If you sell direct to consumer on Shopify and you want one place to put your next 40 hours of SEO effort, put it into collection pages. Not the blog. Product pages matter, but they win long-tail brand-plus-model queries you were going to get anyway. Collection pages are where category demand lands, and on most Shopify stores they are a headline, a grid, and nothing else. That is the gap. Everything else in this post, including how to show up inside ChatGPT and AI Overviews, gets easier once your category pages are actually worth ranking.
Shopify SEO starts with the pages Shopify gives you least help on
Shopify's product template is decent out of the box. Title, description, images, price, availability, usually some JSON-LD from the theme. A merchant filling it in properly gets most of the way there.
The collection template is where the platform leaves you alone. The default is a title, an optional short description that themes often truncate at 150 characters, and a product grid. Meanwhile the query you want, "cotton kurta sets for women" or "whey protein isolate 1kg" or "18k gold chains for men", is a category query. Google is ranking category pages for it. Yours has 40 words of copy above the fold and nothing below.
What we do instead: metafields for a proper intro block, a buying-guide section rendered below the grid, internal links to two or three sibling collections, and an FAQ that answers the questions customer support actually gets. Not 2,000 words of padding. Three to six hundred useful words, placed where a shopper would read them, plus links that give crawlers a reason to treat the page as a hub.
What one collection page position is actually worth
Numbers make this argument better than adjectives. Take a category term with 12,000 monthly searches in India. You sit at position 8 and, assume, pick up 1.5% of clicks: 180 sessions a month. Get to position 3 at roughly 7%: 840 sessions. Difference of 660.
At a 1.8% conversion rate that is 11.9 extra orders a month. At ₹2,400 AOV, ₹28,500 a month, ₹3.4 lakh a year, from one page. The content work behind that is maybe two days of writing and a day of development. The CTR percentages are assumptions you should replace with your own Search Console data, but run the arithmetic for your top six category terms before you decide SEO is slow.
The same calculation is why we tell some clients to stop. If your category terms have 400 searches a month and your AOV is ₹700, organic is not your channel this year. Spend the money on retention.
The technical debt that ships with every Shopify theme
A short list of things we find on almost every audit:
- Tag collections breeding URLs. Legacy tag-based views like /collections/shirts/linen are crawlable, mostly thin, and often left indexable from a theme migration three years ago. On a 4,000-SKU catalogue with a dozen tags per product this turns a 60-page site into thousands of crawlable URLs, and Googlebot spends its budget on the junk.
- Filters pointing nowhere useful. Native Shopify filters use query parameters and roll back to the parent collection. Usually correct. Occasionally wrong, because "linen shirts" is a real query with real volume and deserves a real page, not a filtered view of the shirts page.
- Duplicate product paths. Shopify's product URLs inside collections canonicalise to the clean /products/ path. Most themes handle it. Custom sections and third-party quick-view scripts sometimes do not, and then you have two crawlable copies of every product.
- Incomplete JSON-LD. Dawn-derived themes output product schema, but variant-level offers, GTINs, brand and review data are frequently missing or hardcoded wrong after a developer has been in the file.
- Redirect rot after migration. WooCommerce and Magento URL structures do not map onto Shopify's forced /products/ and /collections/ prefixes, so every migration needs a mapped redirect file. We still see stores that lost a third of their organic traffic on launch day and never diagnosed it. If that is you, our Shopify migration notes cover the mapping properly.
None of this is glamorous. All of it is cheaper to fix than to rank around.
Large catalogues need search and filtering that search engines can follow
Past roughly 800 SKUs, on-site search and faceting become an SEO problem as well as a UX one. Shoppers who cannot find the sub-category bounce, which Google notices in aggregate. Crawlers that find 14,000 parameter URLs waste the budget you need on your money pages. The fix is to decide, deliberately, which facet combinations deserve indexable landing pages and which stay as parameters, then build filters that support both. We built FilterPro partly because bolting a filter app onto a store and hoping for the best is how most of these messes start.
Speed is a ranking input, but it is a bigger conversion input
Core Web Vitals will not lift you from page three to page one. What a 4.2s LCP on mobile does is quietly cost you a chunk of the traffic you already earned, particularly on 4G in tier-2 cities during a festive sale when your app stack is at its heaviest.
