Ecommerce paid advertising is a different trade from lead generation advertising, and hiring an agency that does not know the difference is the commonest expensive mistake in the category. In lead generation, the ad and the landing page carry most of the weight. In ecommerce, a large share of performance is decided by the product feed: the titles, images, attributes and availability data that determine which searches your products are eligible to appear against and how they are presented. An agency that treats feed work as a technical afterthought is neglecting the single highest leverage asset in the account. This page covers what actually drives results, what to ask before signing, and how to read the numbers honestly.
The feed decides more than the bid
Shopping and other product driven formats are matched largely on the data you submit rather than on keywords you choose. That makes the product feed the primary optimisation surface: titles that lead with the attributes buyers actually search, accurate categories and product types, correct availability and pricing, and images that meet the requirements. Google's Shopping ads policies set out what may be advertised and the accuracy standards the data must meet, including that the price and availability shown must match the landing page, and products can be disapproved for mismatches that a merchant never notices until traffic disappears. Ask any prospective agency who owns feed management, whether it is inside the quoted fee, what tooling they use, and how disapprovals are monitored. An agency that says the developer or platform handles the feed is quoting for the smaller half of the job.
Value based bidding needs values that are true
Ecommerce is one of the few advertising contexts where genuine revenue can be sent back to the ad platform, which unlocks value based automation. Google's documentation on target return on ad spend bidding describes it as setting bids to help get as much conversion value as possible at the return target you set, which only works when the values are accurate and complete. Three things routinely break it. Sending order value including tax and shipping inflates apparent returns. Sending revenue rather than margin makes the platform favour high price low margin lines, which can raise revenue while lowering profit. Ignoring returns and cancellations in categories where they are common, such as apparel, makes the whole account look better than the bank account does. Ask an agency which figure it sends back, whether it is gross or net, and how returns are handled. The answer separates people who optimise a dashboard from people who optimise a business.
Attribution, and why the reported numbers disagree
Every ecommerce advertiser eventually notices that the platform reports more revenue than the store does. This is not usually fraud, it is a difference in counting: ad platforms attribute conversions to their own interactions across a lookback window, several platforms can each claim the same order, and the store counts each order once. The fix is not to argue about which is right but to agree in advance which system is the source of truth for decisions, and to look at the aggregate relationship between total advertising spend and total store revenue over a period rather than at per channel claims. Ask an agency how it handles this, whether it runs any form of incrementality testing, and what it would do if the store data and the platform data pointed in different directions. An agency with no answer to that last question will optimise toward whichever number is most flattering, because that is the one in its report.
What to establish before signing
Confirm account ownership first: the ad accounts, the merchant account, the analytics property and the tags should belong to your business with the agency granted access, so that years of learning do not leave with the relationship. Establish the fee shape and interrogate any percentage of spend arrangement, since it pays the agency more when your media cost rises. Get the scope in writing, specifically whether feed management, creative production, landing pages and promotional planning are included or billed separately, because the gaps between those four are where ecommerce engagements usually fail. Ask what the agency needs from you, since ecommerce advertising depends on merchandising decisions, stock information and promotional calendars that only you have, and an agency that asks for none of it has not planned to use them. The same discipline about ownership, fee shape and written scope applies whether the advertising sells products or service appointments, which is worth remembering if you also buy marketing for a service side of the business.
Questions people ask about ecommerce ppc marketing
What return on ad spend should I target?
There is no universal figure, because the right target depends on your gross margin, your repeat purchase rate and whether the campaign is meant to acquire customers or harvest existing demand. Work out the return at which a sale breaks even on contribution margin, then set the target above it with a margin for error, and revisit it as margins change.
Who should manage the product feed?
Whoever is accountable for performance, which normally means the agency running the campaigns, with your team supplying accurate source data. Splitting the feed from the campaign management creates a gap where disapprovals and stale availability data go unnoticed. Whatever the arrangement, it should be named in the contract.
Why does my ad platform report more revenue than my store?
Because platforms attribute orders to their own interactions within a lookback window and several can claim the same order, while your store counts each order once. Agree in advance which system is the source of truth for decisions, and judge the programme on the relationship between total spend and total revenue over a period.
Should I run Shopping, Search or both?
Most stores need both, doing different jobs. Product formats capture buyers who already know what they want and compare on price and image. Text ads cover the earlier, more descriptive searches where a product listing is a poor fit. Splitting the reporting between them is what tells you which is actually incremental.