What is an attribution model?

Before a user performs any conversion action, they may interact with multiple touch points along the way.

For example, imagine you own an online camera store called Online Camera Warehouse. A customer first finds your site by searching "Second Hand Cameras". The next day, she searches "Buy Used Camera" and comes back to your site. Finally, one week later she searches "Online Camera Warehouse", clicks your ad, and makes her purchase.

Attribution models let you choose how much credit each of these touch points (or clicks) gets for the conversion. Google Ads offers five different rules-based attribution models:

  • First Click: Gives 100% of the credit for the conversion goes to the first touchpoint — in this case the first ad for "Second Hand Cameras".
  • Last Click: Gives 100% of the credit goes to the final touchpoint before the conversion — your brand ad for for "Camera Superstore"
  • Linear:  Distributes the credit for the conversion equally across all clicks on the path — in our example a third of the credit would go to each click.
  • Time Decay: Gives more credit to clicks that happened nearer to the conversion, using a 7-day half-life. For example, a click eight days before a conversion would get half as much credit as a click one day before the conversion.
  • Position Based: Gives 40% of credit to the first and last clicked ads, with the remaining 20% spread out across the other clicks on the path.

When choosing an attribution model, think about your marketing goals. For example, if you're trying to acquire new customers it's a good idea to use a more growth oriented attribution model.

If you'd like more help on this, the Google Ads tutorial video goes into more detail.

Why does Opteo recommend a Position Based attribution model?

By default, Google Ads using a Last Click attribution model. This can be misleading because it undervalues the role of every other touch point in the conversion path.

A Last Click model tends to overvalue bottom-of-funnel campaigns (like Brand or Remarketing),  while undervaluing top-of-funnel campaigns which contain high volume, generic keywords

A Position Based model should give you a better understanding of how your ads perform and help to optimise across the user's conversion path, because it takes into account every single touch point that has contributed to the conversion.

How this Improvement works

Where Opteo finds a conversion action that uses a Last Click attribution model, it will prompt you to change this to Position Based.

Opteo will show you the average number of days from when a user first clicks an ad, to them performing the conversion action. If the conversion action typically occurs more than 1 day after the first click, it suggests there are multiple touch points in the conversion path. In this case, using a Position Based model should be more accurate.

You can then choose to:

  • Complete the Improvement and change the conversion action's attribution model to Position Based.
  • Dismiss the Improvement for a week, a month, or forever.
  • Adjust the attribution model and use First Click, Linear or Time Decay models instead.

Be aware that if you decide to change a conversion action's attribution model you may see a decrease in conversions on the day of the change. This is because all the conversions that happened that day need to be redistributed across previous touch points. You'll now need to wait a few days to get an accurate picture of conversions. You can learn more about this in the Google Ads tutorial here.

Technical Notes:

  • Opteo will only prompt you to change a conversion action's attribution model where that conversion action has more than 20 conversions over the previous 30 days.
  • Opteo will only recommend changing the attribution model to Position Based where at least 3% of conversions occur more than 1 day after the first click.
  • Note that, Google only allows you to change the attribution model for Search and Shopping campaigns. Display campaigns will still use the Last Click model.
Did this answer your question?