We target LCP under 1.8s on a mid-range Android on 4G, and the wins usually come from the same three places: hero images served at the wrong size, three review or popup apps loading render-blocking scripts, and a theme carousel nobody scrolls. Our approach to Shopify speed optimization starts by deleting apps, which clients rarely enjoy hearing.
Getting cited by AI assistants
Two different behaviours are getting lumped together, and they need different responses.
The first is AI Overviews on Google for informational queries. "How to wash linen", "difference between whey concentrate and isolate". These clicks are going away. If your blog traffic is mostly this shape, expect it to decline regardless of what you do, and stop optimising for it.
The second is assistants answering shopping questions: someone asking ChatGPT or Perplexity or Gemini for the best options in a category, then following a link. What we see cited in those answers, consistently, is not brand sites. It is editorial roundups, forum threads, marketplace listings and review platforms. So the work splits:
- Be mentioned where assistants retrieve. Category roundups in publications your customers read, genuine participation in the communities that discuss your category, review platform presence with enough recent volume to look alive.
- Make your own pages machine-readable and unambiguous. Specifications in a table, not buried in a paragraph of marketing copy. Materials, weight, dimensions, ingredients, certifications, warranty, return window, all stated in plain text. An assistant that cannot extract a spec will not recommend the product.
- Keep your entity information identical everywhere. Brand name, founding year, city, contact details, product naming. Assistants stitch a picture of your brand from scattered sources and inconsistency makes you less quotable.
- Submit to Bing Webmaster Tools. Several assistant products lean at least partly on Bing's index, and most Indian D2C brands have never touched it.
- Check your robots.txt. If a previous developer blocked AI crawlers as a reflex, you have opted out of the channel you are now trying to enter. Decide that on purpose, not by default.
On llms.txt: we have added it to a few stores, it takes ten minutes, and we have seen no evidence any major assistant reads it. Add it if the ten minutes is free. Do not let anyone bill you for an AI visibility strategy that is mostly that file.
Structured data that earns something back
Product schema with price, currency, availability and variant offers. Review schema only where the reviews are real and the app is writing them into the page source rather than a JavaScript widget. BreadcrumbList, because it changes how your result is displayed. Organisation schema on the homepage with sameAs links to your verified social profiles.
For Indian stores, put the GST-inclusive price in the schema if that is what the customer sees at checkout. A mismatch between schema price and displayed price is the most common reason a Merchant Center feed gets suspended, and it is a ten-minute fix that people spend three weeks arguing about.
Skip FAQ schema on product pages. Google has largely stopped showing it, and the markup clutter is not free.
Content, minus the treadmill
Publishing weekly because a content calendar says so is how stores end up with 90 posts and 200 organic sessions. The blog earns its keep in two ways: supporting category pages with internal links from genuinely useful comparison and sizing content, and answering the pre-purchase questions your WhatsApp support team fields every day.
Timing matters more than volume in India. A Diwali gifting guide published in late September ranks for nothing, because the page needs weeks to accumulate crawl attention and links. Publish festive content by the first week of August, refresh last year's URL rather than creating a new one, and keep the same slug year over year. That is most of what we do on SEO and content for seasonal categories.
If you sell across markets, sort hreflang before anything else
Shopify Markets will happily serve the same catalogue to India, the UAE and the US on subfolders. Without correct hreflang and self-referencing canonicals, you get three versions of every page competing, and Google picks the wrong one for the wrong country. UAE shoppers land on rupee pricing and leave. We see this on maybe half the multi-market stores we audit on the first pass, including ones built by competent developers, because the defaults look fine until you check the rendered head tag on a market-specific URL.
A sensible first month
Week one, crawl the store and count indexable URLs against your actual product count. If the ratio is worse than about three to one, you have a crawl problem before you have a content problem. Week two, fix canonicals, redirects and schema. Weeks three and four, rewrite your top five collection pages properly and measure position change over the following eight weeks.
If you would rather have someone else find the crawl bloat, our free store audit covers the technical layer and tells you which of your category pages is closest to moving. We will also tell you if organic is the wrong channel for your margin structure, which happens more often than you would think.