Paid ads for online retailers, and the data underneath them

Shopping performance is a feed problem before it is a bidding one

What a product appears for is determined largely by the data you submit, not by how the campaign is structured.

Titles do most of the work. A title carrying the brand, the product type and the attributes people search matches far more queries than one carrying an internal product name.

Category assignment, attributes like size and color, and identifiers determine which searches a listing is eligible for at all. Missing attributes silently remove eligibility.

Image quality and availability accuracy affect performance directly, and stock data that lags reality wastes spend on products that cannot ship.

None of that is adjusted in the campaign interface, which is why accounts can be managed diligently and perform poorly. The lever is upstream, in the data.

Feed work is unglamorous, it is usually the highest-return improvement available, and it benefits organic and marketplace listings at the same time since they draw on the same information.

Margin decides everything and the platform cannot see it

An account optimizing on revenue will systematically favor the products you make least money on.

Platforms optimize toward the conversion value you report, which is usually revenue. Revenue and profit are not the same, and in retail they frequently diverge sharply.

A campaign can look excellent while pushing high-revenue, low-margin products and starving the ones that actually pay for the business.

Reporting margin rather than revenue as the conversion value, where the platform supports it, changes what the system optimizes toward and often changes it substantially.

Returns are the second correction and are usually ignored entirely. Categories with high return rates report revenue that never persists, and the account keeps buying more of it.

Neither adjustment is difficult and both require data the business has and the account does not, which is the recurring theme of retail advertising.

Marketplace advertising is a separate economy

Promoted placement inside a marketplace behaves differently from search advertising and needs its own arithmetic.

Advertising within a marketplace reaches shoppers who have already decided to buy something in that category, which produces strong conversion and a narrow audience.

The economics include the marketplace commission on top of the advertising cost, so the number that matters is what remains after both.

It also rarely produces a customer relationship. The marketplace owns the customer, which means repeat purchase and any direct marketing afterward generally are not available to you.

That is a strategic question rather than a channel question. Volume through a marketplace is volume you do not own, and building the business on it has consequences that appear later.

Evaluate it separately from your own advertising with its own reporting, since blending the two hides which is working and lets one subsidize a poor view of the other.

Subscription and auto-renewal offers carry disclosure obligations

Where an ad promotes a recurring arrangement, requirements attach to how it is presented.

Subscription and auto-renewal disclosure is regulated, and negative-option billing has drawn particular enforcement attention.

That reaches the advertisement as well as the checkout, since an offer presented in an ad is where the customer's expectation is set.

Short formats make it harder, which is a reason to move the offer detail to the landing page rather than to compress it out of existence.

Introductory pricing is where this goes wrong most often. A first-period price advertised without the renewal price and the cancellation mechanism is precisely the pattern the rules address.

This belongs with your advisor rather than with an agency's judgment. What we can do is make sure whatever they establish is presented clearly and where people will actually see it.

The path has many touches and the last one takes the credit

Retail purchases involve several visits across several channels, and default reporting gives the whole outcome to whatever came last.

A customer may see a social post, search the brand, click an ad, leave, return through email and buy. Last-click reporting credits one of those and dismisses the rest.

That produces a predictable distortion: brand search and retargeting look outstanding because they sit near the end, and everything upstream looks unaccountable.

Accounts optimized on that logic drift toward capturing demand that already existed and stop creating any, which works until the pipeline of existing demand thins.

The corrections are imperfect and worth using. Longer windows, comparing periods when upstream activity was paused, and watching total new customers rather than attributed conversions.

The most reliable check is whether new customer counts move when upstream spend does, which no attribution model is needed to observe.

Where the account ends and the store begins

Advertising can only deliver the visit, and everything after it is a different discipline.

A campaign that produces qualified traffic into a slow, confusing or expensive checkout has done its job and produced nothing, and the reporting will blame the advertising.

Shipping cost, checkout friction and returns policy affect measured advertising performance more than most bidding decisions do.

That means diagnosing a poor account frequently starts on the site rather than in the campaign, and it is worth checking before restructuring anything.

We cover the store side under conversion optimization rather than here, and the two are usually worth looking at together.