Merchant Center Feed Optimisation: The Highest-Leverage Work in Ecommerce Paid Search

In Shopping and Performance Max, there are no keywords. Google decides which auctions your products enter by reading your product feed, matching your titles, descriptions and attributes against queries. Which means the feed is not a data formality sitting between your store and your campaigns. It is the targeting layer. Every optimisation lever the old search campaigns put in keywords now lives in product data, and most ecommerce accounts leave it wherever their platform's default export dropped it.

That is why feed work is reliably the highest-leverage effort in ecommerce paid search: it improves every campaign that consumes the feed simultaneously, including Performance Max placements you cannot otherwise influence much, a point we made in our look at PMax. This article covers where the leverage concentrates.

Titles: The Keyword Research You Still Control

The title is the single most heavily weighted matching signal, and Google's own guidance is to use the available 150 characters deliberately, knowing only around 70 display in most placements. Two consequences follow.

Front-load what matters. The visible first 70 characters should carry brand, product type and the highest-value differentiator (size, colour, model, material) in roughly the order customers search for them. Category conventions differ: fashion queries lead with brand and gender, parts and components lead with compatibility, consumables lead with product type and count. Match your title structure to how your category actually searches, which your own search terms report will tell you.

Use the full length for matching. Characters 70 to 150 are rarely seen but still matched, so specifications, secondary use cases and variant details belong there. A title that reads "Merino Wool Jumper" enters a fraction of the auctions that "Acme Merino Wool Jumper, Men's Crew Neck, Navy, Machine Washable, Sizes S-XXL" enters, and the second one enters them with better expected click-through, which feeds directly into auction competitiveness.

Ecommerce platforms typically generate titles from product names written for your website, not for search matching. Feed rules (in Merchant Center itself or a feed tool) that concatenate brand, name and variant attributes into proper title structures are usually the first hour of work and the largest single gain.

Attributes Beyond the Mandatory

Google's product data guidance is blunt: richer, more accurate attributes mean better matching and better placement eligibility. The ones that repay attention beyond the required set are GTINs (which unlock matching against Google's product catalogue and competitive benchmarking), product type (your own taxonomy, which supports both matching and clean campaign subdivision), colour, size, material and gender for variant-level matching, and sale price fields handled properly so promotions display. Availability and price must be exact against your landing pages, because mismatches drive disapprovals and account-level trust problems that suppress everything.

Treat disapprovals and warnings in Merchant Center's diagnostics as a weekly operational metric. Products sitting disapproved are auctions you exited silently, and the diagnostics tab is the only place that tells you.

Price Competitiveness: The Uncomfortable Attribute

Shopping is a comparison format, and price is visible in every impression. Google's price competitiveness reporting (unlocked by GTIN coverage) shows how your prices sit against the market at product level. This is commercial input as much as marketing input: products priced far above benchmark convert poorly however good the feed, and bidding harder at them mostly buys expensive proof of the pricing problem. The practical use is segmentation, pushing spend toward products at or under benchmark and pulling back where you are structurally uncompetitive, which is straightforward once your campaign structure follows feed attributes like product type and margin bands.

Treat the Feed as a Campaign

The mindset shift that separates accounts where feed work compounds from accounts where it decays: the feed gets managed on a cadence, like a campaign, not fixed once, like plumbing.

• Weekly: clear new disapprovals, check availability sync after any platform change.
• Monthly: review search terms against titles to find vocabulary mismatches (customers searching "hoodie" while your feed says "hooded sweatshirt" is lost matching), and iterate title rules for the biggest product groups.
• Quarterly: audit attribute coverage, review price competitiveness by category, and prune or fix chronically underperforming SKUs rather than letting them dilute campaign averages.

Title changes alter matching, so expect a re-learning wobble after large rewrites and batch them rather than dripping them daily. Measure feed work the way you would measure any structural change: impression volume and search term breadth first, then click-through, then efficiency, over weeks rather than days.

None of this is glamorous, which is precisely why it is still an edge: most competitors' feeds are their platform defaults. If you want a feed audit, or the title and attribute architecture built properly with your campaign structure on top, our paid search team does this work constantly. Get in touch.