Google Ads for ecommerce: what actually works in 2026

By Ahmed Imran · Updated June 2026 · 8 min read

In 2026, Google Ads works for US ecommerce when you fix conversion tracking and the product feed before touching campaigns, then scale with segmented Performance Max backed by high intent search. That order took Autobuffy to $1.5M in revenue at 6.89x ROAS in 8 months, and it is the same playbook I run on every store I manage.

What should US ecommerce brands fix before spending more on Google Ads?

Tracking. Before I touch campaigns, I verify that every purchase fires once with the correct order value, using enhanced conversions and a deduplicated tag setup, because Smart Bidding optimizes toward whatever numbers you feed it. When spend is large I stress test the data against an outside source. At SwimOutlet we spent $344K in 30 days across 84 campaigns; the platform showed 1.75x while Northbeam verified 3.39x, and that gap changed every budget decision. Texas Hill Country Olive Co is the reverse case: a near dormant account scaled to $467K in revenue at 5.36x ROAS once tracking and the feed were cleaned up, before any clever campaign work.

Why is the product feed the real campaign in ecommerce Google Ads?

Shopping and Performance Max do not read keywords. They read your Google Merchant Center feed, so the feed is where ecommerce accounts are actually won. The work is unglamorous and it compounds.

• ›Rewrite titles so the product type and the attributes shoppers actually search (size, color, fitment, material) sit in the first 70 characters.
• ›Submit correct GTINs so Google can match products to demand and to price benchmarks.
• ›Clear disapprovals weekly; one policy flag can silently pull a bestseller out of the auction.
• ›Add custom labels for margin, price band, bestseller status, and seasonality so campaigns can be segmented by what drives profit.

Standard Shopping vs Performance Max: which one wins in 2026?

Both, for different jobs, and accounts that scale usually run them together. I start newer accounts on Standard Shopping because query level control builds clean data fast, then move proven segments into Performance Max once purchase history can support value based bidding. Pure Sims went from zero to 4x ROAS in four months on exactly that path: segmented bestseller Shopping first, then a new customer Performance Max that held 4.4x. At Aliava, product segmented Shopping ran at 7.15x while the Performance Max engine ran at 9.91x and produced about half of all revenue.

How do you structure Performance Max so it actually scales?

Dumping the whole catalog into one Performance Max campaign does not work. The algorithm concentrates spend on a handful of proven SKUs while brand traffic pads the reported ROAS, and most of the catalog turns into zombie products with zero impressions. I split campaigns by margin and category using custom labels, shape listing groups so each asset group sells one coherent theme, write search themes that mirror real buyer queries, and apply brand exclusions so the campaign has to win strangers. At Autobuffy this structure carried $1.5M in revenue in 8 months at 6.89x ROAS, almost entirely on segmented Performance Max, with Google Ads scripts flagging zombie and overindexed products every week.

Do you still need Search campaigns when Performance Max exists?

Yes. High intent search is still the most reliable profit engine in ecommerce. I run exact and phrase campaigns on category terms and model level queries with clear purchase intent, plus a separate brand defense campaign that I report separately. That separation matters because brand inflates blended ROAS, and optimizing to a blended number lets weak non brand spend hide behind cheap brand conversions. At Aliava, brand search held a 27% CTR at 8.5x through a full rebrand, and keeping it walled off was the only way to prove the non brand engine was genuinely profitable.

How do you scale past Shopping and Search without wrecking ROAS?

Two levers: value based bidding and Demand Gen. Once purchase values are trustworthy I move campaigns to target ROAS and treat the target as a throttle, lowering tROAS in small steps to buy volume and raising it to defend margin. Demand Gen, which now pulls shoppable product ads straight from the Merchant Center feed across YouTube, Discover, Gmail, and the Display Network, is the layer I add once intent capture is maxed out. The ceiling is higher than most merchants assume. Aura Displays went from zero to $944K at 15.02x in about 7 months, and Dandi Fertility did $18K in two months at 6.62x inside a restricted medical category.

What does the weekly operating rhythm look like?

Automation did not remove the work. It moved the work into supervision, and this is what mine looks like on a typical ecommerce account.

• ›Monday: full search term and Performance Max query review, then a fresh batch of negatives.
• ›Merchant Center check for disapprovals, price mismatches, missing GTINs, and items dropped from the auction.
• ›Listing group spend audit, with scripts surfacing zombie products and overindexed winners.
• ›Budget and tROAS moves in small increments, rarely more than 20% at once.
• ›One structural test per week, such as a new asset group or a custom label split.

What results does this playbook produce across different stores?

Same order of operations, very different stores. All of these are public case studies from my own client work.

If you want this run on your store, I work as an independent operator on a flat monthly fee starting at $1,100, tiered by ad spend and never a percentage of it. The free audit below is the fastest way to see which layer is leaking.