Ecommerce PPC marketing: the feed, the margin and the account structure that decide your results

In ecommerce paid search, the product feed is the campaign. Titles, attributes, images, price and availability decide which queries your products are shown for, so feed quality sets the ceiling on everything a bidding strategy can do afterwards. The second decision that separates profitable accounts from busy ones is segmenting by margin rather than by category, because a single blended return on ad spend target across a catalogue with 15 percent and 60 percent margin products overspends on the thin items and underspends on the profitable ones. Get those two right, send the platform clean conversion values with profit rather than revenue where you can, and the structure and bidding questions become straightforward.

This page covers paid search for online retail. For lead generation accounts, see a page on PPC management and account structure.

The feed does most of the work

Shopping and Performance Max match your products to queries using the data you submit, not keywords you choose. That makes the feed the highest return place to spend an afternoon.

• Titles. Lead with the words buyers search. For most categories that is brand, product type, then the distinguishing attributes: model, size, colour, capacity, material. Front load the important terms because titles get truncated in display.
• Product type and category. Set the Google product category accurately and use your own product type field as a hierarchy you can segment campaigns by later. This field is how you split the account by margin without renaming anything.
• Attributes. GTIN, brand, condition, colour, size, material, age group and gender where they apply. Missing attributes restrict eligibility and reduce matching quality.
• Images. The primary image on a plain background, high resolution, no promotional overlays, which are a disapproval cause. Add additional images for the formats that use them.
• Price and availability. These must match the landing page exactly. A mismatch gets items disapproved, and it is the most common reason a feed silently shrinks.
• Custom labels. Use them for margin band, stock depth, seasonality and bestseller status. They cost nothing and they make every later segmentation possible.
• Sale price and promotions. Submit sale price properly rather than editing the price field, and submit promotions so they show in the listing.

Check the diagnostics in your merchant account weekly. Disapprovals are usually a handful of fixable data issues, and every day one sits there is a day those products cannot be shown.

Segment by margin, not by category

Write your margin band into a custom label, then split campaigns on it and set a different target for each band.

Band	What goes in it	How to treat it
High margin, strong demand	Your best contribution products with search volume	Own first position, loosen the target, fund growth here first
High margin, low demand	Profitable but rarely searched	Keep live at a modest budget, support with content and remarketing
Low margin, strong demand	Price competitive commodity lines	Tight target, bid only where you are competitive on price, use for new customer acquisition deliberately
Clearance and overstock	Aged inventory	Separate campaign with a volume goal, judged on stock cleared rather than return on ad spend
New and unproven	Recent additions with no data	Its own campaign so new items get impressions rather than being starved by proven lines

Send profit as the conversion value where your platform allows it, by passing margin rather than order value. Bidding then optimises for the thing you actually want, and the whole account gets better without a single bid change.

Campaign types and what each is for

• Standard Shopping gives you control: you choose the products, see the search terms, and add negatives. Use it where you want command of a product group.
• Performance Max reaches across inventory and placements, and it is driven by your feed, your asset groups and your audience signals. Feed it good assets, set asset groups per product theme rather than one group for everything, and exclude the brand terms you already own if you want a clean read on incremental sales.
• Search campaigns cover the queries no product title matches: category pages, buying guides, and competitor comparisons. Keep brand in its own campaign so you can see what it costs and what it earns.
• Remarketing and dynamic remarketing recover abandoned carts and product views, with frequency capped so a browser is not followed for a month.

The structure decisions that keep an account readable

Split campaigns where you need a different budget, a different target or a different report. Keep them together where you do not, because excessive splitting starves every piece of data. For most catalogues that means campaigns by margin band, asset or ad groups by product theme, and brand separated from non brand everywhere.

Negatives matter more in retail than anywhere else. Exclude the words that signal no purchase or the wrong purchase: repair, manual, used, free, hire, parts for a brand you do not stock, job titles, and in many cases DIY instructions. Review search terms weekly for the first month, then monthly.

Measure what paid advertising added

Four reports answer the questions that matter.

• Contribution by campaign. Revenue minus product cost minus ad spend. A campaign at a healthy return on ad spend can still lose money on thin margin products.
• New versus returning customer revenue. Acquiring a first order at a loss is a decision, not a mistake, provided you know the repeat rate.
• Brand versus non brand. Brand campaigns harvest demand other work created. Reporting them together inflates the paid result and hides what prospecting is doing.
• Product level performance against stock. Clicks on out of stock or soon out of stock items are the fastest avoidable waste in retail accounts.

Be wary of any agency or tool that promises a specific return on ad spend before seeing your margins and your feed. The auction and your product economics decide that number together, and a promise made in advance of both is a promise about something the seller does not control.

What return on ad spend should an ecommerce store aim for?

The break even figure is one divided by your gross margin. At 40 percent margin, break even is 2.5, so a profitable target sits above that with your fixed costs and growth appetite deciding how far above. Calculate it per margin band rather than for the whole account.

Is Performance Max better than Standard Shopping?

They do different jobs. Performance Max covers more inventory and placements with less manual work, Standard Shopping gives visibility and control over product groups you care about most. Many strong accounts run both, with the products that need control held in Standard Shopping.

How much budget do I need to start?

Enough for the chosen campaign to collect roughly thirty conversions a month, because bidding needs data to work with. Calculate it from your average order value and conversion rate, and if the number is beyond you, narrow the product set rather than spreading the same budget thinner.

Should I bid on my own brand name?

Usually yes, because the clicks are cheap, they defend against competitors bidding on you, and they let you control the landing page and the offer. Keep it in a separate campaign and judge it on its own, never blended into the headline figure.

How often should I update the feed?

Price and stock should sync at least daily, and more often during sales. Titles and attributes deserve a proper review each quarter and before every peak season, starting with the products that carry the most contribution.

The question that never meets your feed

Your feed enters you into product searches. A growing share of product research starts as a question: someone asks ChatGPT, Gemini or Perplexity which brand to buy for a specific need, and the answer names two or three, using product pages, reviews and comparison articles rather than ads. Clean feed data helps because it makes your product details match across sources. Test it with the question a shopper would type for your best margin product, and see whether you are named and which pages are cited. AI Knows Us measures which assistants name a business and the sources behind each name.