Ecommerce is the one category where Google Ads can see exactly what it earned you. Every sale has a value attached, so the bidding can optimize toward revenue instead of guessing from form fills. That makes it the easiest category to run well and the easiest to waste money on, because the machine will confidently scale whatever you told it was valuable.
The difference between a good ecommerce account and a bad one is almost never the campaign settings. It is the product feed and the tracking behind it.
The feed matters more than the campaign
Shopping and Performance Max do not read your keywords. They read your product feed. Product titles, attributes and images decide which searches you appear for, which means your feed is your keyword strategy whether you treat it that way or not.
Where feeds usually go wrong:
- Disapproved products sitting in Merchant Center that nobody checks, so part of your catalogue simply never shows.
- Manufacturer titles like a model number, when shoppers search by description.
- Missing GTINs and attributes, which limits where products can appear.
- One flat feed with no split by margin or priority, so your worst products get the same budget as your best.
We rebuilt exactly this on a healthcare ecommerce account. Merchant Center had disapproved products, missing GTINs and generic titles, and conversion tracking counted page views instead of purchases with no value attached at all. Fixing the measurement, then the feed, then the campaign structure took tracked revenue to $2.4M at a ROAS held above 1,000% across a 21 month engagement. The write up is in the Infraheal case study, and our Shopping ads guide covers the feed work.
Track revenue, not orders
This is the single most common ecommerce tracking failure, and it is invisible until someone checks.
If your conversion action counts purchases but does not send a value, Google optimizes toward order count. It will happily find you a hundred $12 orders instead of ten $300 orders, because you told it an order is an order. On the account above, the average value per tracked conversion went from effectively zero to roughly $1,154 once real revenue was passed, because the account finally counted sales rather than page loads.
Check three things:
- Conversion value is passed on every purchase, and it matches what the order was actually worth.
- The purchase conversion fires once per order, not on every thank you page reload.
- Only the purchase is marked as primary. If newsletter signups are primary too, bidding is chasing them.
Our guide to conversion tracking covers the setup and ecommerce PPC strategy covers the wider account.
Know your break even ROAS before you set a target
ROAS targets get picked out of the air more often than they get calculated, and a target that ignores margin is just a number.
Break even ROAS is one divided by your gross margin. At a 40% margin you break even at a ROAS of 2.5, or 250%. Anything below that loses money no matter how good the dashboard looks. At a 25% margin you need 4.0. Work that out per product group, because a 60% margin accessory and a 15% margin appliance cannot share one target.
Our posts on what ROAS is and break even ROAS walk through the math properly.
| Gross margin | Break even ROAS | A healthy target |
|---|---|---|
| 20% | 500% | Above 600% |
| 30% | 333% | Above 420% |
| 40% | 250% | Above 320% |
| 50% | 200% | Above 260% |
What this means for you: if an agency proposes a ROAS target without asking your margin, they are guessing.
Scaling into a peak without losing efficiency
The usual assumption is that scaling spend costs you efficiency. It does not have to.
On a seasonal retailer we took the same peak window year over year, rebuilt Performance Max and Standard Shopping around product priority and margin rather than one flat feed, cleared Merchant Center disapprovals, rewrote titles around shopper search behaviour and added 500 plus generic negatives. Spend rose 110% and conversions 105%, and ROAS still improved from 535.90% to 566.84% rather than dropping. Revenue went from $948K to $2.1M across the same window. The detail is in the Christmas retailer case study.
The lesson is not that scaling is free. It is that most accounts are capped by structure, not by demand, and fixing the structure is what lets budget go up without return going down.
Where Performance Max helps and where it hides things
Performance Max works well in ecommerce because a good feed gives it something concrete to optimize against. It is also the campaign type that shows you the least.
Two things worth doing:
- Do not put your whole catalogue in one asset group. Split by margin or product priority so budget follows the products worth selling.
- Use account level negative keywords to keep it away from traffic you do not want. It is one of the few steering controls you get.
Our Performance Max guide covers what you can and cannot control.
How to choose an ecommerce PPC agency
- “What is our break even ROAS?” If they cannot answer after seeing your margins, they have not done the work.
- “What is wrong with our feed right now?” A good answer names disapprovals and title structure. A weak one talks about campaigns.
- “Are we passing conversion value correctly?” They should want to verify this before touching bidding.
- “How will you split products by margin?” One flat feed is the default and it is rarely right.
- “What would make you tell us to stop?” Anyone with no answer is selling.
Our post on choosing an ecommerce PPC partner covers this in depth, and management pricing covers fee models. Percentage of ad spend deserves a hard look in ecommerce, because it rewards spending rather than returning.
Common questions
Shopping or Performance Max?
Most stores end up running both. Standard Shopping gives you more control and clearer reporting, Performance Max reaches further. Start where you can see what is happening, then add reach.
What ROAS should we target?
Whatever clears your break even with margin to spare. There is no universal good number, and anyone quoting one without knowing your margins is guessing.
Why did ROAS fall when we increased budget?
Usually because the extra budget went into products or audiences that were already marginal. That is a structure problem, not a proof that scaling does not work.
Do we need to fix Merchant Center before advertising?
Yes. Disapproved products cannot show at any budget, and generic titles limit which searches you appear for. The feed is the campaign.
Should we bid on our own brand?
Usually yes if competitors are bidding on it. The clicks are cheap and convert well, and letting a competitor sit above your own name costs more than the clicks do.
How long before it works?
Ecommerce feedback is faster than lead generation because sales happen the same day. Give it thirty days for bidding to settle and ninety to judge the account properly.
Where to go from here
Do one check before anything else. Open Google Ads, look at your conversion value column, and confirm the total roughly matches what your store actually took in the same period. If those two numbers disagree, every decision built on top of them is wrong, and that is where to start.
Our ecommerce advertising page explains how we work, our case studies show two real rebuilds, and you can get in touch for a straight read on your feed and your tracking.
