How to Scale Google Ads and Microsoft Ads for Ecommerce: The 2026 Playbook

The Google and Microsoft Ads playbook for ecommerce changed significantly over the past two years. Smart Shopping campaigns are gone, migrated to Performance Max. Google Optimize shut down in September 2023. Broad match modifier on Microsoft was deprecated. What replaced them requires a different operating model than the keyword-list approach most ecommerce brands still run.

This guide covers the advanced tactics that move performance in 2026: bidding structure, audience signals, keyword management, Performance Max strategy, attribution setup, testing methodology, automation, and cross-platform execution. For foundational setup, start with our Google Ads ecommerce guide.

What bidding strategies actually move the needle in 2026?

Target ROAS is the primary lever for ecommerce accounts with conversion history. It works, but only when Smart Bidding has enough data. You need at least 30-50 conversions within the past 30 days for the algorithm to make reliable bid adjustments. Below that threshold, Target CPA or Enhanced CPC gives the system a guardrail without requiring it to calculate a ratio it doesn't have enough data for.

• Google Ads: Start with Maximize Conversions during the learning phase (2-4 weeks), then shift to Target ROAS once you have 30+ conversions in 30 days.

• Microsoft Ads: Enhanced CPC is a solid default. Move to Target CPA or Target ROAS once your UET tag has 30+ conversions tracked.

• Portfolio bid strategies: Group campaigns with similar margin profiles under one portfolio strategy. Especially useful when individual campaigns have thin conversion volume on their own.

• tROAS targets: Review quarterly against actual margins. A 400% tROAS target set 18 months ago may be too aggressive if AOV or COGS has shifted.

How do you build audience layers that improve bid efficiency?

First-party data is the foundation of audience strategy in 2026. Third-party cookies are largely gone, and platform-native in-market segments are useful but not sufficient. Customer Match — uploading your email list to Google and Microsoft — unlocks bid multipliers, suppression lists, and similarity modeling that no third-party segment replicates.

• Customer Match: Upload your purchaser list (hashed email and phone). Bid up on existing customers for repeat-purchase categories, and exclude recent purchasers from prospecting campaigns.

• In-market segments: Layer them in observation mode first. Collect 2-4 weeks of performance data, then adjust bids on segments that convert above your account average.

• Microsoft's LinkedIn data: Unique to Microsoft Ads — you can target by job title, industry, and company size. High-value for B2B ecommerce or premium consumer goods where professional demographics correlate with purchase behavior.

• Observation vs. targeting: Always add new audiences in observation mode before restricting targeting. Targeting limits reach; observation tells you what's actually converting first.

What does keyword and search query management look like now?

The SKAG (Single Keyword Ad Group) model is effectively dead. Google's semantic matching means one keyword triggers hundreds of query variants, and structuring campaigns around individual keywords creates maintenance overhead without efficiency gains. What matters now is negative keyword coverage and a broad-match-plus-Smart-Bidding structure that lets the algorithm find buyers you didn't anticipate.

• Broad match + Smart Bidding: Outperforms exact match when you have conversion data. Google's bid algorithm adjusts for query-level intent — exact match just restricts the inventory it can bid on.

• Negative keyword hygiene: Your highest-leverage weekly task. Review the search terms report for irrelevant brand overlaps, competitor terms, and research queries that don't convert. Build shared negative lists across campaigns.

• Microsoft query matching: Often broader than Google. New Microsoft accounts commonly see a higher percentage of irrelevant queries in the first 30-60 days — import your Google negative list on day one.

• Phrase match: Still useful as a middle layer between exact and broad when you want coverage on specific query patterns without fully opening the targeting.

How should ecommerce brands approach Performance Max in 2026?

Smart Shopping is gone — all accounts migrated to Performance Max by 2023. PMax consolidates Shopping, Search, Display, YouTube, Discovery, and Gmail into a single campaign type. That breadth is both a feature and a liability: more reach, but less control over where spend goes.

The 50-conversion threshold is real. Below it, Standard Shopping typically delivers lower CPA ($38.87 vs $43.91 in benchmarks) because it stays in high-intent placements. PMax's advantage kicks in once the algorithm has enough signal to allocate across channels intelligently. For how to structure PMax alongside standard campaigns, see our PPC management playbook for ecommerce.

• Asset groups: Structure each asset group around a distinct product category and audience signal. Dumping everything into one asset group prevents the algorithm from learning which creative-audience pairings convert.

• Product feed quality: Matters more in PMax than it did in Standard Shopping. Clean titles, accurate descriptions, strong GTINs, and high-resolution images directly affect which queries your Shopping placements match.

