What do you do when a Microsoft Ads account is profitable at 2.4x ROAS but refuses to scale any further? That was the exact ceiling one of our DTC supplement clients hit. Revenue was coming in, conversion tracking was clean, and the campaigns looked stable on the surface, but performance had flattened.
At Positive Sparks, we dug into the account like we always do: search terms, audience layers, feed structure, and margin logic. Over 4 weeks, three technical fixes moved the account from 2.4x to 2.9x ROAS, lifted revenue by 34%, and reduced CPA from $32 to $26.
The problem
Why was growth stalling if the account was already converting? Because profitable does not always mean efficient. In this case, the brand was spending into low-intent queries, using broad Shopping targeting without audience signals, and giving Microsoft’s automation very little margin intelligence from the feed.
Here’s what the account looked like before the changes:
| Metric | Before |
|---|---|
| ROAS | 2.4x |
| Conversion rate | 1.8% |
| CPA | $32 |
| Revenue growth | Flat |
Fix #1: Negative keyword strategy
What was the fastest win? Search term control.
We audited the Microsoft Ads search term reports and found that 40% of spend was going to non-commercial queries. You know the kind: research-heavy searches, informational supplement questions, and terms with weak purchase intent.
So we added 200+ negative keywords at campaign level to tighten relevance across the account.
| Negative keyword fix | Impact |
|---|---|
| Spend on non-commercial queries | 40% of spend |
| Negative keywords added | 200+ |
| Wasted spend reduction | 18% |
Actionable takeaway: Pull search term reports weekly, classify terms by commercial intent, and build shared negative lists around informational modifiers, DIY language, and low-buying-intent variants.
Fix #2: Audience targeting
Could Shopping campaigns perform better without rebuilding everything? Yes, by layering stronger buyer signals onto what was already working.
We applied in-market audiences for health supplements to the existing Shopping campaigns. Bear in mind, this was not a full campaign restructure. We kept the proven product groups in place and added audience intelligence to help Microsoft find people closer to purchase.
That pushed conversion rate from 1.8% to 2.9%.
| Audience layer | Before | After |
|---|---|---|
| Conversion rate | 1.8% | 2.9% |
Actionable takeaway: If you sell in a defined category like supplements, beauty, pet care, or fitness, test in-market audiences on top of Shopping rather than waiting for generic automation to figure it out alone.
Fix #3: Product feed optimization
What gave the bidding algorithm better commercial direction? Feed structure.
We added custom labels for margin tiers and seasonal relevance, which gave the account a much clearer way to separate high-value SKUs from lower-margin products. That meant Microsoft’s automated bidding could stop treating every item as equally important.
Once those labels were in place, the AI shifted 30% of budget toward higher-margin products.
| Feed optimisation change | Impact |
|---|---|
| Custom labels added | Margin tier, seasonal relevance |
| Budget shifted to high-margin products | 30% |
Actionable takeaway: If your feed only includes basic title, price, and availability fields, you’re leaving bidding efficiency on the table. Add custom labels tied to profit, seasonality, and inventory priorities.
The result
So, what changed after one month?
| Metric | Before | After | Change |
|---|---|---|---|
| ROAS | 2.4x | 2.9x | +22% |
| Revenue | — | — | +34% |
| CPA | $32 | $26 | -19% |
| Conversion rate | 1.8% | 2.9% | +1.1 pts |
That’s the bigger lesson for us: Microsoft Ads for ecommerce usually doesn’t need dramatic reinvention. It needs tighter query control, better audience signals, and a product feed built around profit, not just catalog management.
What are you seeing in your own Microsoft Ads account right now? Are you hitting a ROAS ceiling, or wondering whether search terms, audience layering, or feed structure are holding you back? If you want, we can compare notes.
Category: Positive Sparks News
