Google Shopping ads run on your product feed, not keywords you choose

Unlike traditional search ads, Shopping ads are triggered by Google matching a searcher's query against your product feed data (title, description, category, attributes) — you don't bid on specific keywords directly. This means feed quality is the primary lever for Shopping ad performance, more so than bid strategy. A well-titled, complete, accurate feed with strong images will often outperform a poorly-fed campaign with a much larger budget, because Google can't match you to relevant searches it doesn't understand you're relevant for.

What Performance Max adds

Performance Max is Google's automated campaign type that serves ads across Search, Shopping, Display, YouTube, Gmail, and Maps from a single campaign, using machine-learning bidding to find converting traffic across all of them. For ecommerce sellers, a Performance Max campaign built on a Shopping feed is now Google's primary recommended setup, largely replacing standalone Shopping campaigns for most advertisers.

What you need before starting

A working Google Merchant Center account with an approved, accurate product feed (see Setting Up Google Merchant Center), a Google Ads account linked to it, and — importantly — enough historical conversion data or a large enough budget for Performance Max's automated bidding to optimize against; very low-volume accounts sometimes struggle with Performance Max's automation until they build up conversion history. It's also worth having Google Analytics (or your platform's native analytics) properly configured beforehand so you can independently verify what Google Ads reports.

Campaign structure basics

A Performance Max campaign for ecommerce is typically organized around a Shopping feed plus optional "asset groups" — sets of creative (images, headlines, descriptions, logos, video) that PMax can assemble into ads across its various placements. For catalogs with meaningfully different product categories or margin profiles, many sellers split PMax campaigns by product category or by margin tier, since a single campaign will optimize toward whatever converts most easily, which isn't always your most profitable segment.

Bidding strategies

Performance Max (and modern Shopping campaigns generally) use automated, goal-based bidding rather than manual per-click bids:

  • Maximize conversions — spends the full budget while trying to get as many conversions as possible, with no explicit efficiency target. Best used briefly to gather early data, not as a steady-state strategy.
  • Target ROAS (tROAS) — you set a target return-on-ad-spend ratio, and the algorithm bids to hit that average across the campaign. This is the most common steady-state strategy for profitability-conscious sellers, but requires enough conversion volume for the algorithm to have something to optimize against.
  • Target CPA — you set a target cost per acquisition rather than a ROAS ratio; more common for businesses with a fairly uniform product price/margin, less common for a varied ecommerce catalog.

The trade-off to understand going in

Performance Max's automation reduces manual control significantly — you can't see exactly which placement or audience segment is driving results the way you could in older, more granular campaign types. Sellers who want channel-level attribution clarity sometimes run a smaller manual Shopping campaign alongside Performance Max specifically to preserve some visibility, though this adds management overhead. The search-terms and placement-level reporting available for PMax is deliberately limited compared to older campaign types — treat this as a real cost of the format's convenience, not an oversight to work around.

Feed quality is the highest-leverage input

Accurate, complete product titles and attributes (color, size, material, GTIN/UPC where applicable) matter more to Shopping/PMax performance than almost any other single input — an incomplete or inaccurate feed is the most common reason a Shopping campaign underperforms. Common feed-quality issues to check first when performance disappoints:

  • Missing or incorrect GTIN/UPC/brand fields, which can suppress a product from showing at all in some categories.
  • Product titles that don't include the attributes buyers actually search on (material, size, use case) — Google Shopping titles benefit from being descriptive and keyword-rich in a way a branded website title doesn't need to be.
  • Images that don't meet Google's requirements (background, resolution, watermarks) causing disapprovals.
  • Price or availability mismatches between the feed and the live product page, which can trigger a policy suspension.

Worked example: diagnosing a weak PMax campaign

A seller launches PMax with a $50/day budget on a target ROAS of 400%. After three weeks, spend is inconsistent and ROAS is well below target. A reasonable diagnostic sequence: (1) check Merchant Center for feed disapprovals or warnings first, since a partially-disapproved feed silently shrinks the eligible catalog; (2) check whether the target ROAS is realistic given the account's actual historical performance — an unrealistically high target can cause the algorithm to underspend budget rather than bid aggressively; (3) check asset group quality (images, headlines) since weak creative caps performance even with a clean feed; (4) confirm conversion tracking is actually firing correctly, since a broken or double-counting conversion tag will send the bidding algorithm the wrong signal entirely.

Common mistakes

  • Launching Performance Max with an incomplete or low-quality product feed and blaming the campaign structure instead.
  • Setting an unrealistically aggressive target ROAS immediately, which can cause the algorithm to underspend rather than find volume.
  • Running one PMax campaign across a catalog with wildly different margins, letting the algorithm chase whichever products convert most easily rather than most profitably.
  • Ignoring Merchant Center diagnostics and warnings until a product (or the whole feed) gets suspended.
  • Judging performance too early — automated bidding strategies typically need a real learning period (commonly at least 1-2 weeks of stable settings) before the data reflects steady-state performance.

Best practices

  • Get the product feed genuinely clean and complete before worrying about campaign structure or bid strategy.
  • Segment PMax campaigns by margin tier or product category when the catalog is diverse enough that a single blended target doesn't make sense.
  • Check Merchant Center diagnostics weekly, not just when performance visibly drops.
  • Give a new bid strategy or major campaign change a real learning period before judging or adjusting it further.
  • Independently verify conversion tracking against your platform's own order data periodically.

Troubleshooting

  • Products aren't showing in Shopping ads at all — check Merchant Center for disapprovals first; missing GTIN/brand fields and image policy violations are the most common causes.
  • Spend is well below budget — often an overly aggressive (too low) target CPA or too-high target ROAS relative to the account's actual conversion rate; the algorithm won't spend past what it believes it can profitably bid.
  • ROAS looks great in Google Ads but doesn't match your actual profitability — verify what "conversion value" is being tracked (revenue vs. profit) and whether attribution windows/settings match how you evaluate other channels.

FAQs

Do I still need standalone Shopping campaigns, or just Performance Max? For most ecommerce advertisers, Google now steers strongly toward Performance Max, and standalone Shopping campaigns have reduced functionality or availability in many accounts. Some advertisers who need search-term-level visibility still run a smaller manual campaign alongside PMax, but PMax is the default recommended structure.

How much budget do I need before Performance Max works well? There's no fixed number — it depends more on getting enough conversions per week for the bidding algorithm to have a usable signal than on total dollar spend. A very low-conversion-volume account often struggles regardless of budget size until it accumulates more data.

Can I exclude specific products from a Performance Max campaign? Yes, generally through product feed labels or exclusions at the asset-group or campaign level — useful for excluding low-margin or discontinued items from a campaign optimizing toward broad conversion volume.