Performance marketing for an online store is one of the few agency services where the buyer can check the work honestly, because the money in and the money out are both countable. That is exactly why the category attracts operators who report on the number that flatters them. A store owner who only ever sees blended return on ad spend can be losing money on every incremental order while the dashboard looks healthy. The agencies worth hiring start with your margins, not your budget, and they will say which channels they expect to be unprofitable at your price point. This page sets out how the work is priced, what genuinely moves results, and how to vet a shortlist before you hand over an ad account.
Fee models, and the incentives hidden inside them
Three models dominate. A flat retainer is predictable and neutral, and it is the easiest to compare across candidates. A percentage of ad spend is common and creates an obvious pull toward larger budgets, which is manageable if the percentage steps down as spend rises and if the agency is also accountable for efficiency. Performance based fees tied to revenue sound aligned but are only as good as the attribution behind them, and they quietly reward an agency for claiming credit for orders that would have happened anyway. Whichever model you pick, ask for the hours it funds and the named people delivering them, then ask what happens to the fee if you cut spend by half. The answer reveals whether you are buying a team or a percentage. A hybrid of a modest retainer plus a small spend component is the most common compromise and is usually fine if the numbers are stated plainly.
Contribution margin is the only number that settles arguments
Return on ad spend ignores cost of goods, shipping, payment fees, returns and discounting, and in categories with heavy returns it can be wildly misleading. The number that decides whether a campaign should keep running is contribution margin after all variable costs, measured against the cost of acquiring the order. A competent agency will ask for your product level margins, your return rate and your average discount before it proposes anything, and will build a target acquisition cost from them rather than picking a return multiple out of the air. It should also distinguish new customer acquisition cost from blended cost, because a programme that looks efficient is often just harvesting repeat buyers who would have returned unprompted. Ask a candidate how they would set the target and what they would do in the first month if the target proved unreachable at your prices.
The parts that are not advertising at all
Most stores that think they have a media problem have a site problem. Page speed, product page clarity, review presence, shipping cost transparency and checkout friction all move the same conversion rate that paid traffic has to overcome, and improving them makes every channel cheaper at once. Google publishes product structured data guidance covering how price, availability and review data should be exposed, which affects how listings appear in search results. Core Web Vitals, documented on web.dev, give a consistent way to measure the loading and interaction experience that a mobile shopper actually feels. Ask an agency what it would fix on the site before it raises budget, and whether it can do that work or only recommend it. This is also the boundary where broader ecommerce digital marketing services, covering search, email and lifecycle, get bought alongside pure paid media.
Vetting: the six questions worth asking
Ask for the smallest account size the agency accepts and the contract term, because both are stated in one email and both eliminate candidates quickly. Ask which platforms are managed in house and which are subcontracted, particularly creative production, since creative volume is now the main lever in paid social and outsourcing it slows iteration. Ask who owns the ad accounts, the pixel and the analytics property, and insist they are yours with access granted to the agency. Ask for the reporting template up front and check whether it separates new from returning customers. Ask for two references at your revenue scale rather than their largest client. Finally, ask what they would tell you to stop spending on, because an agency that has never recommended a spend cut has not been paying attention to anyone's margin but its own.
Questions people ask about ecommerce performance marketing agency
What size store is big enough for an agency?
Below roughly the point where fees and spend together consume a meaningful share of gross profit, an agency is hard to justify, and a good freelancer or an in house operator with a strong platform will do better. Ask any candidate to model the fee as a share of your gross profit rather than of revenue and the decision usually answers itself.
How should attribution be handled?
Expect a triangulation rather than a single source of truth: platform reported numbers, an independent analytics view and, when spend justifies it, incrementality tests such as geographic holdouts. Any agency that treats one platform's self reported conversions as the final word is optimising toward a number that platform is grading itself on.
How long does a fair trial period take?
Long enough to gather statistically meaningful data at your volume, which for smaller stores can be a full quarter and for higher volume accounts a matter of weeks. Agree the trial length, the target and the decision rule before the first campaign launches, so the review is a calculation rather than a negotiation.
Should the agency also handle creative?
In paid social, creative is the campaign, and separating strategy from production usually slows testing to a crawl. Either buy both from the same team or make sure your in house production can supply new variants weekly. An agency that requests creative and then waits three weeks for it is not the party that failed.