A workable 90 day AI visibility plan fixes one layer at a time, in six phases over 13 weeks. Five phases take two weeks, and the product data phase takes three. The phases cover crawler access, brand identity, product data, written answers, trust and a final measurement. The order matters more than the speed, since each phase needs the one before it to work.
Trying to fix every layer at once usually leaves every layer half done. Two or three weeks per phase leaves room for the normal work of running a store. It also gives each change time to show up in a scan.
Why does the order matter?
An assistant reads a store from the door inward. If its crawler cannot fetch a page, the product markup on that page earns nothing. If the brand name differs between the homepage and the markup, the assistant has two businesses to choose from.
In our June 2026 study of 216 Shopify storefronts, 41% failed the Front Door layer and 72% failed Entity Integrity. Those two layers take up weeks 1 to 7, since brand identity and product data both belong to Entity Integrity.
What happens before day 1?
Record a baseline so you can see what moved at the end. Run the free scan and save the scorecard. Then run the twenty minute audit by hand and write down each pass and fail with the page you tested. What your AI visibility score means explains how to read the five layers on the scorecard.
What happens in each phase?
Weeks 1 and 2: open the front door
- Open yourdomain.com/robots.txt and confirm that GPTBot, OAI-SearchBot, ClaudeBot and PerplexityBot are not blocked. On Shopify, the file is edited in Online Store > Themes > Edit code, in the robots.txt.liquid template.
- Check that product text appears in View page source on three product pages. Product text should be in the page source itself, rather than loaded later by a script or an app widget.
- Submit the sitemap at yourdomain.com/sitemap.xml in Google Search Console under Sitemaps.
- Check the bot settings of any security app or firewall in front of the site.
The crawler access guide covers each of these checks in detail.
Weeks 3 and 4: settle the identity
- Go to Settings > Store details and use one exact form of the business name.
- View the homepage source and count the Organization blocks. Keep one, with the real name, logo, URL and sameAs links to every profile you own.
- Go to Online Store > Preferences and rewrite the homepage title and meta description so they name the brand and what it sells.
Weeks 5 to 7: fix the product data
- Count the Product blocks on a product page. Turn off the markup in any app that prints a second one.
- Check that every Product block carries price, priceCurrency, availability and an identifier such as a GTIN or SKU.
- Fill in barcodes and SKUs on each variant in Products, starting with the best sellers.
- Compare the markup price with the page price on ten products.
Weeks 8 and 9: write the answers
- Rewrite the first paragraph of the 20 best sellers so it says what the product is, what it is made of and who it suits.
- Add a short FAQ in visible text to the pages buyers ask about most, written the way people ask.
- Write the shipping and returns answers as plain sentences on their own pages.
Weeks 10 and 11: connect the proof
- Check that the business name and address on review sites and social profiles match the site exactly.
- Make sure any rating in the markup matches the rating a shopper can see on the page.
- Add the press and profile links you own to the sameAs list in the Organization block.
Weeks 12 and 13: measure again
- Run the scan again and compare the two scorecards layer by layer.
- Ask ChatGPT and Perplexity "What is [your brand], and what do they sell?" and then ask the buying question for your category. Compare the answers with your notes from day 1.
- Write down the lowest layer that is left. That layer starts the next 90 days.
How does a one person tea shop spend its first 90 days?
A loose leaf tea shop on Shopify has 60 products and one person running it. Its baseline scan shows the Front Door passing and Entity Integrity as the lowest layer. Confirming that the door is open takes the owner one afternoon. The rest of weeks 1 and 2 goes to an early start on identity.
In weeks 5 to 7 the shop finds that its reviews app prints a second Product block with no availability. The owner turns that markup off and fills in barcodes on the 20 best sellers. By week 13 the scorecard shows which layers moved, and the shop has a dated record of every change.
Can I go faster than 90 days?
Phase 1 can take a day on a store with a default theme. The later phases take longer because they touch every product, and most owners do them alongside daily work. Going faster is fine as long as no phase is skipped. Finishing early does not help if the product data is only half done.
The work does not stop at day 90. AI visibility drifts when themes update and apps change, so a scan every month keeps the gains in place.
Questions owners ask
Can I go faster than 90 days?
The first phase can take a day on a store with a default theme. Going faster is fine as long as no phase is skipped, since each one depends on the one before it.
Where do I start if I feel overwhelmed?
Start with a scan. The lowest scoring layer on your scorecard is your first task, and if two layers are close, fix the one nearer the front door.
Do I need a developer for this plan?
From week 3 onward, most of the work happens in the Shopify admin and in your apps' settings. Editing robots.txt.liquid or removing markup printed by the theme is the part most owners hand to a developer.
What happens after day 90?
Scan every month and after every theme change or new app. Each scan shows whether a change broke something, and the lowest layer left starts the next plan.
Sources
- The State of AI Visibility 2026, Visibility Mesh, 216 Shopify storefronts scored in June 2026.