By Nick Quirk
September 29, 2026
8 min read
A home goods retailer came to us spending $42,000 a month on Google Ads at a 1.9x return. Seven months later the same store was spending $68,000 a month at 5.7x, which is the 300% ROAS improvement this e-commerce PPC case study walks through. Almost none of that gain came from clever ad copy, and about half of it came from work that happened outside the ads platform entirely.
The numbers below are real, rounded slightly for client privacy, and presented in the order we actually tackled them. If you are auditing your own account, that order matters more than any single tactic.
The Starting Point: $79,800 Revenue on $42,000 Spend
The store sold mid-priced furniture and decor with an average order value of $164 and a blended gross margin near 44%. Paid search and shopping were generating about $79,800 in attributed revenue monthly, which sounds acceptable until you subtract cost of goods, shipping and the ad spend itself.
At 1.9x, the account was losing money on roughly a third of its catalog and hiding it inside a blended average. That is the quiet problem with reporting revenue instead of contribution margin.
Three structural issues showed up in the first audit:
- Duplicate conversion actions counting purchases twice on about 8% of sessions, inflating reported performance.
- One catch-all Performance Max campaign holding 91% of the budget and every SKU, from $29 candles to $1,200 sofas.
- A Merchant Center feed with 312 disapproved items and missing GTINs on 1,100 more, so a quarter of the catalog was effectively invisible.
Month One: Fixing Measurement Before Touching Bids
We spent the first three weeks doing nothing to bids. Every smart bidding algorithm optimizes toward the conversion data it receives, so feeding it inflated or duplicated purchases guarantees bad decisions at scale.
We removed the duplicate purchase tag, moved to server-side tracking for checkout events, and switched the primary conversion from purchase count to purchase value with a margin-adjusted value rule. High-margin categories reported at full value; low-margin clearance items reported at roughly 60% of revenue.
We also moved reporting off last-click. Shopping ads were getting credit for sales that display and paid social had introduced days earlier, and the reverse was happening too. Reading through the differences between first click, last click and data-driven attribution models changed which campaigns we were willing to defend in the next budget meeting.
Reported ROAS actually dropped to 1.6x that month. Real revenue did not change. We were simply measuring it honestly for the first time.
Months Two and Three: Feed Work Beat Bid Work
Product feed optimization produced the single largest lift in this case, and it is the least glamorous part of the job. We rewrote title structures to lead with brand, product type, material and size, added the missing identifiers, and split the catalog into five custom labels based on gross margin band.
Impression share on shopping inventory rose 61% within 30 days with no bid increase, because previously suppressed products finally entered auctions. Click-through rate on the rewritten titles improved from 0.84% to 1.31%.
Custom labels then let us build a structure that respected profitability:
- Label 1 (margin above 50%): aggressive target ROAS of 3.5, allowed to spend freely.
- Label 2 (35% to 50% margin): target ROAS of 5.0, the volume engine of the account.
- Label 3 (under 35% margin): target ROAS of 8.0, running mainly to defend branded queries.
- Label 4 (bestsellers, any margin): separate campaign with its own budget so hero products were never starved.
- Label 5 (clearance): capped at $2,000 monthly, treated as inventory clearing rather than growth.
Month Four: Splitting the Catch-All Campaign
Breaking one Performance Max campaign into four asset-group-specific campaigns by margin band exposed something the blended average had hidden. Two categories were running at 0.7x and 1.1x while the top category quietly delivered 9.3x.
We cover this topic in more depth in The Power of Case Studies in B2B Marketing and SEO.
We cut the two losers to a $150 daily ceiling instead of killing them outright, since both had strong new-customer rates and reasonable 90-day repeat purchase behavior. Growth decisions get easier when you separate a first-order loss from a lifetime-value win.
Branded search was pulled into its own campaign with a small budget, which stopped shopping campaigns from absorbing brand credit. That change alone reallocated roughly $6,800 a month toward non-brand acquisition.
Months Five and Six: Audience Layering and Retargeting
Cart abandonment on the site was sitting near 71%, roughly in line with the average across dozens of studies compiled by the Baymard Institute. That is a lot of qualified traffic leaving with intent already demonstrated.
We built a three-tier remarketing structure: viewed product in 7 days, added to cart in 14 days, and past purchasers in 180 days segmented by category. Applying dynamic retargeting principles for e-commerce stores to the cart tier produced a 6.9x ROAS at a $4,100 monthly spend.
