+3 votes
We're currently running incrementality tests. One of the things we're looking at is the similarity between networks and organic for a couple of behavioral indicators (eg. Reg rate at D0 or D7).

We noticed very different results when doing these tests with view-through attribution and click-through attribution. As an example, Snapchat is a poorly performing network compared to organic with CTA and suddenly becomes very similar to organics with VTA - indicating that it's cannibalizing our organics with VTA.

Therefore, we're willing to shift away from the standard 1-day view, 7-day click attribution.

Has anyone run tests on changing the view-through attribution window, and do you have any recommendations to share per network?

by (170 points)

1 Answer

+1 vote

It's hard to give a generalized answer for each network, but certainly I have seen that VTA can cannibalize organic installs. But there are a few things to consider when integrating this kind of analysis into an incrementality study:

- How does the delivery of paid installs change when VTA is extended? Extending VTA allows the campaigns to "win" more in the auction, which in turn should theoretically increase your delivery volume (more attributed traffic = more money for the network = higher expected value in the auction). So even though organics may be being cannibalized, the increased delivery could potentially make up for that. 

How does the volume of organics change if you turn off (or scale down) VTA-enabled campaigns? The flip side of cannibalization is that more paid media will almost inevitably drive more awareness, which in turn bolsters organic volumes. If you scale down paid UA because you believe it is poaching organics, you might also see the underlying volume of organics decrease. You need to measure this dynamic.

by (15.2k points)