Google Shopping feed optimization is the single highest-leverage fix most ecommerce brands overlook when their campaigns stall. This is the story of a mid-market DTC brand spending $80K to $120K per month on Google Ads whose Shopping campaigns looked healthy on paper but had completely flatlined on new customer acquisition. The root cause was not bidding, not budget, and not competition. It was feed quality and campaign structure silently funneling spend toward the wrong products and the wrong audiences. Over 45 days, a full Shopping rebuild, starting with the feed and working outward to structure and bidding, doubled new customer volume without increasing ad spend. What follows is how the problems were identified, what specifically got fixed, and what any ecommerce team can take from this Google Shopping campaign case study and apply to their own account.

The Situation: A Growing Ecommerce Brand On A Flat Revenue Trajectory

Mid-Market DTC With Real Budget And Real Complexity

The brand in question sells consumer goods direct-to-consumer across roughly 1,200 active SKUs. Monthly Google Ads spend fluctuated between $80K and $120K depending on seasonality, with Shopping campaigns consuming roughly 65% of total spend. The remaining budget went to branded Search, non-branded Search, and a small Performance Max experiment that had been paused.

The team was lean: one in-house performance marketer managing Google Ads alongside Meta, a founder who reviewed weekly reports, and a feed management tool handling product data exports to Google Merchant Center. They had been running this setup for over two years.

Why Everyone Thought The Account Was Performing

On the surface, the numbers looked strong. Shopping ROAS hovered around 6x. The weekly reports showed steady revenue from Shopping, and the in-house marketer had been progressively tightening the tROAS target over the previous six months to keep the efficiency number climbing. The founder saw a healthy ROAS and assumed Google Ads was doing its job.

But underneath that headline ROAS, two things were quietly going wrong. First, new customer acquisition had been flat for three consecutive quarters. Second, the revenue coming from Shopping was increasingly concentrated in a handful of low-margin SKUs that repeat buyers already knew by name. The account was generating revenue, but it was not generating growth.

The Problem: Shopping Feed Quality Was Masking Real Inefficiency

Title And Attribute Mismatches Were Suppressing Best-Margin SKUs

The first diagnostic step was pulling a search terms report for Shopping and cross-referencing it against the product catalog by margin tier. The finding was immediate: the brand’s highest-margin product lines were barely showing up in auction. They were getting a fraction of the impressions their search volume warranted.

The reason traced back to the feed. Product titles had been auto-generated from the brand’s ecommerce platform and never rewritten for how people actually search on Google. A product the brand internally called “Performance Blend - Midnight” would show up in Google Shopping with exactly that title, while searchers were typing “dark roast coffee beans” or “strong ground coffee.” Google’s matching algorithm could not bridge the gap because the feed gave it almost nothing to work with. Critical attributes like product type, custom labels, and detailed descriptions were either missing or filled with boilerplate.

Meanwhile, the brand’s commodity SKUs, things like accessories and low-margin bundles, had titles that happened to align well with high-volume search terms. So those products ate the budget.

Product Group Structure Was Compounding The Problem

The Shopping campaign used a single campaign with product groups split only by brand and category, the default Merchant Center taxonomy. There was no segmentation by margin, no separation of hero SKUs from long-tail items, and no distinction between products that drive new customers versus products that serve existing ones.

This meant Google’s bidding algorithm treated every product group with equal priority. A $12 accessory with a 15% margin competed for the same budget as an $85 hero product with a 55% margin. The accessory converted more frequently (lower price point, lower consideration), so the algorithm rewarded it with more spend. The hero product, which was the actual growth driver, was starved.

The tROAS Target Was Filtering Out New Customer Queries

The third layer of the problem was the tROAS setting itself. Over six months, the in-house marketer had ratcheted the account-wide tROAS from 4x to 6x. At 6x, Google’s Smart Bidding algorithm was heavily favoring queries from people already familiar with the brand, queries with high purchase intent and high predicted conversion rates.

