When reducing friction is the wrong thing to do for new users
The traditional wisdom tells us that there should be the least amount of friction between the user and their experience of the value proposition. Decrease the time to value prop, retain more users. The traditional wisdom if often wrong though.
What Is Duolingo Ninja?
Duolingo Ninja is a browser extension that lets Duolingo users download their vocabulary but also to use it to translate parts of the websites they visit based on that vocabulary. It also has flashcards.
So, essentially there are three value propositions:
- Download Duolingo vocabulary (which Duolingo doesn’t let you do otherwise)
- Keep practicing the vocabulary by seeing the words translated on websites you visit
- Keep practicing the vocabulary with flashcards
The road to user activation (from discovering the product to being engaged) looks something like this usually:
Reducing Friction Doesn’t Always Reduce Friction
What product managers try to do is to remove the steps. Less steps, more users going through the setup, more engaged users, more paying customers.
It’s a solid idea. In most cases it’s true. You should not complicate or add friction usually. The fewer setup actions the users has to do, the better.
So, this is exactly what I tried to do. I added a freemium plan without signup. This meant that I had three types of users:
- Free plan users without a login (signup)
- Free plan users with a login
- Paid plan users with a login
In the first case, you just install the extension. click on the sync button to get the Duolingo vocabulary, and you’re good to go. In the latter two cases you need to register first.
The hope was that more users would reach the value prop (a-ha moments) this way.
But this is what actually happened:
- I reduced the setup steps for the user
- This actually increased (!) complexity in the back-end
- The increased complexity resulted in more bugs for the user
- More bugs meant more friction for some users and more work for me
Where did the complexity come from?
I somehow needed to separate freemium users from the signed up ones (nevermind if paying or not). This meant a separate table in the database. Which also meant a more complex process for automatically registering the freemium users in the background. Which also meant that bugs were more likely to pop up as the complexity was increased at least 2x across many parts of the app.
It also meant that user tracking got way more complex and bugs got harder to solve.
This all reduced profitability because I was spending my time dealing with the bugs instead making more meaningful improvements (or working on marketing).
Doing the opposite — increasing friction to reduce complexity
I also have another language learning app called Lingo Champion (directed more towards the people who peak out and can’t seem to get to fluency).
I wanted to cross promote Lingo Champion to Duolingo Ninja users. But the problem was I didn’t have the email of the free plan users who just installed the extension but didn’t register. That was the majority of users.
Yes, I nudged them a bit from inside the extension towards Lingo Champion, but I still wanted to be able to email them as well.
I decided to try out pushing them to register (without forcing it or making it mandatory — because this would have also meant a lot of extra work).
So, I changed the Duolingo Ninja homepage links. They now all point to the registration page instead of directly to the Chrome Web Store. A pretty simple change — done in an hour.
The free plan without registration was technically still available but this allowed me to test out “removing it” without actually having to remove it which would have meant migrating databases, rewriting a lot of the code etc.
Note that this quick fix only worked because the majority of the traffic comes through the main website (not through the Chrome Web Store).
I expected to get slightly more email address but slightly less users who reached the a-ha moment. It’s not what happened.
So, what happened is that the number of registrations almost tripled but the number of users who reached the a-ha moments stayed the same.
Which means my email list is now growing 3x faster without any effect on the paid user growth for Duolingo Ninja.
Perhaps one reason why this happened is that users have a strong need but the registration form is super simple — email and password.
It’s too early to say what the long term effects are but there’s an argument to be made that the users who register might be more likely to give products a proper try. Because the already put in a little bit of effort.
Note that even if the nr of users reaching the a-ha moments would have gone down, it would have not automatically meant that the number of paid users would have been reduced as well. If the pain is big enough, the user are willing to jump through some extra hoops.
Some other edge cases
That’s not the only case where adding friction makes sense.
You might want to consider adding friction also where the user needs to make a conscious choice about something. And that choice can’t be automatically detected with good accuracy.
When working at Pipedrive, I saw this being applied in many places. For example, in workflow automations — users needed to be super mindful about what they were doing to not mess up the data across their whole CRM. So, default values etc. (which would have reduced friction) were not a good idea there.
When the data is ambiguous, it also makes sense just to ask the user. It’s why many companies ask you about how you discovered their product. Yes, traffic source tracking / attribution works to some extent but it’s not super accurate.
Don’t be afraid to add extra friction if you feel like it’s justified.
