll evaluate, confirm how you’ll observe it, and agree on a staff-time budget. Record which work the pilot would displace and who will make the day-90 decision.
Copyable pilot brief
Audience and decision: ___
Evidence that this matters: ___
Primary page and the gap we’ll address: ___
Change we intend to make: ___
What we expect to observe if it helps: ___
Existing SEO or conversion measure we’ll monitor: ___
Staff-time budget and work deferred: ___
Accountable owner, sponsor, and review date: ___
For example, if buyers ask whether they can migrate historical data before quarter-end, start with the existing implementation guide. The work might be to add a verified migration section that explains what transfers, what affects timing, and what buyers should check before committing. The test is whether the guide and the answers buyers encounter convey those facts accurately.
Days 1–30: Choose the pilot and capture the baseline
The SEO lead owns the scope. Rank candidate questions by the importance of the customer decision, the gap you can verify today, and the effort to fix it. Choose a question your team can answer with evidence and a page it can improve within the budget. The sponsor signs off on scope and the criteria for expanding the work.
The analyst owns the baseline. Choose a manageable set of exact prompts, including variants where a real constraint changes the answer. Pick the AI search surfaces you have reason to test. Record the date, platform and feature, prompt, search mode, and relevant account or location settings. Decide how many repeat runs you’ll use, and save the original answers and source links before making pilot edits.
For each usable answer, record whether the brand was mentioned, the intended page was cited, the answer explicitly recommended the offer, and the key facts were accurate. Track clickable links separately where useful. Log unavailable features separately and report counts alongside the number of usable answers for each surface. Record any identifiable referral or conversion activity in your usual analytics. Prompt research and AI visibility auditing · Website measurement and conversion basics
By day 30, have a scoped brief, an answer-gap audit, a saved baseline, a budget, and a named owner for every task.
Worked baseline: kentlundin.com on November 9, 2026
We tested ten GEO questions across ChatGPT Search, Perplexity, Gemini, Google AI Overviews, and Google AI Mode. The initial run produced 0 citations of kentlundin.com across 50 results. The site appeared in personalized portions of the signed-in ChatGPT answers, so those appearances were recorded separately from citations. November 9 baseline audit and methodology
We then ran the same ten questions in a fresh, signed-out ChatGPT profile. That follow-up produced 0 mentions, links, citations, or recommendations across 10 answers. Web search was available in the signed-in run and unavailable in the signed-out run. Both the account context and search mode changed, so the two runs provide separate reference points. November 9 baseline audit and methodology
The audit gives us exact questions and documented conditions to revisit. For example, one question asked how GEO differs from SEO. Now that the comparison page is live, a follow-up can use the original question under comparable conditions and check whether the answer cites that URL and accurately describes the relationship between the two practices. Record the number of runs and any changes in test conditions alongside the result. Published GEO vs SEO comparison
Days 31–60: Improve the answer and record what changed
The content owner owns the page. Start with the missing information that would most change a reader’s decision. Give the answer directly, state the conditions that change it, and link to the evidence or the next useful task.
The subject expert verifies the facts. Confirm any consequential claims and attach their supporting evidence. The technical owner checks the published URL, content access, and the intended links. The SEO lead reviews the page in context and confirms that supporting pages lead readers to the right answer.
Record the publication date, the URL, which sections and links changed, and any other site or measurement changes that could affect the follow-up. Track staff time against the budget.
By day 60, have a verified live page, a dated change log, and the same prompt set and scoring rules ready for the follow-up.
Days 61–90: Repeat the tests and make a decision
The analyst owns the comparison. Rerun the baseline prompts with comparable platform, feature, account, and location settings. Use fresh sessions as specified in the original test and save the new answers. If a platform or search mode changes, record those observations under the new condition. Report raw counts and denominators by condition, including examples that show what changed in accuracy or source use.
Review observable referral or conversion activity alongside the prompt results, with the actual staff time and any changes in your existing SEO measures. A citation, an accurate answer, and a qualified lead are different outcomes. A before-and-after pilot can show what changed during the period. Other site changes, platform changes, and variation between runs limit what you can attribute to the pilot.
The SEO lead and sponsor own the decision. Use the standards agreed at day 30. Write down the evidence for the outcome, the cost, and the next action.
| Decision | When to choose it | Next action |
|---|---|---|
| Continue | Repeated observations meet the agreed evidence standard, or buyer feedback shows the page is more useful. The effort fits the budget and existing SEO measures remain within the agreed range. | Apply the same process to one related audience decision. Carry forward the test method and document what you expect to observe. |
| Revise | The question still matters and the evidence points to a specific gap, such as an inaccurate fact, weak supporting evidence, an access issue, or inconsistent test conditions. | Make one targeted correction. Record it and schedule a comparable retest before expanding scope. |
| Pause | There are too few comparable observations to judge the pilot, the time cost exceeds the agreed budget, the question has lost priority, or an essential dependency remains unresolved. | Record what would justify restarting. Keep any published improvement that continues to serve readers. |
By day 90, have a one-page decision record with the baseline, what shipped, the follow-up results and limitations, the actual cost, and the named owner for the next action.
Assign the work
Use names in this worksheet before the pilot begins. One person can fill several roles on a small team.
| Responsibility | Named owner | What they hand off |
|---|---|---|
| SEO lead | ___ | Priority question, page scope, and decision criteria |
| Analyst | ___ | Prompt set, saved baseline and follow-up answers, and result comparison |
| Content owner | ___ | Revised answer and dated change log |
| Subject expert | ___ | Verified claims, supporting evidence, and any qualifications |
| Technical owner | ___ | Checked page access, URL, and links |
| Sponsor | ___ | Budget approval and decision to continue, revise, or pause |
Start by completing the brief for a single priority question. Use the homepage roadmap to place the work within your broader SEO program, then bring the results back to your topical coverage map.