New customer queries, which tend to be more generic (“best coffee subscription,” “premium ground coffee online”), convert at lower rates on the first click. A 6x tROAS effectively told Smart Bidding to avoid those auctions entirely. The account was hitting a ceiling on new customer growth because the bidding strategy was optimized to exclude the very queries that bring in new customers.

This is a pattern that shows up constantly in ecommerce Google Ads accounts: a high headline ROAS masking a shrinking addressable audience. The account looks efficient. It is actually contracting.

The Fix: A Three-Part Shopping Rebuild Over 45 Days

The rebuild addressed feed, structure, and bidding in sequence. Each phase depended on the one before it. This is why Google Shopping performance is a data and structure problem first and a bidding problem last.

Part 1: Feed Surgery, Title Rewrites, And Custom Label Segmentation By Margin Tier

The first two weeks focused entirely on the product feed. Every product title was rewritten to lead with the search terms real customers use, pulled directly from Search terms reports, Google Trends data, and competitor Shopping listings. Titles followed a consistent format: product type, key attribute, brand, and variant. “Performance Blend - Midnight” became “Dark Roast Ground Coffee, Strong Blend, [Brand Name], 12oz Bag.”

Beyond titles, product descriptions were rewritten to include relevant long-tail keywords. Missing attributes (color, material, size, product type, GTIN where applicable) were filled in across every SKU.

The most structural change was implementing custom labels based on margin tier. Every SKU was tagged as Tier 1 (high margin, high growth potential), Tier 2 (mid-margin, steady performers), or Tier 3 (low margin, commodity). This tagging did nothing on its own. Its purpose was to enable the campaign restructure in Part 2.

Part 2: Campaign Restructure, Separating Hero SKUs From Long-Tail Categories

With the margin tiers labeled in the feed, the single Shopping campaign was broken into three:

  • A dedicated campaign for Tier 1 hero SKUs, the products that drive new customer acquisition and carry the highest margin. This campaign received the largest share of budget and the most aggressive bidding targets.
  • A second campaign for Tier 2 products, steady mid-margin items that support lifetime value but are not the primary acquisition driver.
  • A third campaign for Tier 3 commodity and accessory SKUs, set to a lower priority with a tighter efficiency target.

This separation meant budget allocation could be controlled at the campaign level rather than left to Google’s algorithm to figure out across a single mixed product group. Hero SKUs could no longer be starved by commodity items that happened to convert more cheaply.

Part 3: Bidding Reset, tROAS Adjusted By Product Group, Not Account-Wide

The final phase addressed the bidding strategy. Instead of a single 6x tROAS applied account-wide, each campaign received its own target calibrated to its role:

  • Tier 1 hero campaign: tROAS dropped to 3.5x. This sounds aggressive, but the margin on these products justified it, and the goal was acquisition, not efficiency on paper.
  • Tier 2 campaign: tROAS set at 5x, reflecting a balanced goal of volume and efficiency.
  • Tier 3 campaign: tROAS held at 7x, extracting maximum efficiency from commodity SKUs without letting them consume meaningful budget.

The learning phase after these changes was managed carefully. Bid changes were staged rather than applied simultaneously, and conversion volume was monitored daily for the first two weeks to ensure Smart Bidding had enough signal to stabilize.

The Results: What Happened After The Rebuild

New Customer Acquisition Volume Doubled

Within 45 days of the full rebuild going live, the brand’s new customer acquisition from Shopping more than doubled. The hero product campaigns were now showing up in generic, high-intent queries that the account had been invisible in for months. The feed improvements alone drove a measurable increase in impression share for Tier 1 products before bidding changes even took effect.

Revenue Grew Without A Budget Increase

Total Shopping revenue grew meaningfully on the same monthly spend. The shift was not about spending more. It was about spending in the right places. Budget that had been flowing to low-margin accessories was redirected to products with higher average order values and better unit economics. Revenue per dollar of ad spend, measured on a blended basis, actually improved once the higher-margin products took their rightful share of impressions.

