These experiments were building on a revamp of the onboarding flow that we did. The experiment result for that initial iteration is 

In the first iteration of the personalized flow, we defaulted to the most commonly used templates within Airtable. Things like Project Trackers, To-Do lists, etc, that were specific to a function like Sales, Product, or Design.

We ran this as AB test and expected a straightforward incremental win, but the results we saw on the other side were very strange — we saw a major regression in sharing volume and collaboration, which seemed like a metric unrelated to the fundamental pivot toward personalization. Our first reaction was that it seemed like the experiment data was simply wrong. Was it a fluke result? But we didn't want to accept that answer, especially with such an important surface and user experience, and also because we might be leaving critical learnings on the table. We went pretty deep brainstorming to try and figure out what was going on. We also put it in front of some users, and asked them about their experience and whether or not they were going to share content with others as they were filling it out. 

What we learned is that the examples in the new personalized onboarding tended to be single player or solo uses of Airtable - something like a calendar or a to-do list just for you. That meant that we weren't teaching users, in the moment, that Airtable is a multiplayer product, that you should share that content with others and bring them in. In rolling this out, we accidentally were setting our users up to actually work by themselves.

This was the newer improved version where we focused on what we knew about you 

as well as what might be best for your team

. If you were a product manager or designer, we showed you the best examples of templates or use cases you could use with your team at work - things like prioritizing the roadmap or measuring and improving results or launching a project. In doing that, we were able to fully correct the regression and also see an activation gain on the other side, both for the user who was putting this together but also for the team.

As you can see, this experiment got us looking at sharing and collaboration more deeply as an opportunity across the entire product. This led us to do work in a few areas: 

Forms - we were able to unlock a lot more virality and sharing there. 

Share Links - we played around with the default pages and sign up flows for users who clicked on link someone shared with them.

Lauryn Isford

When our team was working on onboarding at Airtable, we ran experiments using personalization to drive the completion rate and provide more value to new users of the product.

When users are onboarding to Airtable, they typically have a specific use case or JTBD in mind. However, we require them to build a generic workflow first. This causes users to have a long and arduous time to value, where they are required to do more work before they see the reward of what’s possible to solve their specific problem.

We believe that providing personalized templates and examples in the new onboarding flow (i.e. most popular trackers for someone working in Product, or Design) will reduce time to value and help more users discover the power of Airtable for their needs. This should increase onboarding completion rates and overall activation rates. This hypothesis is supported by learnings from the initial iteration of our onboarding experience and our template library, which both indicate that increased personalization improves activation for teams.

Before - Build Your First Airtable Baseline (Non-Personalized)

+15% lift in Onboarding Completion Rate (% of new teams that make it to the end of onboarding)

-10% decrease in sharing and collaboration

No change in Activation Rate (W4MUA: Week 4, Multi-User Active Rate)

After additional user research, we believe the regression in sharing and collaboration was driven by the specific template choices surfaced to users. The templates and examples proposed are better suited to single-person use cases as opposed to team collaboration (ex: to-do list vs team sprint planning).

While the new onboarding flow is personalized, it over-focuses on individual use cases and under-focuses on the importance of real-time collaboration to achieve value in Airtable.

Run another version of the personalized flow with a revised focus on more well-rounded multi-player and team use cases.

After - Default to Multiuser Templates in Personalized Flow

+10% lift in Activation Rate (W4MUA: Week 4, Multi-User Active Rate)

Dig deeper to figure out other places where we might better promote sharing and collaboration. Now that we know the concept of sharing/collaboration and team workflows is so important to activation, there are probably other areas in the product where we can be more explicit and supportive in guiding users to work with their teammates and encouraging multi-player use of the product.

Onboarding experiment result for personalization at Airtable

Problem 🧠

Hypothesis 💭

Experiment 🧪 - Iteration #1

Results 📊

Sign up to see more