Thinking Funnels

As a product person, I spend a significant amount of time answering questions such as:

1. Would this change contribute significantly to our company goal?
2. Why is this number suddenly down?
3. What is the impact of the feature we just launched??

Those who work in product would be able to label these questions as problem sizing, RCA and output metric check respectively. The common link among all these however is that these questions require a good understanding of the product funnels. It is no brainer then that many product interview questions revolve around the understanding of these funnels.

While thinking in funnels might come naturally to seasoned product folks, it can be challenging for aspiring PMs and product analysts to develop this approach to solving problems. More-so because it is not advertised as much as other jargons like ‘user empathy’, ‘design thinking’, ‘first principles’ and many more. Nonetheless, understanding the funnels not only helps you diagnose and fix problems quickly, but also helps you know all the leverage points in your system that you can use to improve the output metrics.

To help you understand how to build a mental funnel for any feature, here is a simple example that you can modify and use for any major product or feature.

Given above is the simplified funnel of a fictitious e-commerce app called Amazekart. This type of funnel can however be built for any app or feature. All you have to do is map the user journey from start to finish and also map the features that link each of these steps.

Amazekart’s funnel consists of a landing page which opens up on the app open, the listing page which includes the list of items that the user can select from, Item details where the user sees a detailed description of the selected item and a CTA to add it to the cart, the cart from where the user finally lands to the checkout page and the checkout page where the user makes the payment. Each of these steps is linked to the next steps through one or many features and the strength of these features actually control the conversion from the top level to the next level.?

Most often the company leadership would only be monitoring the output, the daily checkout size or revenue in this case but as a product person knowing how all of these steps are performing can help you predict the impact of feature launches, evaluate the impacts of features, understand the reasons for fluctuations in the output numbers.?

For example, let us consider the following is the revenue graph of Amazekart

It seems like the revenue for Month 9 in the Year 2 is not in line with the projections based on the year before. Such an issue is not uncommon but not having a structured approach to diagnosing this can lead to delays in resolution. Here is how the knowledge of funnels could help you solve this anomaly faster.

Consider the following graphs which trace the trend of each step of the funnel for the same duration as the revenue graph above (all numbers are in 1K). Does this view help you isolate the step with the issue leading to the revenue drop?

There seems to be some problem with the cart addition step of the funnel.

Now that we have been able to isolate this step, we can start analysing each of the links that lead to this step of the funnel, eg. the add to cart CTA, reviews, ratings etc. One of these might have had some issue organic or related to a new change in that feature which could have led to this drop .

Let's take another example that would help understand how we can predict the % change in the output when improving or changing one of the links between any 2 steps.

Amazekart noticed that a new payments app, Dogepay has started building a loyal user base recently and that a few customers raised complaints about not being able to find Dogepay wallet payment option. Since Dogepay offers lucrative cash backs, it is likely that these customers would have bought the same item from the competitor app that supports Dogepay wallet payment option in order to benefit from the cash backs. Since this app has low market penetration, the predicted impact of adding Dogepay wallet payment option is about 1% increment in checkout users/cart users ratio. What would be the revenue impact of adding the new payment option??

Here is how you can solve this question quickly by knowing the math of the funnel we designed.?

Revenue = (Landing Page Users) x (Listing Page Users/Landing Page Users) x (Item Page Users/Listing Page Users) x (Cart Users/Item Page Users) x (Checkout Users/Cart Users) x (Average Checkout Revenue)

This formula which connects input metric (Landing Page Users) to the the output metric (Revenue) through a series of conversion ratios can be used to predict the impact of any movement in either the conversion ratio or the input itself on the revenue of the company.

For the case mentioned, the revenue impact comes out to be about 10% or 100K

The understanding of the funnel and the math that links the input and the output can help you solve any RCA or impact problem. Although the funnel shown above is a simplified version created only for developing the basic understanding. In real systems each step might be linked multiple steps and the change made in variable would not translate to a linear change in the output metrics but once the basic understanding is built, this model can be extended for complex systems as well.