ROAS Settled Lower But Drove More Profitable Growth

The blended Shopping ROAS did drop from the original 6x to approximately 4.5x. On a dashboard, this looks like a step backward. In the P&L, it was a significant step forward. The brand was now acquiring new customers at scale on products with strong margins rather than recycling existing customers on low-margin SKUs. Revenue growth, customer growth, and gross profit all moved in the right direction.

This is one of the most counterintuitive lessons in ecommerce Google Ads: a lower headline ROAS frequently indicates a healthier, more profitable account. The relationship between tROAS settings and actual revenue growth is not linear, and treating it as such is one of the most common ways ecommerce accounts stall out.

How groas Prevents This Problem From Developing In The First Place

The issues in this case study, feed quality degradation, structural drift, and over-tightened bidding, are not unusual. They are the norm for ecommerce accounts managed by a single in-house marketer or a traditional agency where one media buyer juggles multiple clients.

This is exactly the kind of problem groas exists to solve. For DFY (Done For You) clients, a dedicated groas strategist owns the entire Shopping operation: feed, structure, bidding, and landing pages as one integrated system. The proprietary engine trained on over $500 billion in profitable ad spend runs execution around the clock, identifying feed gaps, impression share losses, and structural inefficiencies far faster than any human could catch them in a weekly check-in.

Feed quality is not treated as a separate workstream from campaign structure. Landing pages are not treated as someone else’s problem. groas works on everything from the first click to the final conversion because that is the only way to prevent the silent compounding of small inefficiencies into flat growth.

For in-house teams who want to stay in control, groas DWY (Done With You) puts the engine underneath your team’s execution while a senior strategist works alongside you, flagging exactly these kinds of structural issues before they cost you a quarter of new customer growth. The engine surfaces what needs to change. Your team decides when and how.

For agencies managing ecommerce client accounts, the groas DIY product gives your media buyers access to the same engine, so feed audits, structural recommendations, and bidding calibration happen at a speed and depth that a single person checking an account can never match.

No onboarding fees. Month-to-month commitment. groas earns the next month by performing.

What In-House Teams And Agencies Can Apply Today

The Three-Question Shopping Diagnostic Any Team Can Run This Week

Before hiring anyone or rebuilding anything, answer three questions about your Shopping campaigns:

  1. Pull your search terms report for Shopping and cross-reference it against your product catalog sorted by margin. Are your highest-margin products getting proportional impression share, or are low-margin SKUs eating the budget?

  2. Look at your campaign and product group structure. Is there any segmentation by margin tier, or is everything lumped together in a way that forces Google to allocate budget blindly?

  3. Check your tROAS target against your new versus returning customer split. Has the target been ratcheted up over time? If so, pull new customer acquisition data for the same period. If ROAS went up and new customers went flat or down, your bidding strategy is filtering out growth.

If you find problems in all three areas, you are looking at the same pattern described in this case study. The fix is not a single bid adjustment. It is a structural rebuild that touches feed, campaign architecture, and bidding as one connected system.

When Shopping Complexity Warrants Fully Managed Support

The three-question diagnostic is free and any competent in-house marketer can run it. But executing the fix, especially across hundreds or thousands of SKUs with margin-tiered feed tagging, multi-campaign restructures, and staged bidding resets, is a different matter. It requires sustained, detailed execution over weeks, not a single afternoon of changes.

If your team has the bandwidth and expertise, the playbook above gives you the roadmap. If your team is stretched, if your agency is capped at what one person can get through in a week, or if you simply want someone to own this end-to-end, that is exactly what groas is built for.

DFY clients can apply here and groas will figure out the right plan on the call. DWY teams can get started with the engine and a strategist alongside. Agencies can start a 7-day free trial and put the engine underneath their client accounts immediately.

Shopping performance is a feed and structure problem first. Solve the data, then solve the bidding, and the growth follows.