Helping Kanary users protect even more with Profile Suggestions

video demonstrating suggested attributes, showing how Kanary proactively suggests pieces of personal information that was found in initial scans, to add to your profile.

During my time at Kanary, their core product worked like this:

1. You enter your personal information.
2. Kanary's bots scan data brokers for matching information.
3. Kanary works to remove the exposures.

Signup needed to be quick, smooth, and low-stakes. To keep it that way, we required minimal info up front: name, age, city, state.

This was enough data to run an accurate scan for most people and provide an impactful first experience.

However, the ideal experience required adding as much information as possible. A more complete profile would lead to more accurate and thorough scans and better safety outcomes.

So, how can we reliably increase profile completeness, while maintaining a speedy onboarding?

Keeping the whole family safe meant more work

The above problem was multiplied by Kanary's strong support for multi-user plans: depending on the year someone signed up, the Family tier started with 2–5 additional member slots.

It was in their best interest to make the most out of those member slots. This required taking the time to add extra members along with lots of data for each person.

Reconsidering how members add profile data

We explored a bunch of routes to encourage profile completion: guided onboarding, completion metrics that nudged them to hit a minimum baseline, alerts, emails…

But what if they didn’t have to do the work at all?

Here’s a sample of the type of exposures we removed for members:

To find these, Kanary scanned data brokers for info matching our members. But we’d end up saving all the data around it to run more advanced matching algorithms. We realized that data could serve and additional purpose:

First experiment: Member Suggestions

We decided to try using this already-scraped data to suggest members first. It had a few benefits: it was tightly scoped, easy to detect, and supported company goals of increasing member usage. (Each user could have multiple members on their account.) For the user, this would act as a shortcut to creating a new member.

After design iteration, I narrowed to a basic card showing potential relatives with matching last names. While adding members from the card directly could have been nice, we had an existing table interface that could be re-used to better support that. (And we moved fast.)

After letting this run for a while; we saw positive movement on engagement and new member creation. We moved to extend it to the next sensible feature: Profile Suggestions.

Supporting rapid low-effort profile completion with Profile Suggestions

While suggesting members saves some steps of adding a member, profile-level suggestions like alternate names or previous addresses were something that could be added in one click. This was a significant improvement over manual entry for multiple reasons:

1. ⚡️ Speed: one-click is hard to beat.
2. 😅 Mental load: it can take effort to list all prior addresses or contact details. You might not remember some, and you might not even know those of the other members on your account

After iterating on designs, we finally arrived at a pattern which achieved our goals for clarity, speed, and ease-of-use, and as a bonus could also accommodate suggested members. Instead of surfacing these in the profile section, we'd add a card to their dashboard with simple questions:

Did you ever live in Seattle, Washington?

Did [member] ever use the phone number 555-555-5555?

Having found a pattern that met our goals, I built a new card that was able to handle both suggested attribute and suggested member functionality:

This video shows the Kanary profile suggestions feature in use, where users are presented with questions like "Did you ever live in Seattle?" with yes/no buttons for quick confirmation, making it easy to add previously detected information to their profile.

This ended up being an obvious-in-hindsight feature. Users spoke positively about us proactively suggesting data they could protect, and we saw improvements in profile completeness.