I open a lot of accounts running Target CPA or Target ROAS where every conversion — every lead, every sale — gets fed into Google Ads as the exact same value. A $50,000 enterprise lead counts the same as a tire-kicker who filled out a form to get a discount code. A $30 impulse purchase counts the same as a $400 order with 45% margin. Smart Bidding doesn't know the difference. It just does exactly what you told it to do: get more conversions at that value, as cheaply as possible. If you've flattened every conversion into one number, you've told the algorithm that volume is the goal. It will deliver volume. It won't deliver profit.
Smart Bidding Optimizes for the Number You Give It — Nothing Else
This is the part most accounts get backwards. People assume Target ROAS is optimizing for actual return on ad spend. It's not. It's optimizing for the ratio between spend and whatever conversion value you passed into the account. If that value is static — the same $1 placeholder on every lead, or average order value applied uniformly across every product — the algorithm is blind to which conversions actually make you money. It will happily shift spend toward cheap, low-value conversions because they're easier to win, and your reported ROAS will look great while your actual margin quietly erodes.
Every lead form fill fires the same conversion value. Smart Bidding chases whichever audience and keywords produce the most form fills, regardless of whether those leads ever become customers.
Leads are scored by CRM stage, deal size, or close rate before the value hits Google Ads. Smart Bidding shifts spend toward the traffic patterns that historically produce the leads that actually close.
Building the Value Model for Lead-Gen
For lead-gen accounts, the fix starts in the CRM, not in Google Ads. You need a way to differentiate a lead worth pursuing from a lead worth ignoring, and to get that differentiation back into the ads platform as an actual dollar value or a weighted score. This is usually a job for offline conversion imports or a CRM integration that pushes conversion value updates once a lead is qualified, moves to a pipeline stage, or closes.
- Score leads on data you already have: job title, company size, lead source page, time-to-respond, or a qualification checklist your sales team fills out
- Assign a real value per stage — MQL might be worth a small placeholder, SQL worth more, closed-won worth the actual deal value
- Push value updates back into Google Ads via offline conversion tracking so bidding recalibrates on real outcomes, not just form fills
- Give the model at least 30-60 days of closed-loop data before trusting it — Smart Bidding needs volume in each value tier to learn from it
This only works if your conversion tracking is actually trustworthy in the first place. If leads are getting double-counted, misattributed, or tracked as generic form-fill events with no source detail, no value model will fix that — you're feeding garbage into a system designed to optimize garbage. That's the exact problem I cover in The Conversion Tracking Audit That Explains Why Your Best Campaigns Might Be a Mirage, and it's worth fixing before you touch value-based bidding at all.
Building the Value Model for Ecommerce
Ecommerce accounts have an easier data source — order value is already sitting in the checkout system — but most still get value-based bidding wrong because they pass revenue instead of margin. A $200 order on a low-margin category and a $200 order on a high-margin category are not the same conversion to your business, even though they look identical to Google Ads. If margin data lives in your product feed or ecommerce platform, you can pass margin-adjusted values instead of raw revenue, so Target ROAS optimizes toward the products that actually make you money rather than whichever SKUs move the most units.
Target ROAS isn't optimizing your return on ad spend — it's optimizing whatever number you handed it. Hand it revenue with no margin attached, and it will happily sell you into a loss.
This is also where Standard Shopping vs. PMAX structure matters — if you're running Standard Shopping vs. PMAX: How to Structure Both Without Them Fighting Each Other, make sure margin-adjusted values are consistent across both campaign types, or you'll get bidding that behaves rationally within each campaign but irrationally across your account as a whole.
- ✓︎Confirm conversion tracking is accurate before layering value data on top of it
- ✓︎Define a scoring or margin logic that reflects real business value, not just conversion count
- ✓︎Push values via offline conversion import or margin-adjusted feed data, not static placeholders
- ✓︎Keep value tiers consistent across every campaign type touching the same conversion action
- ✓︎Give Smart Bidding a full learning cycle before judging results — don't panic and revert after a week
Value-based bidding isn't a setting you flip on. It's an ongoing data pipeline between your CRM or ecommerce platform and Google Ads. Get that pipeline right and Smart Bidding starts working for your P&L instead of against it. Skip it, and you're paying to acquire volume that was never worth acquiring in the first place.