The 90-Day AI Visibility Playbook

If you read only one section of this playbook, read this one. Most of the AI-visibility losses we see in week eight are caused by what teams didn't do in week zero. There are four foundation pieces, all under an hour of total work.

1. Define your category in one sentence

AI engines retrieve by topic match. If your category sentence is generic ('we help SaaS companies grow'), no AI will know when to surface you. If it's specific ('we're the pricing analytics tool for usage-based SaaS billing teams'), the engines have a hook. Write your sentence with one industry noun + one specific problem + one customer type. That's it.

2. List the 30 prompts your buyers actually use

Stop guessing. Open ChatGPT, type 'what kinds of prompts would someone asking about [your category] type into me before they bought?'. You'll get 50-100 ideas. Pick the 30 most realistic. These are your tracking baseline.

3. Audit where you stand today

Run each prompt through ChatGPT manually OR use a visibility tool like Rankply. For each one, record: were you mentioned (yes/no), if yes what position (first/middle/last), and which competitors got mentioned. Save the spreadsheet. This is your day-zero baseline.

4. Pick your 'beachhead' niche

You can't win all 30 prompts in 90 days. Pick five. Make them the ones where (a) you have the best product fit, (b) your buyers are most likely to convert, (c) your current visibility is weakest. Now you have a focused 90-day target.

Week 1: technical AI-readiness audit

AI engines crawl. If they can't read your site, nothing else matters. Use Google's URL inspection tool + Rankply's free tools to check:

• Every page has a unique <title> (25-70 chars) and <meta description> (120-160 chars).
• FAQ schema (JSON-LD) on every page that has a Q&A section.
• Article schema on every blog post.
• Clear H1, H2 hierarchy — H2s phrased as questions where possible.
• robots.txt does NOT block /blog or /resources.
• Core Web Vitals pass on mobile (LCP < 2.5s, CLS < 0.1).

Most teams have 3-5 issues in this list. Fix them this week — it's a one-time cost that pays dividends for years.

Week 2: keyword research with the right filter

Forget the volume-first SEO playbook. For AI visibility you want keywords where the current AI answer is BAD. That means: only 2-3 brands mentioned, no clear winner, niche-specific.

For each of your 30 baseline prompts, ask ChatGPT and look at the answer. Score each one: how confident is the AI? How many brands does it name? Pick the 10 weakest answers as your target list. These are your winnable prompts.

Week 3-4: the three core pages

AI engines cite comparison pages and listicle-style guides far more than self-promotional content. Ship these three pages by the end of week 4:

• 1) A 'best X for Y' listicle covering your category — yes, include competitors honestly. AI engines see this as authority signal.
• 2) A 'X vs Y' comparison between you and your biggest competitor — be fair, be specific, link to real screenshots and pricing.
• 3) A 'what is X' definitional guide for your category — written so an AI can quote the definition verbatim.

The publishing cadence

Two posts per week, for the full 30 days. That's 8-10 pieces. NOT 30 — quality beats quantity by an order of magnitude in AI search. Each post should be 1,500-2,500 words, structured with H2-as-questions, with one clear definition near the top and an FAQ section near the bottom.

The content mix

Across the 8-10 pieces, hit these formats:

• 3-4 'how to X' guides — answer the literal questions your buyers ask.
• 2-3 'X vs Y' comparisons covering your other major competitors.
• 1-2 case studies — anonymised if needed but with real numbers.
• 1 deep listicle ('top 10 tools for X in 2026') where you're honest about the field.

Distribution: don't skip this

Publishing isn't enough. Each post needs at least 3 outbound signals to earn AI retrieval indexing:

• Repost or excerpt to LinkedIn the same day.
• Submit to relevant subreddits where you're a community member (not a drive-by).
• Send to your newsletter (even if small — newsletter mentions get indexed).
• Submit comparison posts to community sites like IndieHackers, ProductHunt blog, Hacker News (if newsworthy).
• Pitch one outlet per month for a guest contribution that backlinks to a foundation page.

Tracking weekly

Every Monday, re-run your 10 target prompts. Track the trend, not the daily noise. If you're showing up in 3 of 10 by week 8, you're on track. If you're at 0/10, your content isn't hitting the AI's retrieval set yet — likely a distribution or schema issue.

Backlinks: quality only

Forget link-buying schemes. AI engines weigh backlinks from trusted, topically-relevant sources heavily. Aim for 5-10 genuine backlinks in this 30-day window from:

• Industry publications you have a relationship with (or pitch cold — most editors accept guest contributions if the topic is sharp).
• Customer case studies on partner sites or vendor blogs.
• Podcast interview show notes (every interview = a backlink + a citation source).
• Reddit + Hacker News threads where your content genuinely answers a question.

Brand mentions in podcasts and newsletters

Pitch yourself to 10 niche podcasts in your category. Even small ones (500-2,000 listeners) get transcribed and indexed. Each podcast appearance = 1 citation source the AI can pull from. Three podcast appearances in 90 days is realistic and high-leverage.

Keep foundation pages fresh

AI engines weight recency. Your 'best X for Y' listicle should be updated quarterly with new entrants, new pricing, new features. Add a 'Last updated: [date]' badge prominently. Refresh signals are cheap and compound.

The 90-day checkpoint

At day 90, re-run your full 30-prompt audit. Compare to day-zero baseline. You should see:

• Visibility on at least 5 of your 10 target prompts (50% target).
• Sentiment scores trending positive (AI describes you favourably in the answer).
• Position improvement (mentioned earlier in the answer than competitors).
• Citation source diversity (your domain cited as a source, not just mentioned).

Problem-aware (high traffic, low conversion)

• How do I do X?
• Why is my X not working?
• What is the best way to X?
• What's the difference between X and Y?
• How much does X cost?
• How long does X take?
• What are the risks of X?
• When should I use X vs Y?
• What's a good alternative to X?
• Is X worth the money?

Solution-aware (medium traffic, high conversion)

• Best X tool for Y
• Top X software in 2026
• Recommend an X for [specific use case]
• Affordable X under $Y
• Most reliable X for enterprise
• Easiest X to set up for non-technical users
• X with the best customer support
• Open source alternative to X
• Best X with [specific integration]
• Top-rated X according to Reddit

Brand-aware (low traffic, highest conversion)

• Is [your brand] worth it?
• [your brand] vs [competitor]
• [your brand] pricing in 2026
• [your brand] alternatives
• [your brand] review
• Does [your brand] support [feature]?
• How does [your brand] compare to [competitor 2]?
• Who uses [your brand]?
• Is [your brand] better than free alternatives?
• Can I cancel [your brand] anytime?

Drop these into your Rankply tracking list (or whatever tool you use). Replace 'X' and 'Y' with your specific category nouns. Run them weekly during your 90-day sprint. Track the trend, not the noise.