Paid search for a shop is not mostly about keywords. The majority of the spend usually runs through shopping campaigns, and those are driven by the product feed: the titles, the images, the attributes, the prices and the availability you send. An agency that cannot work on the feed is optimising bids on data it has not looked at, which is the commonest reason an ecommerce account plateaus while everybody is busy.
The feed decides which searches you appear for
There are no keywords in shopping: the system matches queries to your product data. Titles carrying the brand, the product type and the attributes people search by, correct categories, complete identifiers, real images and accurate availability are therefore the campaign. Ask a prospective agency what they would change in your feed in the first week, and treat a bid-management answer as evidence they have not looked.
Margin, not revenue, decides what a click is worth
Different products carry very different margins and return rates, and a single return-on-spend target across the catalogue systematically overspends on the cheap high-volume lines and underspends on the profitable ones. Feed products with their margin and segment the campaigns by it. This is the largest available improvement in most shop accounts and it needs data from finance rather than from the platform.
Brand and non-brand must be separated and reported apart
Searches for your own name convert cheaply and would largely have arrived anyway; searches for the product category are the ones you are actually buying. Blended together, the account's average looks excellent and hides whether the money spent on new customers works at all. Separate them, report them separately, and judge the non-brand campaigns on their own numbers.
Out of stock and price changes have to reach the feed fast
Advertising a product you cannot ship wastes the click and produces a complaint, and a price mismatch between the feed and the site gets products disapproved. The update frequency of your feed is therefore an advertising setting, not a technical detail. Establish how often it refreshes and what happens on a busy sale day before increasing any budget.
Judge on new customer profit, not on blended return
The honest measures in ecommerce are contribution after cost of goods, shipping, fees and returns, and the cost to acquire a customer who was not already yours. Both require data the advertising platform does not have. An agency comfortable being measured that way will ask for it in the first meeting; one that is not will report the platform's own arithmetic and call it return.
Questions people ask about ecommerce ppc agency
Why does the product feed matter so much?
Because shopping campaigns have no keywords: the system matches searches to your product data. Titles, attributes, categories, identifiers, images and availability decide what you appear for.
Should we use one return-on-spend target?
No. Margins and return rates differ enormously by product, so a single target overspends on cheap high-volume lines and underspends on profitable ones. Feed the margin and segment by it.
Why separate brand campaigns?
Because searches for your own name convert cheaply and would mostly have arrived anyway. Blended with category searches they make the account's average look excellent and hide whether new-customer spend works.
What should an ecommerce agency be measured on?
Contribution after cost of goods, shipping, fees and returns, and the cost of acquiring a genuinely new customer. Both need data the platform does not have, which is why the request itself is informative.