Best practices when switching to data-driven attribution

After you change a conversion action's attribution model to "Data-driven," you should expect to see some changes to your "Campaigns" reporting. You should also monitor and update your bids and targets.

Changes to expect in "Campaigns"

After changing your attribution model, you may notice changes to the reporting in your "Campaigns" tab.

  • Fractional credit: Credit for a given conversion is distributed between contributing ad interactions, according to your selected attribution model. As a result, you'll see decimals in your "Conversions" and "All conv." columns for the first time.
    Example 1
    You select the "Linear" model. A user follows the path keyword1 > keyword2 and then converts. In this case, each keyword will display 0.5 in the "Conversions" column from that conversion.
  • Time lag: Google Ads reports conversions according to the date of the ad interaction. Since a non-last click attribution model shares conversion credit between multiple interactions, each of which happened at a different point in time, the time lag associated with your "Campaigns" reporting may increase. As a result, you may see a temporary, slight dip in conversions for very recent days after changing your attribution model.

    To understand how this time lag impacts your business reporting, look at "Avg. days to conversion" in the "Path metrics" attribution report. We recommend that you wait to evaluate performance until the average number of days to conversion have passed.

Update your bids and targets

If you change your attribution model, you should update your bids and targets for each conversion included in the "Conversions" column. Otherwise, the change in conversion attribution may result in over- or under-bidding.

If you only use manual bidding, performance will not be impacted (positively or negatively) until you begin to optimize based on the new data. However, changing targets is especially important if you're using Target CPA or Target ROAS bid strategies, as illustrated in the following example:

Example 2

You use Target CPA bidding. You change to the "Data-driven" model for your conversion action. You have two campaigns with the following last-click performance in the "Conversions" column over the last two weeks:

  • "Brand" (lower-funnel) campaign: $5 CPA, 200 conversions
  • "Generic" (upper-funnel) campaign: $20 CPA, 50 conversions

Because you use Target CPA bidding, the targets above are also the targets used in bidding.

Now, let's say that you look at the "Conversions (current model)" column, which reflects historical data-driven attribution performance, and observe the following:

  • "Brand" (lower-funnel) campaign: $6.67 CPA, 150 conversions
  • "Generic" (upper-funnel) campaign: $10 CPA, 100 conversions

If you were to switch from the "Last click" to the "Data-driven" model while leaving tCPA targets at the original $5 and $20, the changes in conversions would imply a set of bid changes that would underbid on the "Brand" campaign and overbid on the "Generic" campaign.

Calculate adjustments to your bids and targets

For Target CPA

  1. In the "Campaigns" tab, look at the performance for conversion actions according to your newly selected attribution model by adding the "Conversions (current model)" and "Cost / conv. (current model)" columns to your reports. Learn how to add and remove columns in your statistics table.
  2. Compare these columns to the "Conversions" and "Cost / conv." columns to see the change from the previous model to the current model.
  3. Calculate the percentage change in "Cost / conv." column.
  4. Set your new CPA targets by adjusting your previous CPA targets by the same percentage change, at the campaign level, as illustrated in the example below.

Note: When selecting the date range for your analysis, a good rule of thumb is to exclude the most recent few weeks to avoid the effect of time lag between click and conversion.

Example 3

You change the attribution model for your conversion action on September 1. Because you should exclude the last 14 days of data (August 18-31) from your analysis, you use the data from August 4-17 to help you calculate your Target CPA adjustments.

Campaign Cost Conversions Cost/conv Conversions (current model) Cost/conv (current model) tCPA adjustment
Generic 1 $250 200 $1.25 280 $0.89 +29%
Generic 2 $500 600 $0.83 520 $0.96 +16%

For the "Generic 1" campaign, there is a $0.36 decrease in cost per conversion. To account for this decrease, you should adjust your tCPA down 29% (0.36/1.25=0.29).

For the "Generic 2" campaign, there is a $0.13 increase in cost per conversion. To account for this increase, you should adjust your tCPA up 16% (0.13/0.83=0.16).

For Target ROAS

The steps to calculate your new ROAS targets are similar as above but with different columns:

  1. In the "Campaigns" tab, look at the performance for conversion actions in your newly selected attribution model by adding the "Conv. value (current model)" and "Conv. value / cost (current model)" columns to your reports. Learn how to add and remove columns in your statistics table.
  2. Compare these columns to the "Conv. value" and "Conv. value / cost" columns to see the change from the previous model to the current model.
  3. Calculate the percentage change in "Conv. value / cost" column.
  4. Set your new ROAS targets by adjusting your previous ROAS targets by the same percentage change, at the campaign level.

Note: When selecting the date range for your analysis, a good rule of thumb is to exclude the most recent few weeks to avoid the effect of time lag between click and conversion.

Was this helpful?
How can we improve it?

Need more help?

Sign in for additional support options to quickly solve your issue