Every marketing manager or business owner would like to know which steps their customers are taking, where they drop off, and why.The problem is that in most cases, people are guessing which steps they need to monitor, so consequently they might set up a “pre-defined” funnel.
However, there are many types of funnel analysis techniques and tools. Today, funnel analysis has evolved into a grained analysis where you can define things like:
- Conversion between steps (See figure 1)
- Global conversion of the whole funnel to completion (see Figure 1)
- Funnel within a session
- Breakdown based on properties within each steps, ( See Figure 2)
Figure 2: funnel with breakdown by country.
The ability to target specific users that churned during a particular step is also important, not just to see the overall picture, but also to automate campaigns targeting those specific users.
This is also available in some tools. In CoolaData for example, using our CoolaSQL behavioral extensions, you can select the specific user_ids or emails, and target them.
for example: SELECT USER_ID,email FROM cooladata WHERE date_range(last 30 days) CLUSTER FUNNEL BY user STARTS WITH “item_viewed” FOLLOWED BY “add_to_cart” NOT FOLLOWED BY “purchased” END CLUSTER .
From Funnel to Path
It’s easy to define funnels for common steps such as register, login, play song, buy, etc., but you may run into problems when you don’t really know which “funnels” lead to a specific action. For example, there are different routes that users take, and you want to know what they are.
Ideally, you would be able to find out which funnels you don’t know about yet.
These funnels could be issues in the business flow that you were previously not aware of, or technical issues that result in users not ever reaching a specific step in a funnel.
That is where path analysis come into play.
Funnel analysis is a investigating a predefined set of steps.
Path analysis is exploiting a collection of many funnels.
Path analysis is more exploratory: you can find out which are your most popular funnels, or what are the shortest and longest funnels.
When you perform path analysis, you will usually want to understand how a specific path is different from other paths, or you can also choose a specific property within a path and see how it varies among the others. (see figure 4)
So what is the connection between path and funnel?
When you focus on a specific path, it’s actually a funnel. You would use path analysis in order to explore the most relevant paths and then “drill into” a funnel and see how it behaves.
There is a direct connection between the exploratory mode, and the daily monitoring of a funnel conversion.
So does Path overlap with Funnel?
Not really. While you explore things in a path, in a funnel you usually have a specific business question you want to investigate and usually monitor.
Session path and user path
Typically when analyzing path within a session it’s more for a technical or operational purposes. You are trying to see what kinds of routes your users take, where they get stuck, and how long specific steps take for example.
If you decide to look at path analysis over a longer period of time, however, you are able to answer more business related questions rather than technical ones. Think of someone who put a few items in a cart but doesn’t check out for few days, then maybe re-visits the cart to remove some items, and then continues.
So how do you use the power of path and funnel together?
The most important thing is to enable the analyst to move easily from identifying a path that is of interest to analyzing and monitoring it as a funnel.
Furthermore, it is vital to get the actual uses/devices/cars/etc… that perform a specific path, so you can target them directly later.
In CoolaData, we enable the analyst to select the actual users that performs a specific path, and target them.
Is path and funnel analysis relevant to only the traditional web?
The truth of the matter is that path and funnel analysis on the traditional web are just easier to explain, and very data savvy in terms of analyzing time-series data. However, path analysis is relevant to many other areas from IoT, healthcare, or even other industries that need to monitor how devices behave over time.