Paid Advertising · E-Commerce
A good return on ad spend and a profitable month aren’t the same thing.
The revenue figure at the top of an ad dashboard is not the amount your business keeps. Cost of goods, shipping, fees, discounts and the spend that bought the order all come out of it first — and the platform can’t see any of that.
What survives the sale
One order, taken apart.
A sale is reported as a single revenue figure. Underneath it sits a set of costs that vary enormously between businesses and even between products in the same catalogue. What’s left after all of them is contribution — and that tells you considerably more about whether paid acquisition is sustainable than a return-on-ad-spend figure does on its own.
What one order actually contains
Segment widths are illustrative, not a benchmark or a projection. The real structure differs by business, by product, and often by season — a made-to-order item and a drop-shipped one behave nothing alike. The point is only that the revenue number is not the amount kept.
What return on ad spend compares
Return on ad spend compares the top of that bar to one segment inside it.
Everything in between is outside the calculation — which is why it’s a useful input and a poor standalone verdict.On benchmarks
There is no universally good ROAS
The return a campaign needs in order to be worth running is set by the economics of the product and the business, not by an industry figure.
A hypothetical 3:1 could be comfortable for one store and inadequate for another, depending on margin, fulfilment costs and how much discounting is happening.
On acquisition cost
Cheaper isn’t automatically better
A falling cost per acquisition is usually welcome, but it can sit alongside worse economics — if discounting increased, if the product mix shifted toward thinner-margin items, or if order economics changed underneath it.
Where a business has genuine repeat purchase, what a customer is worth beyond the first order can reasonably change the acquisition cost it can afford. Not every store has that, and we don’t assume it.
On attribution
Platform revenue is attributed revenue
Reported revenue is attributed under that platform’s measurement methodology. Attribution windows differ, view-through conversions may be counted, and two platforms can each claim influence over the same sale, so it may not reconcile exactly with your books.
Estimated revenue is also not the same as directly tracked revenue. Attribution is useful and imperfect; we report it as such.
On reading the numbers
Averages hide the products losing money
A campaign can look healthy overall while specific products inside it are acquiring customers at economics you’d reject if you saw them alone. We read performance at product and category level as well as campaign level, so spend can move toward what’s genuinely worth selling. Where the product page or checkout is what’s limiting conversion, that’s a web development problem rather than a bidding one.
Contribution is not the same as net profit — fixed overhead, cash flow, inventory and repeat-purchase behaviour all still matter, and those sit with you. But it’s a far more decision-useful number than return on ad spend alone when deciding whether to scale. How acquisition, the storefront and retention fit together is covered in our complete guide to e-commerce marketing.
The upstream constraint
Your product feed decides how well the ads can possibly perform.
Economics decide whether the traffic is worth buying. The feed decides how much of it you’re even eligible for. Shopping, Search and dynamic remarketing all read from the same product data underneath them, which is why we audit it before touching a campaign.
What we usually find
Feed left as-is
- Shopping campaigns
- Search campaigns
- Dynamic remarketing
Products show for fewer searches, lose auctions they should win, and the campaigns still look fine in the dashboard — quietly working under a lid.
What we build
Feed built properly
- Shopping campaigns
- Search campaigns
- Dynamic remarketing
Products become eligible for more of the right searches, compete on their real merits, and give bid management something worth optimising.
Same product. Two ways of describing it to the platform.
Weak feed entry
“Sweatshirt”
- Brand
- Not provided
- Type
- Clothing
- Size / Color / Material
- Not provided
- GTIN
- Not provided
- Availability
- In stock
- Landing page
- Links to general category page
Useful feed entry
“Men’s Cotton Crewneck Sweatshirt, Charcoal, Large”
- Brand
- Example Brand
- Type
- Apparel > Men’s > Sweatshirts
- Size / Color / Material
- Large / Charcoal / 100% cotton
- GTIN
- 0123456789012
- Availability
- In stock — updated daily
- Landing page
- Links directly to this exact product
Same product, two ways of describing it. The right column gives Shopping, Search and remarketing something to actually match against.
This is the difference between an account that responds to optimisation and one that quietly refuses to. It’s also why we start at the feed rather than the campaigns, even when the campaigns are what you came in asking about.
How it runs
Four stages, in this order.
The order matters here. Building campaigns before fixing the feed means rebuilding them afterward.
Feed & account audit
The product feed, existing campaigns, and purchase tracking before anything else. Titles, attributes, availability, pricing accuracy, image quality. Most stores are losing eligibility at this level and have never been told.
Shopping & Search build
Campaigns structured around your actual catalogue rather than a template — segmented by product and category so budget can be steered deliberately. Where you’re able to share cost or margin data, that segmentation can follow profitability rather than revenue. Microsoft added where it earns its place.
Remarketing & purchase tracking
Dynamic remarketing wired to your real product data, so someone who abandoned a specific item sees that item. Purchase tracking installed at cart and checkout level, so revenue is attributed to the campaign that earned it — with the attribution caveats above understood rather than ignored.
Ongoing optimisation against the economics
Performance reviewed by product and category, not just by campaign, and judged against the economics you’ve shared with us rather than revenue alone. A product that sells well at a bad margin is not necessarily a successful advertising outcome, and the reporting should say so. Where we don’t have cost data, we’ll tell you what we can and can’t see rather than implying otherwise.
Measured outcome
Product sales, remarketing, and honest arithmetic.
