This guide covers 10 specific tactics to improve your Google Ads ROAS: negative keyword cleanup, conversion tracking fixes, campaign segmentation, bid strategy tuning, Shopping feed upgrades, audience layering, ad copy testing, and more. Each one includes what to do and why it works.
1. Clean Up Your Negative Keywords
This is the fastest win on the list. Most accounts we audit are wasting 15-25% of their Search budget on queries that will never convert. We're talking about terms like "free," "how to," "DIY," "jobs," "salary," and "reviews" triggering product ads.
Here's what to do: pull your search terms report for the last 90 days. Sort by cost (descending). Look at every query that spent more than $20 and didn't convert. Add those terms as negative keywords at the campaign or ad group level.
Then build a shared negative keyword list at the account level. Include universal ecommerce negatives: free, coupon code, cheap (if you're a premium brand), wholesale, DIY, tutorial, salary, hiring, jobs. This list prevents the same junk traffic from hitting new campaigns you create later.
Do this weekly for the first month, then biweekly. The wasted spend drops fast.
2. Fix Your Conversion Tracking
Your bid strategy is only as good as the data you feed it. And in roughly 40% of the accounts we look at, conversion tracking is either broken, duplicated, or measuring the wrong things.
The most common problem: double-counting conversions. This happens when you have both the Google Ads tag and the Shopify Google channel (or GA4 import) firing on the same purchase event. Google sees two conversions per sale and thinks your ROAS is twice what it actually is. The algorithm then bids more aggressively, overpaying for traffic.
Check your conversion actions in Google Ads (Goals > Conversions > Summary). You should have exactly one primary "Purchase" action. Everything else (add to cart, begin checkout) should be set to "Secondary" so it's tracked but doesn't influence bidding.
If you're not sure whether your tracking is accurate, run a free Google Ads audit to check.
3. Segment Branded from Non-Branded
If your branded and non-branded traffic lives in the same campaign, your ROAS number is lying to you. Branded traffic converts at 5-10x the rate of non-branded. Blending them hides the fact that your prospecting campaigns might be losing money.
Create separate campaigns for branded search (your store name, product names) and non-branded search (category terms). Give each its own budget and its own ROAS target. Branded should target 8x+. Non-branded should target your break-even ROAS plus a margin, probably somewhere around 2.5x-4x depending on your category.
This doesn't "improve" your total ROAS on paper. But it gives you clarity on what's actually working, which lets you make better decisions about where to spend. For more context on how these numbers compare, check the ecommerce ROAS benchmarks by category.
4. Tune Your Bid Strategy
The right bid strategy depends on your conversion volume, and most accounts get this wrong.
Target ROAS needs at least 50 conversions in the last 30 days to work well. If your campaign has fewer than that, tROAS is guessing, not optimizing. For campaigns under 50 conversions/month, use Maximize Conversion Value without a target. Let Google learn what converts first.
When you do switch to tROAS, set your initial target at 80% of your trailing 30-day ROAS. Not your aspirational target, your actual performance. Then increase it gradually (5-10% per week) as the algorithm adjusts. Setting an aggressive target from day one just restricts volume.
And never change bid strategies during a sale, promotion, or any period with unusual traffic. Wait until performance normalizes.
5. Upgrade Your Shopping Feed
Your Shopping feed directly determines which queries trigger your products and how relevant Google considers them. A poor feed means higher CPCs and worse placements.
Three changes that make the biggest difference:
- Product titles: Front-load with the search term a buyer would use. "Women's Running Shoes Nike Air Zoom Size 10" beats "Air Zoom Pegasus 40." Include brand, product type, key attribute, and size/color.
- Descriptions: Write at least 150 words per product. Include natural variations of your target keywords. Shopify's default descriptions are usually too short.
- GTINs and brand fields: Fill in every GTIN you have. Missing GTINs reduce your eligibility for certain placements and make it harder for Google to match your products.
Stores with well-optimized feeds see 15-30% lower CPCs on identical products. The feed is free to fix and it compounds over time.
6. Layer Audiences on Search Campaigns
Audience layering means adding audience segments to your Search campaigns as "Observation" (not "Targeting"). This doesn't restrict who sees your ads, but it lets you see which audiences convert better and bid up on them.
