Reference use: Ad spend, clicks, CPA, or ROAS are not turning into qualified revenue. Budget keeps moving while the account, page, offer, or tracking leak stays hidden. Keep this as an authority reference, then use the decision view to decide the next check.
Concept signal
Business problem
What to review before changing the plan
Next step
Symptom match
Ad spend, clicks, CPA, or ROAS are not turning into qualified revenue.
Compare the concept to the visible business symptom before changing the channel, page, or budget.
What are the Shopify Google Ads best practices for 2026?
Answer
Start with the parts Google and Shopify need to process correctly: feed data, conversion data, campaign controls, and landing page match. A pretty PMax summary is not enough. Check these first:
Feed clean? Product title, GTIN, brand, variant, category, image, price, and availability match the site.
PMax controlled? Brand exclusions, final URL rules, asset groups, search themes, and Merchant Center links are deliberate.
ROAS honest? Branded demand and non-brand acquisition are claimed separately.
Tracking right? Purchase-complete conversion action passes order value. Add-to-cart and initiate-checkout are not primary revenue actions.
Page matched? The ad promise lands on the right product, collection, offer, or explanation page.
Split
Report branded and non-brand contribution separately
Account evidence
Reconcile
Compare platform-attributed revenue with store orders and margin
Store data
Feed
Check product attributes against Google's current specification
Google Merchant Center
Verify
Confirm the primary conversion event is a completed purchase
GA4 and Shopify reconciliation
Branded share of PMax spend, five Shopify audits. Range: 28-62%. Source: Stan Consulting audits, 2026.
§01The branded-attribution bias
Before anything else.
Illustrative account: the dashboard shows strong ROAS while store revenue stays flat. The query and category report then shows how much branded demand PMax claimed.
Option A
Exclude brand from PMax
Account-level negative keywords plus a brand-exclusion list applied to the campaign. PMax now numbers non-brand only.
Best for: single brand defense, simple catalog.
Option B
Brand on its own campaign
Separate Standard Shopping or Search for branded terms. Each campaign judged on its own non-overlapping work.
Best for: brands competing on the SERP.
Watch the clock
Month four is too late
Most discover this in month four. By then the wasted non-brand budget is roughly a quarter of cash. Run the split inside month one.
Cost of delay: ~25% of paid budget per quarter.
Branded clicks would have converted at 30-60% on their own. PMax counts them because they fall inside its attribution window. Claimed ROAS rolls in the free revenue. The bank account only sees what new revenue was generated.
§02Campaign structure
The five rules that survive defaults.
Same algorithm, different shape. The structure most accounts run produces a nice summary. The structure below produces honest numbers.
The rules.
Fix attribution before you tune a single campaign.
Exclude brand from PMax or judge it on inflated numbers.
Track purchases only · not add-to-cart, not initiate-checkout.
Segment margin once catalog exceeds 5,000 SKUs.
Defend brand cheaply on Standard Shopping or Search.
Old way
One PMax catch-all. Branded included. Judged on blended ROAS. The wizard produces it; the operator doesn't know there's another option.
New way
Brand defense + PMax non-brand + margin tiers. Three campaigns minimum. Claimed numbers track the bank account.
What's being claimed?
Claimed ROAS vs actual contribution.
§03Feed work + tracking
What actually compounds.
Title, image, and structured-attribute work compounds. Description and bullet-list polishing rarely does. PMax checks structured signals far more aggressively than prose.
Checks first
Title
Checks first
Image
Checks first
Category
Checks first
GTIN
Checks last
Description
Checks last
Bullet copy
Claimed ROAS climbs while actual non-brand ROAS falls. The gap is the branded-attribution claim.
"The ROAS summary shows what the campaign claimed. The bank account shows what the campaign actually generated. Two numbers, not the same story."
· Stan Tscherenkow, Principal · Stan Consulting audits, 2026
"The claimed ROAS is what the campaign claimed. The bank account is what the campaign delivered."
§04What audits find
Six findings to check.
Calculate each finding from the account's own query, conversion, feed, catalog, and revenue data.
01check
Branded-search attribution claimed by PMax. Report the branded share and cost directly.
02check
Conversion tracking firing on the wrong event. Confirm completed purchase is the primary revenue action.
03check
Feed approvals not addressed. Review Merchant Center issues and affected inventory.
04check
Product titles missing useful attributes. Compare the feed with Google's current product-data specification.
05check
One PMax campaign managing the entire catalog. Review margin, demand, and creative differences before deciding whether to segment.
06check
No documented brand strategy. Decide from auction competition, organic coverage, and incremental revenue evidence.
Why most accounts hide the leak.
Claimed ROAS is the metric. The dashboard says the campaign is profitable. The board is happy. The leak stays.
Agencies optimize the ads, not the structure. The handoff between paid media and conversion architecture is somebody else's problem.
Google's wizard ships the catch-all. One campaign, branded included, judged on blended numbers. Different shapes require deliberate builds.
Branded share looks small until you measure it. 30-60% sounds like a stretch until you pull the Search categories report.
§05Where Stan Consulting fits
If you ran this and it's still bleeding.
The Conversion Marketing Plan identifies which structural layer is costing the budget. Written marketing plan, no retainer.
Symptom path
Use the Shopify traffic-without-sales page when clicks, sessions, or carts fail to become purchases.
Why does my Shopify Google Ads ROAS look great but my bank account doesn't agree?
PMax includes branded search by default. Branded clicks would have converted at 30-60% anyway. PMax claims credit because conversions happened in its attribution window. Exclude brand, run a separate brand campaign, judge each on its own non-overlapping work.
What campaign structure should I run?
One PMax for product-discovery with brand excluded. One Standard Shopping or Search for branded protection. Separate PMax budgets per margin tier if catalog has wide margin variance.
Which feed optimizations actually compound?
Title, image, and structured-attribute optimizations compound. Description and bullet-list optimizations rarely do. PMax checks structured attributes (product_type, google_product_category, gtin, brand, condition, age_group, gender, color, size, material) far more aggressively than description prose.
Should I bid on my own brand name?
If a competitor bids on it, yes. If no competitor bids, the answer depends on keeping PMax from claiming branded traffic. Bid on brand in a separate campaign + exclude brand from PMax = cheap defensive coverage without attribution bias.
What conversion tracking should I have first?
A Google Ads conversion action firing on purchase complete (not add-to-cart, not initiate-checkout) with order value passed dynamically. Use Enhanced Conversions or server-side tracking. If tracking fires on initiate-checkout, ROAS overstates by 30-60% vs actual revenue.