Supplement manufacturer
708.22%
Value-to-cost ratio
Display and remarketing built around what a purchase is actually worth
Image ads with ongoing testing, Search campaigns after keyword research, and remarketing that brought back visitors who hadn’t yet purchased. The account was managed against purchase value rather than click volume.
- Directly tracked conversions increased 245%, from 490 to 1,689
- Average cost per conversion: $9.96
The $1,935,403.02 figure is estimated total revenue, not directly tracked revenue. It includes $219,085.12 in estimated view-through revenue, calculated as a $62.56 average purchase price multiplied by 3,502 total view-through conversions, then added to directly tracked conversion value.
Online retailer
47%
PPC revenue growth in two months
An inherited paid search account restructured and given a negative keyword strategy. PPC revenue grew from $6,567.07 in September to $9,677.27 in November, and cost per conversion fell 16%, from $25.97 to $21.83.
Or see what one looks like: a real, anonymized e-commerce analysis, from research to recommended sequence.
Scope & fit
Where these campaigns run.
Search and shopping platforms, where someone is actively comparing products and ready to buy. Platforms are how the work gets executed, not how it gets decided.
- Google Shopping
- Google Search
- Performance Max
- Google Display
- Dynamic remarketing
- Microsoft Shopping & Search
Facebook, Instagram, TikTok and Pinterest are handled separately under Paid Social — different targeting, different creative, different management. Plenty of stores run both, and they work better together once the feed is right — though on a limited budget it’s often better to start with one. Meta Ads vs. Google Ads for e-commerce covers how to choose. This service sits under Paid Advertising; if you sell through calls and quotes rather than checkout, lead generation is the other path. This service is one part of our wider e-commerce marketing work. On the organic side, we’ve published SEO work for an e-commerce apparel brand, which is separate from paid advertising but deals with the same product-page economics.
A good fit
- You sell physical products online and have product data we can work with
- You know roughly what an order costs you, or are willing to work it out
- You accept that the feed is foundational, not an afterthought
- You’re willing to fix product pages or checkout if that’s what’s limiting you
- You want spend judged against economics, not revenue alone
Probably not
- You have no product data that can become a working feed
- You want ads running without Shopping as part of the foundation
- You judge performance by clicks rather than purchases
- Return on ad spend is the only number you want to look at
- You’re shopping for the cheapest management fee available
Questions
E-commerce FAQs.
What is a good return on ad spend?
There isn’t a universal figure, and anyone offering one without knowing your costs is guessing. The return a campaign needs depends on what the product costs you to make, ship and process, and how much discounting is involved.
The same ratio can be healthy for one store and unworkable for another. What matters is what’s left after those costs, not the ratio in isolation.
Why can a high return on ad spend still produce weak economics?
Because the calculation only sees revenue and ad spend. It has no visibility into cost of goods, shipping, payment fees, or the discount codes applied at checkout. A campaign can push revenue efficiently while the orders it produces contribute very little — particularly if the growth came from heavy promotion or from lower-margin products.
What’s the difference between return on ad spend and profit?
Return on ad spend is revenue divided by ad spend. Contribution is what remains from an order after the variable costs of fulfilling it and acquiring it. They answer different questions.
Contribution is also not the same as net profit — fixed overhead, cash flow and inventory still sit outside it. It’s a considerably better basis for deciding whether to scale than the ratio alone, not a complete picture of the business.
How do margins affect how much we can spend?
Directly. A thin-margin product can only afford a small acquisition cost before an order stops contributing anything. A high-margin product can absorb considerably more. That’s why the same campaign settings can be right for one catalogue and wrong for another, and why we ask about product economics rather than optimising to revenue by default.
We walk through the full method in how much an e-commerce business should spend on marketing.
Does customer lifetime value change the picture?
Where a business has genuine repeat purchase, yes — what a customer is worth beyond the first order can justify a higher acquisition cost on that first sale. But plenty of e-commerce businesses are largely one-off, and assuming repeat value that doesn’t exist is a common way to over-spend. We won’t assume it; if you have the data, we’ll use it.
What actually is a product feed?
The structured data behind your product listings: titles, descriptions, prices, images, availability, brand, condition, identifiers. Google and Microsoft read it to decide which searches your products can appear for and how they’re presented.
Thin titles, missing attributes, or stale availability quietly limit which auctions you’re eligible to enter. You never see the searches you didn’t show up for, which is what makes this so easy to miss.
Do I really need Shopping campaigns?
In nearly every case, yes. Shopping is usually the backbone of e-commerce advertising and performs best paired with Search rather than run alone. If someone is selling you e-commerce ads without Shopping, ask why.
We already have an agency. How do we know if the account is being managed?
Look at the account’s change history rather than the monthly report. You’re looking for deliberate decisions with reasons behind them — budget shifts, search-term exclusions, feed fixes, new tests — not a particular number of changes. How to tell whether your e-commerce ad account is actually being managed sets out what to check.
Weak results don’t automatically mean weak management, either. Margins, demand and the storefront can hold back a well-run account. When to change e-commerce marketing agencies covers how to tell the difference.
How long before this works?
Faster than SEO, but the first 30 to 60 days go to feed refinement, campaign structuring, and cutting traffic that doesn’t convert. Accounts improve as purchase data accumulates, because that’s what the bidding has to learn from.
Next step
Start with the feed and the arithmetic.
If you’re already running Shopping campaigns, the feed usually explains a good part of the gap between what you’re getting and what the catalogue could do. What the orders are actually worth explains most of the rest.