Operating cadence and attribution

Read the daily series by topic. Pick the prompts where competitors appear and you do not, open the answers, note the cited sources. Decide one on-site action (a page to write or fix) and one off-site action (a source to earn). Ship, then watch the same prompts.

• Monday: read visibility and share of voice by topic; list new losses and gains.
• Pick one page and one source; brief the page from the winning answers.
• Publish, request indexing, and keep the prompts tracking daily.
• Two weeks later: compare the prompt's visibility with its pre-change baseline.

Because every prompt moves on its own, a before-and-after comparison on one prompt is weak evidence. Two habits help. Keep a set of prompts you deliberately do not act on as a control, and compare the change in the acted-on prompts with the change in the controls over the same period. And judge changes on two weeks of daily data, not on the day after publishing.

Buyers do not experience search and AI answers as separate channels, and your report should not either. For each topic, show the Google positions for its keywords and the AI visibility for its prompts, with the same competitors on both. Movement in one with none in the other is itself a finding.

Chasing single-day movements, writing for prompts nobody asks, publishing comparisons that hide a competitor's strengths, and blocking crawlers you forgot about in 2023. Each shows up in the data quickly once you track daily.