Smart Bidding Strategies

Smart Bidding is Google's auction-time, machine-learning bidding. For each auction it predicts conversion likelihood using signals like device, location, time, and audience, then sets a bid to hit your goal.

It replaces manual CPC tuning with automated optimization toward conversions or conversion value.

Smart Bidding strategies split into volume goals and value goals. Volume goals chase number of conversions; value goals chase total conversion value (revenue).

Choose value strategies when conversions differ in worth, and volume strategies when every conversion is roughly equal.

Target CPA (tCPA) tells Google the average cost you will pay per conversion. The engine raises bids on likely converters and lowers them on unlikely ones to hit that average.

It is an average, not a cap: some conversions cost more, some less. Set tCPA near your real historical CPA, then adjust gradually.

Target ROAS (tROAS) tells Google the conversion value you want per unit of spend, expressed as a percentage. A 400% tROAS means 4 in revenue for every 1 spent.

tROAS only works well with accurate conversion values flowing into the account. No values, no usable tROAS.

Maximize Conversions spends the full budget to get as many conversions as possible. Maximize Conversion Value spends the full budget for the most total value.

Both are budget-constrained: they will spend it all. Add a tCPA or tROAS target only once you have a stable goal you want to hold.

The choice flows from your business model. Lead gen with equal-value leads leans volume; ecommerce with varied basket sizes leans value.

Always prefer value-based bidding if you can pass meaningful values, even in lead gen, by sending offline conversion values back to Google.

Smart Bidding needs data. Sparse conversions make predictions unreliable. A rough guide is 15 conversions in 30 days for tCPA and 15+ with values for tROAS, though more is better.

Below that, use Maximize Conversions without a target, or broaden your conversion action to feed the model.

Aggressive targets choke delivery. A tCPA far below your real CPA or a tROAS far above your real ROAS will cause the system to under-spend or stall.

Start near historical performance, then tighten by 10 to 20% at a time, allowing a learning window between changes.

After you set or change a bid strategy, the campaign enters learning. Performance is volatile and should not be judged yet.

Every significant edit, target change, budget swing, conversion-action change, can reset learning. Batch changes and then leave the campaign alone to stabilize.

Smart Bidding offers seasonality adjustments for short, predictable spikes (a flash sale) and data exclusions to ignore broken tracking periods.

Use seasonality adjustments sparingly for 1 to 7 day events. For long-term shifts, just let the model learn the new normal instead of forcing adjustments.

Conversion value rules let you adjust value by audience, location, or device at bid time, for example valuing new customers higher than returning ones.

This sharpens value-based bidding so PMax pursues the customers your business actually wants, not just the cheapest conversions.

Test your understanding of Smart Bidding strategies.

Smart Bidding sets auction-time bids using ML. Volume goals use Maximize Conversions and tCPA; value goals use Maximize Conversion Value and tROAS.

Prefer value-based bidding when conversions differ in worth, set targets near historical performance, respect conversion-volume thresholds and learning periods, and sharpen results with value rules.