Sometimes Google Analytics does things in a quirky way, or has technical limitations you don’t anticipate. Here are four such “surprises,” some of which come up commonly in reporting (the details of sequences maybe not so much). None of these are secrets, but they are all things you might expect to work differently than they do.
1. Exit pages have 0 time on page (if no event occurs).
GA calculates time on page via the timestamp of page load and the timestamp of the subsequent hit. When there is no subsequent hit, such as on an exit page where no engagement hit occurs, the time is 0. You can see this by applying the Bounced Sessions system segment to the Behavior > Site Content > All Pages report. You can set up an event to fire later on to give GA timestamp data if you want more accurate numbers, though you’ll never get it exactly without blowing up your event hit count.
Loves Data has a nice video explaining Google Analytics time calculation, or you can read Google’s session duration documentation page.
2. Revenue attribution is different in Conversions reports than in Acquisition reports.
In the Multi-Channel Funnels and Attribution reports under Conversions, Direct-channel sessions get credit for conversions that happen during them. In the Acquisition reports, that credit is given to the most recent previous session that was in some non-Direct channel (unless there isn’t one in the previous six months). Of course, there’s one exception as well: the Conversions report under Acquisition > Social uses MCF-style attribution.
Annielytics has a post on why Acquisition differs from MCF and some more details around it. If you need more comparable numbers, the Model Comparison Tool allows you to define your own attribution models, so you can give Direct a weight of 0.
3. Non-Interaction Events count as sequence steps.
A session with a single pageview is not bounced if it also contains a typical event – the event counts as a user action beyond the pageview. An event labeled “non-interaction” (the opposite of “engagement”), however, does not prevent bounced sessions, or affect the time on page calculation. That label is intended to be applied to “utility” events that are simply carrying data payload to GA for EEc or similar, not representing active choices by the user. However, in sequence segments they are counted as distinct steps, meaning in some cases bounced sessions can satisfy 2-step sequences.
I have no official references for this; I discovered it while experimenting to work out all the details of sequence segments.
4. Goal funnels have no effect on goal completions.
When you set up a destination goal in Google Analytics, you have the option to define a funnel of preceding pages. You can mark the first step of the funnel as “required” if you’d like to. However, this has no bearing on whether the goal is completed or not! If the destination page is reached, a goal conversion is counted – funnel or no funnel, required first step or optional first step. The only places in GA where the funnel matters are in the Funnel Visualization and Goal Flow reports (Funnel Visualization is a fantastic report, I hasten to add, although you can’t segment it). The only place a required or optional first step matters is in Funnel Visualization: only sessions that completed that first step will be included, which may not include all sessions in which a goal completion happened. All sessions that converted will be included in Goal Flow and the other goal reports.
Google’s page of destination goal examples seems to be the best reference for funnels, oddly. Their page comparing Goal Flow and Funnel Visualization may also be useful.
California ground squirrel photo by JoeBreuer on Pixabay.