Customer match lists of past buyers were uploaded as exclusions on prospecting campaigns and as value-based seeds elsewhere. Prospecting cost per acquisition fell 18% once the algorithm stopped optimizing toward people who already owned the product.
Month Seven: Post-Click Changes Worth 31%
Traffic quality had improved, so the remaining bottleneck was the site. Three changes moved conversion rate from 1.82% to 2.39%, a 31% relative lift that flowed straight into the ROAS calculation.
- Collection page load time cut from 4.1 seconds to 1.9 seconds by deferring third-party scripts and compressing hero imagery.
- Shipping cost shown on the product page rather than at the final checkout step, which reduced checkout-stage abandonment by 14%.
- Review counts and stock status added above the fold on the 40 highest-traffic product pages.
None of that is PPC work in the narrow sense, but paid traffic pays for every wasted session. A 10% conversion rate gain and a 10% cost per click reduction compound into roughly a 22% ROAS improvement.
What Did Not Work
Two experiments failed and are worth naming. A broad match test on non-brand terms with target ROAS bidding burned $5,300 over three weeks at 1.2x, mostly on informational queries that never converted.
An Amazon Ads expansion also underperformed initially, because the same creative and pricing logic does not transfer between marketplaces and owned storefronts. It became profitable in month nine, well after this reporting window closed.
Dayparting restrictions, which we expected to help, made almost no difference. Conversion rates across the day varied by less than 0.2 percentage points for this catalog.
The Seven-Month Numbers
- Monthly ad spend: $42,000 to $68,000 (up 62%).
- Monthly attributed revenue: $79,800 to $387,600.
- ROAS: 1.9x to 5.7x, three times the starting return, reported as a 300% gain.
- Average order value: $164 to $181, helped by bundle promotion on high-margin SKUs.
- New customer share of revenue: 58% to 64%.
Scaling spend while improving efficiency is unusual, and the sequence is why it worked here. Measurement first, feed and structure second, audiences third, site experience fourth.
How to Apply This Strategy to Your Own Account
Start by exporting 90 days of product-level performance and joining it to your margin data in a spreadsheet. Most stores discover that 15% to 20% of SKUs consume a third of the budget at a negative contribution margin.
Then set a daily review habit around the small set of PPC metrics that actually predict revenue rather than opening 40 columns each morning. The same discipline applies outside retail, which is clear in our breakdown of PPC strategies for plumbers, electricians and roofers, where lead quality plays the role that margin plays here.
Context also matters for forecasting. E-commerce accounts for roughly 16% of total US retail sales according to quarterly Census Bureau data, so category growth alone will not carry an inefficient account.
Frequently Asked Questions
How Long Does It Take to Triple ROAS on an E-commerce Account?
Expect five to nine months for a meaningful 300% improvement on an account spending $25,000 or more monthly. Faster results usually mean the account had obvious tracking errors, while accounts already above 4x rarely triple without a pricing or product change.
What Counts as a Good ROAS for an Online Store?
Break-even ROAS equals 1 divided by your gross margin, so a 40% margin store needs 2.5x just to cover costs. Most healthy e-commerce accounts target 4x to 6x on blended paid spend, though subscription models can profitably run near 1.5x on first order.
Does a Higher ROAS Always Mean Higher Profit?
No, and this is the most common trap in scaling paid media. Cutting spend to the top 5% of keywords will push ROAS above 10x while total profit falls, because volume disappears faster than efficiency improves.
Should You Fix Tracking Before Changing Bids?
Always, since smart bidding needs roughly 30 conversions in 30 days of accurate data per campaign to perform. Changing targets on top of duplicated or missing conversions produces optimization decisions that are confidently wrong.
Which Change Drove the Largest Share of the Gain?
Merchant Center feed cleanup and margin-based campaign structure accounted for roughly 45% of the total improvement in this case study. Retargeting contributed about 20%, site speed and post-click work about 20%, and attribution corrections the remainder.
Get a Real Read on Your E-Commerce Ad Account
If your reported return looks fine but your bank balance disagrees, the gap is usually sitting in your feed, your conversion values or your campaign structure. SEO Quirk will walk your account with you, and you can also read why working with a local agency tends to shorten that diagnosis.