Smart Bidding Strategies

Smart Bidding is Google's machine-learning auction-time bidding system. It decides each bid based on user signals (location, device, time, query, history, audience, browser, etc.) that no human can manually evaluate. Used correctly, Smart Bidding meaningfully outperforms manual CPC. Used badly, it amplifies bad inputs — converting bad conversion tracking, mis-set targets, or fragmented account structure into worse performance than manual would have produced. This module covers what each strategy does under the hood, when to use which, the learning phase, seasonality adjustments, and the choices that compound over time.

1. What Smart Bidding actually does

For every Google Ads auction, Smart Bidding considers signals that are not available to manual bidders:

• Device — mobile, desktop, tablet, with model-specific learnings
• Location — geo down to neighborhood-level for some categories
• Day and time — hour, day of week, with seasonality memory
• Search query — specific intent reading, including close-variant interpretation
• Remarketing list — member of your audience or not
• Browser and OS — combined with category-specific conversion patterns
• Language — UI language preferences
• Ad characteristics — which RSA combination is showing
• Demographics — age, gender, household income tier
• Weather (in some categories) — for weather-sensitive products
• Operating system patterns — iOS users with the same query convert at different rates than Android

Smart Bidding combines these signals with your conversion history to predict the probability and value of conversion for each auction, then bids accordingly. The signals available to manual bidders (campaign, ad group, keyword, time of day, geo, audience) are a tiny subset.

Almost any question can be answered, cheaply, quickly and finally, by a test campaign. And that is the way to answer them — not by arguments around a table.

— Claude Hopkins, Scientific Advertising (1923)

2. The six Smart Bidding strategies

3. Target CPA vs Target ROAS — the value-based decision

Target CPA

You tell Google: "Acquire conversions at an average cost of $X." Google bids each auction so that the predicted CPA, averaged over time, lands at your target. Each conversion is treated as equal in value.

• Use when: Every conversion is roughly equal in value (lead gen, free trial signups, app installs).
• Volume requirement: 30+ conversions per campaign in the trailing 30 days for stable performance (Google's minimum is 30; in practice 50+ is better).
• Setting the target: Calculate your maximum allowable CPA from your unit economics (LTV × allowable acquisition margin %). Start with the trailing 30-day actual CPA; adjust toward goal in 10-15% steps.

Target ROAS

You tell Google: "Get me $X of conversion value for every $1 of spend." Google bids each auction based on predicted conversion value. Higher-value conversions get higher bids; lower-value get lower bids.

• Use when: Conversion values vary meaningfully (e-commerce with different product prices, lead gen with modeled value per lead source).
• Volume requirement: 50+ conversions per campaign in trailing 30 days (Google's minimum is 50; in practice 100+ is much better for stability).
• Setting the target: Calculate break-even ROAS from gross margin (1 / gross margin %). Set target ROAS at or slightly above blended-target ROAS. Start conservative; lower target ROAS to scale spend.

4. Maximize Conversions vs Maximize Conversion Value

The "Maximize" strategies don't have explicit targets. Google spends the daily budget while optimizing for either conversion count or conversion value.

Maximize Conversions

• Use when: New campaigns without enough history for tCPA. Lead gen where every lead is similar value.
• Add Target CPA constraint when you want a cost cap.
• Watch out: without Target CPA, you can get conversion volume that's cheaper than optimal but lower-quality.

Maximize Conversion Value

• Use when: New e-commerce campaigns without enough history for tROAS. Growth-mode campaigns where you want to spend the full budget.
• Add Target ROAS constraint to set a minimum efficiency.
• This is the default for Performance Max in growth mode.

Not sure which to run? Answer four questions and the selector below recommends a starting strategy — with the setup notes and the honest caution that comes with it.

5. Enhanced CPC (legacy)

Enhanced CPC (eCPC) sets manual bids that Google adjusts up to ~2x or down to floor based on auction-time conversion probability. eCPC was the bridge between Manual CPC and full Smart Bidding (2010-2019).

eCPC is mostly deprecated in 2026. Google has been retiring it. Move new campaigns to Smart Bidding directly. The eCPC use cases (transition campaigns, low-volume) are better served by Maximize Conversions.

6. The Learning Phase

When you launch a new campaign with Smart Bidding, or change the bid strategy on an existing campaign, the campaign enters a Learning Phase. Google's model needs conversion volume to calibrate the prediction model.

What "Learning" means

• Bidding is more volatile during Learning (Google explores to gather signal).
• CPA / ROAS may not hit target during Learning.
• Learning lasts approximately 7 days, but conversion volume matters more than calendar time.
• Google labels campaign status: "Learning", "Active", or "Limited" (if there isn't enough conversion volume to exit Learning).

