Ecommerce PPC agency: what a good one actually does

What you are actually buying

In e-commerce, most of the work an agency does is upstream of the ad account. Bidding is largely automated now. Creative and copy matter. But the biggest movements in performance usually come from the product data, the account structure, and knowing which products deserve budget in the first place.

That is why "we will optimize your campaigns" is such a weak pitch. Optimizing campaigns is a Tuesday. The work that changes the numbers is fixing a feed where two thousand products have no product identifiers, or splitting a campaign so that your 8% margin bestseller stops absorbing the budget that should be going to the 60% margin range.

So when you assess an agency, assess what they intend to do in the first ninety days. If none of it touches the feed, the structure or your margins, you are buying account maintenance.

The product feed is the campaign

Shopping ads have no keywords. Google matches searches to your products using the data you send it, which means your feed is doing the job that keywords do in a search campaign. A weak feed is not a technical annoyance. It is an untargeted campaign.

What a competent agency fixes in the first month:

• Titles rebuilt for search, not for your website. The most valuable words go first: brand, product type, then the attributes people search on (size, color, material, model). Google’s title attribute guidance sets the constraints, and titles are the single highest-leverage field in the feed.
• Identifiers filled in. GTINs, brand and MPN. Missing identifiers limit how your products can compete and how they can appear.
• Product types and categories set deliberately. These are what let you carve the catalog up later. Skip this and every structural decision afterwards becomes harder.
• Attributes completed. Size, color, gender, age group, material. Sparse attributes mean you match fewer relevant searches.
• Disapprovals cleared, and monitored. Every store accumulates them. Nobody notices until a bestseller has been offline for three weeks.
• Custom labels for margin and strategy. This is the one that separates real e-commerce specialists from generalists. Labeling products by margin band, season and stock level is what makes profit-based bidding possible at all.

The full technical reference is Google’s product data specification. We covered the practical side in how to optimize Google Shopping ads and the mechanics in how Google Shopping ads work.

Structure: what should be separate, and why

The default approach is to put the whole catalog into one Performance Max campaign and let it sort itself out. Sometimes that works. More often it means your top twenty products consume the budget while the rest of the range never gets shown, and you cannot see enough to know whether that is the right outcome.

Sensible separations, in rough order of usefulness:

1. 1By margin. High-margin and low-margin products should not compete for the same dollars at the same ROAS target. This is the split most accounts are missing.
2. 2By brand versus non-brand. People searching your brand name were already coming. Blending them in flatters every other number in the account.
3. 3By intent stage. Shoppers searching a specific model number are close to buying. Shoppers searching a category are not. Same budget, very different value.
4. 4By stock and seasonality. Advertising a product that ships in six weeks is a good way to buy refunds.
5. 5New customers versus returning. If you can separate them, you can bid properly for growth rather than paying to re-acquire people who already buy from you.

The best account structure is the one a human can understand six months later. If your agency cannot draw the account on a whiteboard in two minutes, neither will the person who inherits it.

ROAS is not profit, and target ROAS is not a strategy

ROAS is the number that makes everyone happy in a meeting and nobody rich. It measures revenue against ad spend, and it ignores cost of goods, shipping, payment fees, returns and the discount that got the order over the line.

The number that matters is your break-even ROAS, which is simply one divided by your gross margin. At a 40% margin, you break even at 2.5. At 25%, you break even at 4.0. Every target above that is profit and every target below it is a subsidy, no matter how good the dashboard looks.

Which is why a single account-wide target ROAS is usually wrong. A catalog with margins ranging from 15% to 65% needs different targets for different products, and that is only possible if the feed has been labeled properly in the first place. The two halves of this article are the same problem.

Ask any agency what your break-even ROAS is. If they do not know your margins, they cannot tell you whether the account is making money. They can only tell you it is busy.

The other half of the equation is the store itself. Baymard puts average cart abandonment near 70%, which means the checkout usually has more headroom than the bidding does. And recovering those carts is an email job, not an ads job. Our guide to e-commerce email flows covers the sequences that pick up the difference.

Questions that reveal whether they know retail

Five questions, asked in a first call, will tell you more than any case study deck.

1. 1"What would you change in our feed first?" They should be able to look at your store and name something specific within minutes. Titles, usually.
2. 2"How would you handle our margin differences?" If the answer does not involve custom labels or separate campaigns, they manage accounts rather than catalogs.
3. 3"How do you separate brand from non-brand?" Anyone who does not do this is reporting numbers inflated by traffic you already had.
4. 4"What happens when something goes out of stock?" A real answer involves feed rules and automation, not somebody remembering.
5. 5"What is our break-even ROAS?" They cannot answer on day one, but they should immediately ask for cost of goods, shipping and return rates. If they never ask, they will never know whether they are helping.

One more thing worth checking: whether they will tell you to spend less. Performance Max in particular can look excellent while quietly harvesting branded and returning-customer demand. Our assessment of whether Performance Max is worth it covers how to test that honestly.

When you do not need an agency yet

Under roughly $10,000 a month in ad spend, an agency retainer is often a large share of your budget going to management rather than to media. At that stage you are usually better served by a one-off setup and a quarterly review, with the day-to-day handled in-house. Most agencies will not say this, which is exactly why it is worth saying.

Other times to hold off:

• Your store converts poorly. If your conversion rate is well under 1% on qualified traffic, more traffic is an expensive way to confirm the problem. Fix the product pages, the shipping costs and the checkout first.
• Your margins do not support paid acquisition. Some catalogs simply cannot carry the cost of clicks. If break-even ROAS is 6 and the category averages 3, paid media is not the growth channel. Email, organic search and retention are.
• Your tracking is broken. If purchase conversions do not reconcile with your store’s own numbers, nobody can optimize anything. Fix measurement, then hire.
• You have one product with no repeat purchase. Possible, but hard. Get the economics working on a small budget yourself before paying someone to scale something that is not yet profitable.

If you are past all of that, our paid media work and our e-commerce Google Ads service start with the feed and the margins, in that order. And if your organic side needs the same attention, what a Shopify SEO agency should do covers the other half.

Frequently asked questions