Evolution of a product feature - Assessments
Explore the journey of a feature through feedback and iteration via a real-life example
A real danger in product development is building something in the dark for months or years without getting user feedback or having real customers use it, then having it flop when you finally launch it.
We’d all love to achieve the end goals of our product or feature from day one, but as you know, it takes time. I’m sure you’ve heard by now that you need to get your MVP out there and get feedback ASAP, but what does that look like? And how do you structure your product development roadmap to make that happen?
I present a case study on how to break up a feature and stagger its development. This feature was built for a product I worked on a few years ago. This is close to how it all panned out. Some iterations of this feature might seem very reactive, but these were all roads we had contemplated going down at some point. In some cases, the priority of things changed based on user feedback. We kept moving toward our original goal, which we had set through our initial findings from our original customer interviews.
Needs Assessment feature
A needs assessment is a survey life insurance brokers use to help determine how much insurance coverage you need. It asks about things like marital status, debts, salary, etc. Internally, we called this a Financial Needs Assessment tool (FNA).
The evolution of this feature has several stages. Some evolutions set the feature on a new trajectory, while others are incremental upgrades or improvements. Before we start on any product or feature, having an idea of the end goal can be helpful. This way, we don’t paint ourselves into any corners, and we can hopefully avoid having to rebuild large portions of it.
During one of our town hall meetings, I explained this process as a branching tree of if-then statements. Once we release one version of the feature, we stop, gather feedback and evaluate our next move. We often have a sense of what that next move will be, or at least several options. Sometimes, we’re surprised, and things go in another direction.
Our end goal:
Ultimately, our goal is a fully customizable needs assessment where brokers can build their own templates, use pre-configured templates, add their own questions, customize recommendations, and make sales by embedding assessments directly through their websites.
Before getting this far, we interviewed many of our brokers to review their current digital and paper workflows. That discovery led to the planning of the needs assessment feature. There were a few options and ideas for what this could eventually evolve into, but we had to decide on a starting point.
The needs assessment feature actually breaks down into the following four sub-features, which can then be broken down into even smaller increments or improvements:
Assessment composer: Used for adding custom questions and creating and managing assessment templates.
Default templates
Turn questions on or off
Add custom questions
Save new templates
Survey: The actual end-user experience of answering the assessment questions.
Assessment notifications and reminders
Auto saves progress
Intuitive UI
Encourage the client to complete
Analysis: The processing of the user’s answers to the assessment questions.
Include retirement planning analysis
Add an amortization table
Printable answers page
Recommendation: What we ultimately end up doing with the analysis. The final screen of the assessment.
Recommend coverage needs
Recommend specific products
Allow clients to buy insurance products on the spot
Allow brokers to adjust recommendations
Before we could do any of this, there were a few prerequisites we had to make sure were in place:
The ability for clients to log-in securely.
Brokers need the ability to request a client to log-in and complete the assessment.
It’s important to consider whether there is work to be done or other features to exist before you can move forward with your feature implementation.
Version 1 - MVP: Hardcoded assessment
It’s important to remember the future plans for this feature, even at this stage. One known future state was having different question templates and possibly adding new ones. So, we implemented the concept of a template from day one.
Goal:
Allow brokers to send their clients a pre-determined set of assessment questions to answer. Notify the broker when the client has completed the question and allow the broker to view the client’s answers.
Technical considerations:
Considering the architectural impacts of customized assessments, we designed our system to support templates from the start but hardcoded everything for our users for the MVP.
Feedback questions:
At this stage, we’re seeking feedback about the default question set, broker, and client experience. Since this is the first version of this feature, we’re looking for the most significant areas of concern and will prioritize those next.
Actual feedback:
Most brokers were okay with the default assessment questions. Still, specific questions (or groups of questions) weren’t always required, depending on the type of insurance they sell.
Version 2: Customize assessments - composer
This is our first iteration of the composer for creating an assessment. There’s expected to be lots of feedback around UX/UI. We’ve shown some mockups to brokers already, but we’ll need to get it in their hands to learn more.
Goal:
Allow brokers to turn off some questions and add a custom message before sending an assessment. In this iteration, we created an interface where brokers could preview the assessment, toggle questions as needed, and add a message before sending the request to complete to their client.
Technical considerations:
Since our default assessment is a template, we also saved each customized assessment as a template and referenced that template ID when sending it to the client. This way, we’d know which questions were disabled.
Feedback questions:
What is the end client experience like?
What’s the completion rate of assessments?
What are brokers asking for next?
Actual feedback:
Brokers want custom questions. Turning questions off is helpful but not enough for many brokers.
Version 3: Save and re-use templates
Since we used the concept of templates in the background for Version 1 & 2, we’re already part of the way there. This workflow involves a lot of UI work, so it gets its own increment.
Goal:
Allow the broker to save the configured assessment questionnaire as a template they can reuse. They can then quickly select this template to send to future clients.
Feedback questions:
Have we allowed the broker enough customization to focus on the next phase, recommendations?
Actual feedback:
Brokers still want the ability to add custom questions. That will have to wait, though, because they’ve also seen the value in having recommendations for themselves and their clients. Many brokers were still taking the responses and copying them into their own spreadsheets to calculate coverage needs.
Version 4: Provide recommendations after the assessment
As part of our research with brokers, we’ve obtained copies of their needs assessments and decided on the most commonly used formula for calculating insurance needs. We’ve also consulted with our internal insurance experts to test and ensure this calculation is adequate.
Goal:
Save brokers from copying assessment information to their own templates for calculation. Through the needs assessment tool, give the broker and the client recommendations on insurance coverage needs.
Feedback questions:
Is the formula we used adequate?
Are there parts of the recommendations brokers want to leave out?
Actual feedback:
The recommendation formula could use some tweaking, especially for high-net-worth individuals. Brokers are again asking for custom questions.
Version 5: Add custom questions
As mentioned before, some brokers wanted the ability to add their own questions. We’re revisiting this after making some progress in other areas of the product.
Goal:
Allow brokers to add custom questions to the needs assessment.
Feedback questions:
Do brokers require a way to store these custom questions for use across multiple templates?
Actual feedback:
Custom questions are OK for now, but more options to organize them and use different fields (e.g., checkbox, short text, multiple choice) would be helpful. More importantly, though, brokers want the ability to customize recommendations, so that’s what we’ll focus on next.
Wrapping up
I could go on and on about this one area of our product. The above-mentioned changes happened over the course of about one year of development. We jumped between this and other product feature areas during that time. Our cadence was good; it gave us time between each iteration to see how the assessment tool was used, gather more feedback, and plan our next move.
Instead of building this in the dark for a year, we launched the MVP within a few weeks, got user feedback, and iterated. We may have ended up in a similar place regardless, but many small things (not mentioned here) made it into our product due to this critical feedback from real-world users.
There’s no one-size-fits-all solution for breaking a feature into multiple iterations. What will work depends largely on the feature, existing frameworks, and the business. Part of your job as a Product Manager is to figure out how to get value to your users and the business quickly. Don’t be afraid to get creative about how you do that, and get feedback with as little investment as possible to avoid building the wrong things.