A store selling trail shoes has a few hundred products, a blog nobody reads, and a list of SEO advice long enough to fill a quarter. The temptation is to start with the buying guides, because writing feels like progress. Hold off: the platform under the store is almost certainly leaking rankings already, through duplicate URLs, a slow theme and product pages that say nothing, and a guide that feeds a collection Google has folded away feeds nothing. The first thirty days stop the leaks, record the baseline, and build the keyword map the next sixty days are built on.
Why leaks come before content
Why some stores rank and yours doesn't explains the pipeline: Google has to find a page, understand it as distinct, and trust it before it can rank. Store platforms strain the middle step because they generate pages in bulk: one product at its own URL and under every collection it belongs to, a page for every tag, another for every filter combination. Google's guidance on consolidating duplicate URLs describes what happens next: it picks one version to represent the set and folds the others away, with no penalty, and the folded versions never rank. Crawl attention spent on copies is attention not spent on the collections you want seen.
Fixing that once is cheaper than fixing it under twenty new guides. The baseline matters because in ninety days you'll want to know whether any of this moved revenue, and you can only answer if you wrote down day one.
Week 1: see what Google sees, then write the baseline
Three jobs, in this order.
- Put measurement in place. Verify the store in Google Search Console and set up analytics so organic and AI-referred sessions are separated and tied to orders, following Tracking revenue from search and AI.
- Write the baseline. One sheet: pages Search Console reports as indexed, clicks and impressions by collection and product, organic revenue, and the top queries. Add an AI row: ask ChatGPT, Gemini and Perplexity the "what should I buy" question a customer would ask and note whether your products are named, as AI shopping answers describes.
- Find the leaks. Run The 30-minute store quick-wins audit and check the settings in Platform SEO defaults. Then read the reasons Search Console's Pages report gives for pages it hasn't indexed. Duplicates, "crawled, currently not indexed" and pages blocked by robots rules are the leak list.
In place by day 7: Search Console and analytics collecting, a baseline sheet, and a leak list.
Week 2: canonicals, speed and the product basics
Fix in order of what stops Google seeing a page, not what's easiest.
- Duplicates first. Duplicate URLs and canonicals covers collection-path product URLs, tag pages and parameters; Variants and duplicate content covers sizes and colours. Every copy needs a canonical tag pointing at the one URL you want ranked. Faceted navigation and filters decides which filter URLs earn a place in the index.
- Speed on the collection and product templates. Store speed and themes walks through apps, scripts and images. web.dev's Core Web Vitals thresholds count a page as good at a Largest Contentful Paint of 2.5 seconds or under, an Interaction to Next Paint of 200 milliseconds or under, and a Cumulative Layout Shift of 0.1 or under, measured from real visitors.
- The product basics, on your best sellers only. Rewrite titles and descriptions for the buyer's search and the buyer's doubt using Product titles and descriptions. Validate the structured data using Product schema and rich results: Google's structured data gallery lists Product among the types that earn a rich result, which is how price and availability appear under a listing. Fix image weight and alt text using Product images and alt text.
- Pages that still rank for products you no longer sell, following Out-of-stock and discontinued products: redirect, keep or replace.
In place by day 14: one canonical URL per product and collection, filter URLs decided, both templates measured, and the best sellers rewritten and validated.
Week 3: build the keyword map
Start from how buyers describe what they want, not from how the catalogue is organised, following Product and category keyword research. Sort by intent using Buying intent for stores: "best", "cheap", "vs" and brand searches each want a different page.
Which page catches which search assigns each keyword to a page type: category searches to collections, product searches to products, questions to guides. Collections worth creating runs every candidate collection through three tests, so you build the ones with demand and skip the ones that would be thin.
Keep the map to what you can serve in sixty days; anything larger is a spreadsheet, not a plan.
In place by day 21: one primary keyword per collection and per product group, a short list of collections to create, and a list of guide topics with the collection each one feeds.
Week 4: structure, links and the calendar
Arrange the collections so the hierarchy mirrors how people shop and every product sits a few clicks from home, using Collection structure and navigation. Create the collections the map asked for, each with a few lines of text saying what the range is and who it's for.
Then move the links. Internal links to money pages shows how to link existing blog posts to the collections and products they already talk about. Old posts that name a product without linking it are the cheapest links you'll ever add.
Start the routines that run for the rest of the plan: add the map's keywords to rank tracking so history starts today, book a weekly review slot, and draft the calendar for days 31 to 90 from the guide topics list, one cluster per top collection.
In place by day 30: a store Google can index end to end, a keyword map, tracking running, and a dated calendar for the next sixty days.
What to take away
- Store platforms create duplicate pages in bulk, Google keeps one version of each set, and fixing that comes before anything new is published.
- Week one records the baseline and finds the leaks; week two fixes them in order of what stops Google seeing a page.
- The keyword map gives one primary keyword to each collection, product group and guide, and stays small enough to serve in sixty days.
- By day 30 you have clean URLs, validated product pages, a keyword map and a calendar: everything the next sixty days need.
Next
Days 31 to 90: content, feeds and AI turns the calendar into guides that feed collections, a clean product feed, and the signals AI assistants use.