Increasing user retention by 7% with a new onboarding
Talent seekers were taking too long to engage with their job openings after posting them, leading to a high dropout rate in Torre.
We identified that the main cause was the complexity and unclear descriptions of some features on the platform, which were also difficult to understand for new and existing users.
To address this, I deisgned a self-guided product tour that explained the most important features of the platform.
This effort resulted in a 7% increase in user retention and a 14% improvement in positive user feedback.
Company:
Torre.ai
Role:
Product Designer
Tools used:
Figma, Notion, Metabase
About Torre.ai
Torre connects job seekers with opportunities and provides talent seekers a platform and ATS to manage their entire recruitment process. For talent seekers, the three most important sections are:
The job post: Where talent seekers provide all the necessary details about the opportunity, including responsibilities, compensation, location, required skills, contract type, and so on.
The pipeline of candidates: A Trello-style board that shows the recruitment flow. It allows talent seekers to review and move around candidate cards as they move along the different recruitment stages.
The candidate's profile: A compact view of the candidate's information, including skills, interests, professional experience, compensation preferences, reputation, and how well they match the role.
Let me give you a quick thirty second tour of these three views.
Please ignore any weird candidate name, since this is a test job opening 😬
Houston, we have a problem…
During one of our weekly metrics review with the Product team, we identified two worrying metrics related to talent seeker engagement:
Activation time was too long. This meant many talent seekers weren't interacting with their job openings after posting them, meaning no distribution or reviewing candidates.
Drop-off rate was too high. Many talent seekers weren’t posting a second job after they posted their first. In other words, users were churning.
This meant we were losing users early and struggling to retain them long enough to monetize. And for any startup, that's a big problem.
Identifying and tackling the root of the problem
Step 1: Research
With the research team, we gathered information from talent seekers in different stages of the conversion funnel, including recently acquired, active, and retained; to better understand their challenges with Torre's platform. We found the following:
Only 30% of talent seekers were able to hire someone.
They felt Torre asked for too much information to post a job.
They felt Torre required too much time investment, especially while facing tight deadlines to hire.
Most of them mentioned it was hard to understand some features and concepts within the platform.
Together with the Product Manager, we reviewed this analysis and concluded that:
The first issue could be a consequence of the two negative metrics we identified, as hiring is rarely a quick process.
Issues two to four had a potentially strong correlation with the negative metrics we identified, so we decided to tackle them.
After brainstorming with members from the Operations and Engineering teams, our overall feeling was that most talent seekers didn't understand how Torre's features could help them be more efficient in their recruitment process.
The most common emotions from recently aquired users.
The most common emotions from active users.
Step 2: Looking for inspiration
We needed to improve how we explained users about the platform's features, especially talent seekers.
It couldn't be manual, because we didn't had enough resources to handle all and we didn't want to throw more work to our Operations team. So we aimed for something simple, scalable, and automated to integrate directly into the platform.
So with that constraint in mind, I analyzed how other complex platforms onboarded new users for inspiration.
I reviewed these ideas with the Product Manager and we agreed to move forward with a self-guided product tour, similar to Canva's or Toggle's, because:
It could be implemented faster than a knowledge base.
It allowed us flexibility as steps could be easily added, edited or removed to update the tour's experience.
We could ensure users interact with it, unlike a checklist or knowledge base which could be ignored.
It allowed us to track at which step talent seekers left the tour and analyze its engagement.
It could be implemented with a third-party provider solving our engineering constraints.
Step 3: Hypothesis and defining metrics
Our hypothesis was that the self-guided product tour for the platform's main flows would increase talent seeker engagement, positively impacting the two metrics we had issues with:
Activation time for talent seekers (in hours) — how long it took for talent seekers to interact with their job after posting it.
Talent seeker retention (%) — how many talent seeker returned to post a second job after posting their first.
Other secondary metrics we though were important to analyze the effort's impact were:
Overall product tour completion (%) — To see how many users who started the tour completed it.
Overall positive experience feedback from talent seekers (Net Promoter Score)
Step 4: Analyzing alternatives
As I mentioned before, due to engineering constraints, the initial idea was to have a third-party provider implement the self-guided product tour. I researched and compared several providers in terms of features and pricing to propose the best solution.
List of key features and pricing to consider when evaluating providers.
Final review of the options, with a brief summary and a 5-star rating review for each.
After comparing providers we decided to do a demo with Chamaleon. However, branding constraints and a proof of concept (POC) developed surprisingly fast by one of our engineers made us realize we had made a mistake.
So, we took a U-turn and developed the effort in-house. This approach allowed us to implement it faster and with a design that was fully aligned with our design system. This mistake delayed the release by three days.
POC for the 'Posting a job' flow
POC for the 'Reviewing a pipeline of candidates' and 'Reviewing a candidate's profile' flows.
Step 5: Design, specification, and deployment
After a couple of days, I handed off the design and specs of the self-guided product tour for the three most important flows for talent seekers:
Posting a job
Reviewing a pipeline of candidates
Reviewing a candidate's profile
Given that the tour relied on widely-used and proven UX patterns from companies with great design, I argued that a usability test was unnecessary. This also meant we could recover the days lost from the third-party test hiccup. Both the Product Manager and the CEO agreed.
However, after development, I conducted a thorough quality assurance process with the engineering team in a feature-flag environment to ensure:
Smooth transitions between steps
Accurate positioning of the hotspots
Correct behavior when pausing, restarting, or ending a product tour in each flow
Key takeaways and learnings
The effort was released in September 2023. Within the first month of deploying the self-guided product tour, we analyzed the metrics defined in our hypothesis with the data analytics team. The results were:
The activation time for talent seekers remained too variable month-over-month to attribute changes directly to this effort.
Talent seeker retention increased by 7%, measured by whether users returned to the platform to post a second job.
The overall product tour completion rate was 37%.
The positive experience feedback from talent seekers improved by 14%.
Based on these results, we decided to keep the self-guided product tour, and moving forward, our focus will be on increasing its completion rate.
Some personal notes:
It is crucial to keep all close team members updated about your work. The POC was developed by one of the engineers in my squad when he became aware of the effort. My mistake was aligning and requesting feedback only from the Tech Lead and the engineer assigned to the project.
Due to constraints, like a tight deadline, 'genius design' can be applied sometimes, but it isn't ideal.
⚠️ Update (April 2025)
In order to increase its completion rate and highlight new relevant aspects of the platform, the self-guided product tour was updated. The new version is estimated to be launched in Q3-2025.