Add these audiences to every Search campaign:
- All website visitors (last 30, 60, 90 days)
- Past purchasers (customer match list)
- In-market segments related to your products
- Similar audiences to your converters
After two weeks of data, set bid adjustments. Past visitors who didn't convert might warrant a +30% bid adjustment. Past purchasers might warrant +50%. Cold audiences stay at baseline.
7. Test Ad Copy Against Real Metrics
Most ad copy tests measure CTR. That's the wrong metric for ROAS. An ad with lower CTR but higher conversion rate will often produce better ROAS because it pre-qualifies clicks.
Run at least 3 responsive search ads per ad group. Pin your primary value proposition to headline position 1. Test different angles in headlines 2-3: price-focused vs. benefit-focused vs. urgency-focused.
Measure by conversion rate and ROAS, not CTR. After 100+ clicks per ad variant, pause the lowest ROAS version and test a new one. This is a slow process, but it compounds. A 10% improvement in ad-level conversion rate means 10% better ROAS from the same spend.
8. Cut Low-Performing Placements
If you're running Performance Max or Display campaigns, check your placement reports. Certain placements eat budget without converting: mobile game apps, parked domains, and low-quality content sites.
For Performance Max, you can't exclude placements directly (thanks, Google). But you can add account-level placement exclusions through your Google Ads rep or through the Placements tab at the account level. Exclude mobile app placements (adsenseformobileapps.com) and any domain that's spending without converting.
For Display campaigns, review placements monthly. Anything with 500+ impressions and zero conversions should be excluded. This alone can shift Display ROAS by 20-30%.
9. Adjust Device and Schedule Bids
Not all hours and devices convert equally. Pull your device performance report and your hour-of-day report from the last 90 days.
Common patterns we see: mobile CPC is lower but conversion rate is also lower, making mobile ROAS roughly equal to desktop. But some categories (fashion, beauty) convert well on mobile, while others (B2B, high-ticket) skew heavily to desktop.
For time of day, most ecommerce accounts see the strongest ROAS during evening hours (7pm-11pm) and the weakest during early morning (2am-6am). Set bid modifiers accordingly. Decrease bids by 20-30% during your worst hours and increase by 10-15% during your best.
Don't over-optimize here. Small bid adjustments are fine. But cutting mobile bids by 50% when mobile is 60% of your traffic is going to tank your volume. For more granular analysis, consider building a custom dashboard that breaks this down.
10. Tighten Geographic Targeting
Check your geographic performance report. Most accounts have a few states or regions that consistently underperform. Maybe shipping costs to those areas are higher, or demand is just lower.
Sort locations by ROAS. Exclude any location where ROAS has been below break-even for 90+ days with meaningful spend. For the remaining locations, set bid adjustments based on performance. Your top 5 performing states might warrant a +15% bid boost. Your bottom 5 might need -20%.
Also check your location targeting settings. Make sure it's set to "Presence: People in or regularly in your targeted locations" and NOT "Presence or interest." The latter shows your ads to people who are searching about your target area but aren't actually there, which rarely converts for ecommerce.
Frequently Asked Questions
Some changes like negative keyword cleanup and bid adjustments can show results within 1-2 weeks. Structural changes like campaign segmentation or feed optimization usually take 3-4 weeks to fully impact performance as the algorithm relearns.
It depends on the campaign type. Branded campaigns with low ROAS usually have a tracking issue, not a campaign issue. Non-branded campaigns below break-even for 30+ days should be restructured rather than paused, because pausing loses all the learning data.
The fastest wins are usually negative keyword cleanup (stops wasted spend immediately) and fixing conversion tracking (makes sure the algorithm is working with accurate data). These two changes alone can shift ROAS by 20-40% within two weeks.
Not necessarily, but it often does in the short term. Increasing budget by more than 20% at once forces the algorithm to find new traffic, which is usually less qualified. Increase budget in 15-20% increments every 1-2 weeks for the smoothest scaling.
Find Out What's Costing You
COREPPC's free audit checks your Google Ads and Meta accounts in 60 seconds. Get your performance score, spot wasted spend, and see exactly where to improve.
Start Free Audit