• Brand exclusions: Campaign-level negative keywords for brand terms are now available in PMax. Add them to prevent PMax from taking credit for branded conversions you'd capture anyway through a separate branded search campaign.

• Audience signals: PMax uses audience signals as suggestions, not hard targeting. Upload your customer list and in-market segments as signals — the algorithm learns from them but isn't constrained to them.

What attribution setup do you actually need for accurate performance data?

Attribution gaps are the most common reason Smart Bidding underperforms. If Google Ads isn't seeing the conversions happening on your site, it can't bid toward them. Enhanced Conversions for Web sends hashed first-party data to improve conversion matching and typically increases reported conversion volume by 10-25% without changing actual performance — it surfaces conversions that were already happening. For full-funnel measurement setup, see our ecommerce marketing strategy guide.

• Enhanced Conversions for Web (Google): Implement via Google Tag or GTM. Sends hashed email and phone at conversion time. Required if you're running Smart Bidding seriously.

• GA4 import: Link GA4 to Google Ads and import key events as conversions. Gives you cross-device attribution that the ads tag alone misses.

• Microsoft UET: Universal Event Tracking tag plus conversion goals configured in Microsoft Ads Manager. Don't launch auto-bidding campaigns until UET has 30+ days of data.

• Attribution window: 7-day click, 1-day view is a solid default for ecommerce. 30-day windows inflate ROAS by attributing conversions from branded searches and repeat purchasers that your ads didn't actually drive.

• Data-driven attribution: Google's default since 2022. Don't switch to last-click unless you have a specific reason — data-driven is more accurate for accounts with sufficient conversion volume.

How do you run meaningful tests without burning budget?

Google Optimize shut down in September 2023. For ad-level and bidding tests, Google Ads Experiments (under Campaigns > Experiments in the UI) is the native replacement. It splits traffic 50/50 between a base campaign and a draft, runs on the same budget, and reports statistical confidence directly in the results view.

• One variable at a time: Test bid strategy, landing page, or creative separately. Testing multiple variables at once makes it impossible to identify what drove the result.

• Run time: Minimum 2-4 weeks per test, preferably 4-6 weeks for ecommerce where conversion cycles can span several days. Calling winners at day 7 is one of the most common optimization errors.

• Statistical significance: Google Ads Experiments shows confidence level in the results view. Wait for 95% before acting. Directional trends are not test results.

• Microsoft Ads testing: No native experiment tool. Run sequential period tests: same campaign, 2-week control period, then 2-week test period. Control for seasonality and external factors.

Where do automation and scripts fit in a well-run account?

Google Ads scripts (JavaScript, runs on a schedule) are useful for anomaly detection, budget pacing checks, and n-gram analysis on search terms. They create a monitoring layer that catches issues before they turn into budget bleed. But scripts are not a substitute for strategy — start with automated rules in the UI before writing any code.

• Automated rules first: Budget adjustments and pausing underperformers can be handled in the Google Ads UI rules interface. Scripts are for automation that the rules UI can't handle.

• Performance Max and scripts: Many Shopping-era scripts that adjusted product-level bids don't apply to PMax. Spend more time on asset group performance analysis and less on bid scripts.

• Microsoft Ads automation: Smaller native script library than Google, but automation rules are comparable. Most Google-side automated rules have a direct Microsoft equivalent.

How do Google Ads and Microsoft Ads compare for ecommerce?

How do you run both platforms together without cannibalization?

Microsoft Ads has a direct import feature that pulls your active Google campaigns in about 15 minutes. It maps campaigns, ad groups, keywords, and ads — you review and launch. Worth re-importing quarterly to keep the accounts in sync. For a head-to-head on where each platform wins, see our Google Ads vs Microsoft Ads comparison.

• Budget split at launch: Allocate 75-80% to Google and 20-25% to Microsoft. Microsoft's lower volume means it needs less budget to generate comparable learning signal.

• Conversion overlap: Microsoft and Google don't deduplicate conversions against each other. A user who clicks both platforms in the same window will appear in both reports. Factor this into blended ROAS calculations.

• Where Microsoft wins on margin: Categories where Google CPCs are inflated by competitive bidding often show stronger CPA on Microsoft. Pull CPA by category on both platforms after 60 days and reallocate toward the more efficient channel.

• YouTube for upper funnel: If you're scaling Google, YouTube ads are the natural expansion for top-of-funnel. The same Google Ads account, audience lists, and conversion data all apply.