Smart Bidding guide for display campaigns

Determine the results of a display campaign that uses Smart Bidding

While you evaluate overall performance, don’t focus too narrowly on individual ad groups or short time periods (such as performance on a specific day, as short spikes can occur), especially if you set a target at the campaign-level.  Make sure that the Target Cost Per Acquisition (CPA) or Return On Ad Spend (ROAS) is realistic based on the historical average of your Display campaigns.

Example

In the chart below, Ad Group C is performing above the goal CPA, whereas Ad Groups A and B are meeting the goal CPA. Since the campaign is hitting the goal, you shouldn’t pause Ad Group C. If you pause Ad Group C, the overall volume will suffer. Smart Bidding will learn attributes about Ad Group C, which then lead to a CPA above goal and lower bids.

For best results, you can combine Ad Group C with the other ad groups to allow Smart Bidding to optimise across your traffic. You can also pay for conversions when using eligible target CPA campaigns.

Smart Bidding guide goal CPA chart | Google Ads | Display

Smart Bidding accuracy and volume

Smart Bidding is more likely to perform at your target CPA or ROAS if you evaluate a time period with more conversion volume. The table below shows expected accuracy given the conversion volume in a time period or campaign hierarchy evaluated.

Impact of conversion volume

Number of conversions in ad group or time period

CPA or ROAS performance fluctuation

Reaction time to changes

Initial learning period

Fewer than 30

Medium to high (up to 100%)

Slow (up to 2 weeks)

Up to 4 weeks

50

Medium (up to 50%)

Medium

Up to 3 weeks

100

Low (up to 20%)

Fast

Up to 2 weeks

500

Very low (less than 20%)

Very fast (as few as 6 hrs)

Up to 2 weeks

Example

If one ad group out of a large campaign only has 15 conversions during a time period analysed, you'd expect daily CPA volatility in that ad group. The overall campaign might be consistently hitting the CPA over multiple days. The chart doesn't apply for longer time periods (all time or multiple months).

Focus on overall campaign performance trends rather than individual ad groups or short date ranges with more volatility. The algorithm is designed to react and meet the target over time in each ad group. Given the dynamic nature of the auction, you're likely to see some fluctuation in CPA or ROAS performance.

Smart Bidding and campaign scale

Our programmatic bidding algorithm is designed to provide your campaigns with the most conversion volume at the chosen performance goal, CPA goal for target CPA and ROAS goal for target ROAS. As the market shifts (for example, due to changes in user behaviour), conversion volume and spend may fluctuate in order for this algorithm to continue to meet your performance goal. There are a few steps you can take to scale your campaigns:

  • Increase your CPA goal or decrease your ROAS goal: These will allow our algorithm to bid higher in each auction which results in a higher chance of winning impressions that result in conversions.
  • Expand your targeting: With more impressions to bid on, our algorithm is able to find more conversions.

How does a change in conversion rate or bid impact your scale?

Typically, there’s a strong correlation between bid and scale, and conversion rate and scale. A bid or conversion rate increase is likely to result in more scale, and a bid or conversion rate decrease is likely to result in less scale.

Switching to Maximise conversions from target CPA or target ROAS does not increase scale at the same performance goal. For example, switching from target CPA to Maximise conversions will not increase scale at the same CPA. Instead, you may see higher conversion volume, but at a higher and less efficient CPA. The same goal can be achieved by remaining on target CPA and increasing the CPA goal.

Was this helpful?

How can we improve it?
Search
Clear search
Close search
Main menu
10694293634750763396
true
Search Help Centre
true
true
true
true
true
73067
false
false
false
false
false