The concept of assisted conversions was introduced by Universal Analytics. Although analytics tools use them less often today, they remain an important metric in conversion attribution analysis.
Previous article (part 12): Assisting or cannibalizing?
What is an assisted conversion?
We always talk about assisted conversions in reference to a traffic source dimension (this can be the source, medium, campaign, etc.). Assisted conversions are conversions in which the given source appeared on the path but was not the last click.
Let’s consider some example conversion paths. We have 6 + 5 + 6 = 17 conversions:
The assignment of last-click and assisted conversions will be as follows:
| Source | Last-click conversions | Assisted conversions |
|---|---|---|
| Cross-network | 5 | 17 |
| Paid Social | 6 | 0 |
| Referral | 0 | 5 |
| Display | 6 | 6 |
Assisted conversions are an additional measure indicating the potential impact of a given source on conversions, which may not be visible in conversion reports using last click attribution.
The assisted conversions report in Google Ads
We can see an assist report e.g. in the Google Ads attribution reports:
In this case, however, the report doesn’t show assisted conversions, but assist clicks and views. These indicate how many times a given campaign was clicked or viewed before a conversion, except for the click that was the last one before the conversion. These numbers are much larger than the number of assisted conversions.
If we applied this methodology to our example above, we would get the following results:
| Source | Last-click conversions | Assist clicks |
|---|---|---|
| Cross-network | 5 | 17 |
| Paid Social | 6 | 108 |
| Referral | 0 | 20 |
| Display | 6 | 18 |
Problematic assisted conversions
Assisted conversions are often misinterpreted. One typical mistake is adding assisted conversions to last-click conversions — in order to obtain the number of conversions influenced by a given campaign. This is not a correct methodology, because in some conversions a given channel will appear both as the last one on the path and at earlier stages, as in the conversion path Cross-network > Referral x 4 > Cross-network from our example.
The actual number of conversions involving a given source is as follows:
| Source | Last-click conversions | Assisted conversions | Conversions involving the source |
|---|---|---|---|
| Cross-network | 5 | 17 | 17 |
| Paid Social | 6 | 0 | 6 |
| Referral | 0 | 5 | 5 |
| Display | 6 | 6 | 6 |
While last-click conversions (or conversions assigned in another attribution model) can be summed, and their sum should represent the total number of reported conversions, assisted conversions for individual channels cannot be summed.
That number would have no meaningful interpretation and would change depending on the source dimensions used. The more granular the breakdown we apply, the more assisted conversions there will be in total.
The terms assisted conversions and assist clicks are not the most fortunate ones. The word “assist” suggests that the given click somehow supported the occurrence of the conversion — that it increased its probability.
In reality, however, we don’t know that. It’s possible that the impact of a given click is negligible; one can also theoretically imagine situations in which a given interaction has a negative impact on conversions. That’s why a much better term would be accompanying clicks. See also the article Assisting or cannibalizing?.
In reports using multi-touch attribution, the concept of assisted conversions starts to lose its meaning (the definition of assisted conversions refers directly to the last click model).
Conversions involving a source
It seems that a sensible supplement to attribution reports is a report of conversions (key events) and revenue obtained with the involvement of a given source — that is, those conversions (key events) in which the given source appeared anywhere on the path (see the table above). Such a measure makes sense regardless of the attribution model used.
In Google Analytics we don’t have direct access to such data, but it can easily be obtained from the attribution paths report. The report should be downloaded to Excel and then filtered by the name of the channel for which we want to determine the conversions it was involved in. For example, we want to see the key events and revenue in which the Affiliates channel took part:
After summing, we get the number of key events and the revenue in which the given channel took part (in this case Affiliates).
If we remove the last interaction on each path in the downloaded file, the sum calculated this way will be a measure of assisted conversions in the traditional sense.
To avoid drawing wrong conclusions from these summaries, it’s worth first making sure that the total number of key events and the total revenue value in the attribution paths report match the conversions (key events) and revenue reported for that period in the acquisition reports.