Triggers that re-enter Learning

• Bid strategy change (Manual CPC to tCPA, tCPA to tROAS, etc.)
• Target value change beyond 15-20%
• Major budget change (50%+)
• Conversion goal change (adding/removing conversion actions)
• Account-level changes that affect serving (new conversion action, new audience strategy)

7. Volume requirements

Smart Bidding strategies have different conversion volume floors below which they don't work well:

Below the floor, the model can't differentiate signal from noise. Common workaround: combine multiple campaigns into a portfolio bid strategy to pool conversion volume.

8. Portfolio bid strategies

Portfolio bid strategies apply one bid strategy across multiple campaigns. The advantage: pooled conversion volume for learning. The trade-off: campaigns must share the same target (tCPA or tROAS).

When portfolios help

• Several small campaigns that individually lack conversion volume for stable Smart Bidding.
• Multiple campaigns that should share the same ROAS or CPA target (e.g., all brand campaigns, or all SMB-segment campaigns).
• You want one place to update targets across many campaigns.

When portfolios hurt

• Campaigns have meaningfully different conversion economics (lumping together $200-LTV and $2,000-LTV customer segments distorts bidding).
• You need different targets per campaign (each campaign should have its own strategy).

9. Seasonality adjustments and data exclusions

Smart Bidding handles normal seasonality automatically (Christmas, Mother's Day, etc.) by learning patterns over time. But for unusual events, manual signals help.

Seasonality adjustments

Tell Google about a short-term conversion-rate change (1-7 days typical, max 14 days). Example: you're running a big promo from Friday-Sunday and expect 2x conversion rate. Set a seasonality adjustment of +100% conversion rate for that window. Google bids more aggressively to capture the lift.

Use seasonality adjustments for:

• Major promotions (Black Friday, Cyber Monday, anniversary sales)
• Product launches with expected demand spike
• Brand-spike moments (PR coverage, partnership announcements)

Data exclusions

Tell Google to ignore specific date ranges in its learning. Use for:

• Conversion tracking outages (bad data shouldn't train the model)
• Site outages or major site issues
• Unusual events not representative of normal performance (one-time PR event, unforeseen viral moment)

10. Target Impression Share

Target Impression Share bids to achieve a specified SERP impression share. Three variants:

• Top of Page (60-100%) — bid to appear in top SERP positions.
• Absolute Top of Page (10-100%) — bid for the very top slot above all organic.
• Anywhere on SERP (10-100%) — bid for any SERP position.

Use Target Impression Share for:

• Brand defense — bid to capture top slot on brand queries; target 90-100% impression share.
• Strategic visibility goals — defending against new competitive entries.
• Awareness campaigns — reach goals not directly tied to CPA/ROAS.

Set a max CPC cap to prevent runaway bidding when competitors push prices up.

11. Value rules

Value rules adjust the conversion value Google uses for bidding (without changing the actual conversion value passed in). Use to bias bidding toward higher-value segments.

Common value rules

• Audience-based: +20% conversion value for first-party customer-list audiences (existing customers who repeat-purchase have higher LTV).
• Location-based: -50% conversion value for shipping-cost-heavy geos.
• Device-based: +30% conversion value for mobile (if your mobile customers have higher repeat-purchase rate).

Value rules let you encode unit-economics knowledge into Smart Bidding without changing your actual conversion-tracking values. Particularly useful when you have rich first-party data but a flat conversion-value-per-purchase reported to Google.

12. The 10 most common Smart Bidding mistakes

13. Anti-patterns: what NOT to do

14. Migration from manual to Smart Bidding

Pre-migration audit

1. Conversion tracking quality. Validate conversion actions, values, and Enhanced Conversions / CAPI setup. Smart Bidding amplifies conversion-tracking quality — both good and bad.
2. Conversion volume. Check trailing 30-day conversions per campaign. Plan portfolio bidding if volume is fragmented.
3. Current CPA / ROAS baseline. Document current performance as the migration baseline.
4. Bid history. Review your manual bid logs; identify the implicit CPA / ROAS you've been targeting.

Phase 1: Maximize Conversions (or Maximize Conversion Value)

Start with Maximize Conversions (lead gen) or Maximize Conversion Value (e-commerce). No target. Just let Smart Bidding spend the budget while optimizing for the goal. Run for 2-4 weeks.

Phase 2: Add target

Once Learning completes and performance stabilizes, add Target CPA or Target ROAS at current actual values (don't change behavior; just lock in the new baseline). Run 2-3 weeks.

Phase 3: Optimize target

Move target in 10-15% steps toward goal. Lower target ROAS (more spend, more volume) or higher (less spend, more efficient). Run each step 2 weeks before next change.

Quick reference: the “good Smart Bidding setup” checklist