Blended ROAS hides which products actually make money. Plenty of stores scale a 4x ROAS campaign that is selling their worst-margin SKUs and wonder why profit did not move. Feed quality and margin data fix that.
Usually in the feed and in the reporting, not in the ad copy.
In Shopping and Performance Max, feed quality determines what you show for far more than bids do. Poor titles, missing attributes and stale availability cap performance before anything else matters.
Feeding revenue but not margin means bidding optimises toward turnover. A high-revenue, low-margin SKU looks like a winner right up until you check the bank account.
Run without structure or exclusions, PMax will happily spend most of your budget re-buying customers who were coming back anyway, and report it as new revenue.
Brand search converts beautifully and mostly would have converted anyway. Blended into one number it flatters everything else and hides whether acquisition is actually working.
Feed and measurement first, then scale.
Before anything gets scaled, the data has to be right. GA4 and Google Tag Manager audited and rebuilt where needed, conversion actions deduplicated, Meta CAPI or server-side tagging where it applies, and offline conversion import so the platform optimises toward revenue instead of form fills.
Titles, attributes, GTINs, availability and custom labels structured so Shopping and PMax can match you to the right queries. This is usually the highest-leverage work in an ecommerce account.
Where your platform allows it, margin or product-level profitability fed back as conversion value so bidding chases profit rather than revenue.
Paid traffic dies on a bad page. I build and iterate the landing pages the campaigns point to, so the conversion path is one system instead of two vendors blaming each other. Fixed scope, tied to the campaigns, not open-ended web design.
A Looker Studio dashboard tied to your real conversion data, plus a written read on what changed and why. No 40-tab spreadsheet, no vanity metrics, no 'impressions are up' as a headline.
Brand and non-brand always get reported separately. If a vendor shows you one blended ROAS number, ask to see it split.
Some engagements have verified public numbers. Others are longer-running or confidential roles where I can speak to scope and approach but not exact percentages. I would rather show you the real mix than invent a case study.
Roughly $20K per month in Google Ads spend, in a market with $45 to $80 CPCs.
Ongoing role as Senior Paid Media Manager, leading strategy and a growing paid media team in a heavily regulated industry.
I go through the account, the tracking and the conversion path and tell you what is actually broken. You get the findings whether or not we work together.
Tracking, conversion actions and structure get corrected first. Optimising on top of bad data just makes you confidently wrong.
Campaigns, ad copy, audiences and landing pages built around how your buyers actually search.
Weekly optimisation against the metric that matters to your business, with a monthly written read on what moved and what is next.
Long-form writing on this topic from the blog.
All of them. The platform matters less than whether it can export product-level data and accept proper conversion tracking. Shopify is the easiest, custom builds usually need more tracking work up front.
Usually yes, but with structure: brand exclusions, proper asset groups and product segmentation. PMax run as a single catch-all campaign tends to take credit for demand you already had.
If you can export product-level margin, yes. It is one of the more effective changes available to an ecommerce account and relatively few stores do it.
Ecommerce economics tend to need somewhere north of $5K per month before the data supports confident decisions. Below that, feed and tracking fixes still help, but ongoing management is harder to justify.
A short form beats a long discovery call. Give me the basics and I will come back with a straight read on whether I can help, usually within one business day.
Rather book a time directly? Schedule a call.