Key facts
- Start with a focused set of questions tied to revenue, not every possible keyword.
- Measure each engine separately; they rarely agree.
- Fix crawl access and wrong facts before writing new content.
- Report trends with a likely range so the team acts on real movement.
Week 1: questions and baseline
- 1List the questions buyers ask when choosing in your category, including comparisons with your main rivals.
- 2Group them by topic and buyer stage, and pick the ones closest to revenue.
- 3Ask each AI engine those questions as a buyer in your market and record who is named, in what order and from which sources.
Week 2: fix the foundations
- Allow AI search crawlers in robots.txt and your firewall.
- Correct wrong prices, features or policies wherever they appear.
- Add or fix structured data for your organization, products and FAQs.
- Consider an llms.txt file that points AI to your most important pages.
Weeks 3 and 4: content and sources
- Write answer first pages for the questions you lose, with specifics AI can quote.
- Publish honest comparisons with the brands AI names instead of you.
- Reach out to the reviews, communities and media that the engines cite for your topic.
Every month after
Re-measure the same questions, compare against the baseline and its likely range, and pick the next set of pages and sources to work on. Add markets and languages once the first one is moving.
Put this into practice with Variask Prompt research, Variask Brand accuracy and the free AI agent readiness checker. See Variask pricing or read the Variask FAQ.