Winning when every price is the same

A campaign for every kind of search

ServicesGoogle Ads · Microsoft Ads · Feed Management · Server-Side Tracking

Google Ads carries the business. We split the account by what the person searching already knows.

• A trademark search campaign with a capped maximum CPC, so we collect every branded click without overpaying for traffic that was coming anyway
• Dedicated campaigns for the biggest brands in the catalog
• Google Shopping in 2 tiers. High intent takes brand, model and long-tail searches from people close to buying. Low intent takes the broad ones, like “RC cars”, on a separate budget
• Performance Max, plus a small Demand Gen campaign

Splitting Google Shopping by intent lets us push impression share on the searches that matter without the generic ones swallowing the budget.

Dynamic Search Ads that know what is in stock

Dynamic Search Ads run 2 ways on this account. One campaign reads Google’s index of the site. The other runs off the product feed, and that version has a trick. When a product sells out it drops out of the feed, and the ads drop with it. When it comes back in stock, the ads come back too. Nobody pays to send a shopper to a page they can’t buy from.

Room for the new arrivals

Performance Max runs on performance buckets. Every product gets a custom label in the feed based on its recent clicks, conversions and return on ad spend, and the labels refresh automatically as the numbers change:

• Top sellers, which convert well and get the most budget
• Steady performers, which sell at a decent return and get room to grow
• Products with clicks but no sales, which get a lower target so they stop draining spend
• Products Google barely shows, which get a separate campaign so they can be seen at all

Each bucket runs in its own campaign with its own budget and return target, so we can back the proven sellers and push the ones Google is overlooking.

New products get a Performance Max campaign of their own. On a catalog this big, Google keeps spending on what already sells and gives a fresh launch almost no chance to prove itself. So when new stock arrives we tag it with a custom label in DataFeedWatch and move it into the launch campaign, where it gets traffic from day one.

Sales, on the day they start

DataFeedWatch picks up changes from the site once a day, so a sale that goes live in the afternoon would normally reach Merchant Center the next morning. We don’t wait. The moment a promotion starts we force the site, the feed and Merchant Center to sync, so the sale price is in the ads the same day. Black Friday gets the same treatment.

During a sitewide sale every campaign turns promotional. We rewrite the ad copy everywhere and swap the creative, including videos a graphic designer built for us for the bigger events.

Microsoft Ads

Microsoft Ads runs the same structure, imported from Google and managed on manual CPC, on a much smaller budget. It adds searchers Google misses.

Server-side tracking that counts new customers

The standard setup was undercounting. The thank-you page did not pass enough customer details, like an email or a phone number, for enhanced conversions to match orders back to clicks. We moved to server-side tracking on Stape, wrote the data layer workaround and the code for it, and handed it to the client’s developer ready to install. A separate pixel checks every buyer against the Stape database, so we can tell a new customer from a returning one and bid for the ones worth finding.

Should the store add a new brand?

The client once asked whether to start selling another popular brand. A well-known name brings search volume with it, so we built a forecast before anyone placed an order. It weighed 4 things.

• Margins on the new range against the ranges already on the site
• Search volume for the brand, and how much of it the store could realistically win
• Warehouse capacity and the extra orders to pick, pack and support
• The risk of shoppers trading down, since the new brand sells at a lower average order value and could pull buyers away from the higher-value products

At best the numbers broke even, with thousands more orders to handle for the same revenue. We recommended against it, and the client agreed.

Meta catalog ads: a holiday test that paid its way

ServicesMeta Ads · Creative

Meta Ads started as a one-off test over the Black Friday and Christmas period, to see what the channel could do for the store. It took little effort to run and it kept selling, so the client asked us to keep it going alongside everything else.

The account runs mainly on Meta catalog ads, on a modest monthly budget. CPMs sit under $10 and an outbound click costs around $0.15, lower than almost anything we see in the US, so the channel reaches a lot of RC fans on a small budget.

Reading Meta fairly

Since launch, Meta’s own reporting puts the catalog campaigns at a ROAS of more than 28 on click attribution alone, with view-through already taken out, and individual months ranging from around 20 to 30. Click attribution can over-credit Meta, because it also counts shoppers who clicked an ad and then came back through Google or another channel before buying. Last-click reporting in GA4 goes the other way and makes Meta look weak next to Google Ads. The truth sits between the two, and we report both so the client can judge Meta for what it adds to the business beyond the last click: people discovering the store, and a brand they recognize when they search.