Product Management 101: How to deal with high friction steps.

As product managers, we always strive for simplicity. The fewest steps, the least friction, the fastest process. However, sometimes there is a step our product has a step that is high friction: entering a SSN, making a phone call, writing a description, entering credit card information, or something similar.

Ultimately, a good product will reduce that friction as much as possible. How we do that and when we chose to do it is part of the art of product management.

We'll talk about the pitfalls of high-friction steps, but if you're busy here's the TLDR:

TLDR:
  1. Sometimes your product will have a high-friction step
  2. Beware of "silver bullet" thinking: making the process easier will not solve all your problems
  3. Always precede a high-friction step with a low-friction indicator of intent

The problem with high friction steps

Almost by definition, high friction steps tend to have low conversion. More importantly, they tend tend to have low conversion obscurely. We know customers aren't completing a step, but we don't really know why they aren't completing it.

Beware the silver bullet

Many a time I have thought "oh, if only we make this poorly-converting step easier, our conversion numbers will skyrocket." This is silver-bullet thinking and is very risky—there could be many reasons why customers don't complete a high-friction step:

  • There is something buggy with our implementation (ex: the phone number listed doesn't work).
  • The customer doesn't trust us
  • The pricing is wrong
  • They weren't serious about buying and just exploring.
  • Our product is simply not compelling

Betting on a "silver bullet" can be a very bad idea: we devote significant resources to "solving" the friction problem, only to see no conversion bump. Many a company has confused not having product-market fit with "our app is too high friction" and failed on account of that.

The answer: an intent indicator

One way to go around this is to add an additional step before the offending one. This step should:
  1. Be designed to gauge intent.
  2. Be ultra-low friction

Let's look at an example:


Example: A hypothetical email marketing campaign

Imagine we manage a product that requires getting on the phone with a sales rep before purchase. Our main customer acquisition channel is email marketing. The top of our funnel looks like this:
  1. Send emails to potential customers
  2. Customers open email
  3. Customers click through email and hit a landing page
  4. Customers are presented with content and a call to action: “Call xxxxxxx” to proceed.

And the conversion rate for each of these steps is the following:

Step # Completed Conversion
Customers that receive an email 1000 NA
Email opens 200 20%
Site visits 100 50%
Phone calls received 5   5%

The lowest converting step is visit → phone call, and obviously has the highest friction, so we decide to optimize conversions on this step. 

This is what our landing page looks like:

Our first instinct is probably to jump right in and start simplifying the application process. Obviously we need to get rid of that pesky phone call, right?

We think about possible ways to get rid of the phone call.
  • We could replace the phone call with a phone number field and then we call back automatically. We could do this in about a week. 
  • We try to move part of the sales process online. We can do this in about 2 months.

Each of these is a significant project wit added complexity. The first option is technically simple, but will add some complexity to our back office and isn't obviously "better." The second one is a significant undertaking.

It seems like we don't have enough data to make a decision.

Adding an intent indicator

Instead of jumping into a solution without data, we add a lightweight intent indicator. 

We hide the copy that says “call us” and replace it with a button that says “I’m interested and serious about buying this widget.”. After clicking the button we re-display the phone call call-to-action.

It looks something like this:

How is this better? Won’t we get the same five phone calls anyway? Probably, but consider the two following scenarios:

Scenario A

Step # Completed %
Customers that have an email sent 1000 NA
Email opens 200 20%
Site visits 100 50%
Button clicks 6 6%
Phone calls received 5 80%

Scenario B

Step # Completed %
Customers that have an email sent 1000 NA
Email opens 200 20%
Site visits 100 50%
Button clicks 80 80%
Phone calls received 5 6%

Notice that the total number of phone calls received is the exact same, but the intermediate step (button clicks) are very different. 

In Scenario A, it seems like almost everyone who is interested in calling actually makes a call. Our hypothesis of “people aren’t calling because calling is hard” seems to be wrong–our efforts are probably better spent making the landing page more compelling.

In Scenario B, it seems like a big chunk of the drop-off is coming from the phone call. Nonetheless, we’re still better off than before. The decision to dedicate resources to removing the phone call step are now backed with hard data. ( Or maybe we learn that most of the drop-offs are happening when customers land on the page late at night–perhaps they are calling in, just not during business hours. A “call me back” feature wouldn’t help here.)

Regardless of what we end up building, we have far more insight into our customer’s behavior, and it only cost us maybe 3 lines of javascript and 30 minutes to build.