Individual vs Account Level Insights

Regardless of the role you are in or the team you are on, data is critical in understanding what actually happens within the walls of your product, what your customers need from you, and what you’re missing. While I think general data fluency has increased across businesses, a very specific lens is usually applied that prevents teams and individuals from seeing the whole picture.

• Marketing teams focus on throughput and whether or not the people they drove to the product became users and customers.

• Customer support tends to analyze common support requests to track down issues that may be impacting many users.

• Design tends to be more interested in watching session replays, looking at heat maps, and understanding how users navigate each screen or page.

• Product managers are supposed to pull it all together and tie it back to the business metrics so you can put dollars to effort.

The last bit is where I’ve seen many teams struggle. The hard part isn’t tying product or user insights back to business metrics. It’s connecting those insights to the right business metrics to justify (and take) the most impactful actions.

Here are a few examples of bad conclusions from data that I have seen:

• “Oh, there’s a segment of power users!? We should send them a marketing email to upgrade their account to enterprise.” Turns out the power users aren’t the decision makers in that company and have no idea who handles contracts and licensing.

• “20% of users drop off of this flow on the third step, so we need to change it.” Yeah, but that’s because they read the copy and realized it’s a free trial and not a free account, the other 80% flew through it without any problems.

• “We see users on this page hesitate to take action, we need clear direction and more CTAs.” No, it’s a dashboard, they are just soaking in all of the beautiful information.

All that being said, I thought it would be helpful to dive into the difference between finding actionable user insights and actionable account insights and provide some specifics on the type of actions you can take from them.

Another hat tip to Elena Verna, for sparking the idea for this post. Her chat with Lenny on Product Led Sales and follow-up post planted the seed for talking through that shift in how we approach data analysis. It's a reminder that our strategies must evolve with the landscape.

Navigating :allthedata: is about experimenting, learning, and iterating based on a blend of data, intuition, and feedback. As we dip and dive into this topic, my goal is to help you identify the right mental models for leveraging the right data in a meaningful way.

Early-stage startup without a lot of data to pull from? Check out my Worst Case Impact approach to getting buy-in on big ideas.

Start with the Money

Regardless of what team you are on, what initiative you are a part of, or what your role is, you need to remember that businesses run (and live for money). Your impact should be tracked back to money in some way. Maybe your internal tool improves efficiency and saves the company from having to hire. Maybe your research helped an acquisition campaign perform 2x better than average. Maybe you created a tutorial that was added to the onboarding process, helped activate 20% more users on a key feature, and resulted in longer-term retention and increased LTV.

While some of these things are harder to pin down, they’re all driven by objectives tied to increasing revenue or profitability, right? While there are lots of ways to generate revenue, the primary source in SaaS tends to be from subscriptions and licensing. The pricing model used here is what determines which type of analytics you need to evaluate for specific tasks.

Individual Self-Serve: This model relies on users purchasing their own plans. Some might say this is B2C, but plenty of companies use it as an entry point for B2B sales. Allowing individuals to pay for access to the product before the company does is a pretty great PLG tactic. There are also things like company reimbursement that get factored in outside enterprise-level packages that still make me consider it part of B2B. Figma and Grammarly are the shining stars in this space. User-level data is where you will spend most of your time,

B2B Self-Serve: Subscription plans that limit the number of seats/users or allow multiple users but limit feature usage are designed for teams and companies. If your pricing model is PLG/self-serve, you probably look at the individual user who purchased the plan and then shift to looking at account-level data to understand when they are primed and ready for upsells/upgrades. Even in this scenario, you are splitting your time between looking at the account as a whole and user-level data — be it segmented by account or grouped across multiple accounts.

B2B Sales-Led: If your pricing model requires some level of customization or calculation based on the needs of the customer, you’re basically in the same bucket as B2B self-serve with the disadvantage of not having a systematic/universal way to identify when the account is primed for upgrading their plan. For example, if your plans set a seat limit of 10 in the B2B self-serve model, you know when the account is at 8+ seats, they are almost ready to add more and can set up some automation. You may have sold them 23 seats in this model and another customer 47. This means that your sales and accounts team have a bit more leg work to do on the per-account analysis front.

Hey There User

User-level analysis is a magical thing. It tells you all about how people(real humans) use and behave within your product. More than that, it helps you understand where the differences exist across your user base. Some users may lean heavily into a specific feature while a subset of others are barely using it at all. These types of insights are absolutely critical to building great products.

You want to blow some minds? Of course, who doesn’t?!

Pair your user-level insights with qualitative insights from support requests/conversations, onboarding, surveys, and outreach conversations. These insights really give your data meaning and give you a clear path forward.

I want to consider user-level data when my goal is:

1. Improve Usability:

   • Report Type: User Session Replays

   • Data Points: Look for moments of pause or rapid clicking, signifying confusion or frustration. Note the steps leading to drop-offs or abandonment, particularly in critical flows like checkouts or sign-ups.

   • Application: Simplify or clarify UI elements that cause confusion. For example, if users hesitate a lot on a form with many fields, consider reducing the number of fields or adding tooltips for guidance.

2. Optimize Throughput of Key Flows:

   • Report Type: Funnel Analysis

   • Data Points: Identify the steps with the highest drop-off rates in your key workflows, such as onboarding or feature adoption.

   • Application: Streamline the step with the highest drop-off. If users are dropping off at a tutorial step, it might be too complex or not engaging enough. Consider breaking it down into simpler parts or making it more interactive.

3. Enhance Feature-Level Discovery and Adoption:

   • Report Type: Feature Usage and Discovery Funnels

   • Data Points: Look for features with low engagement levels compared to others. Also, pay attention to features that have high discoverability but low usage.

   • Application: For features with low discoverability, incorporate them more prominently in your UI or highlight them in user communications. If a feature is highly visible but rarely used, gather feedback to understand why and improve its value proposition.

4. Cross-Promote Value-Add Partnerships:

   • Report Type: User Activity Segmentation

   • Data Points: Segment users by feature usage and look for correlations that align with partner offerings. For example, users who frequently use project management tools might benefit from time-tracking software.

   • Application: Introduce relevant partner offerings through targeted emails or in-app messages. Use specific examples to show how the partnership adds value, such as "Users like you saved X hours a week with our new time-tracking integration."

5. Identify Segments Based on How They Use the Product:

   • Report Type: Behavioral Clustering

   • Data Points: Group users based on activity patterns, such as those who predominantly use collaboration features versus those who use planning tools.

   • Application: Customize your product development, marketing, and support initiatives to address each segment's specific needs and preferences. For the collaboration-heavy users, for instance, introduce enhancements or tips focused on teamwork and shared workflows.

   

By zeroing in on these specific reports and data points, you can take targeted actions that significantly improve user satisfaction and engagement with your product. Always validate your changes with A/B testing and continue to solicit user feedback to ensure your decisions are driving the desired outcomes.

User-level optimizations in the wild

• Improving User Onboarding: Slack is well known for their streamlined onboarding experience, but I find Linear my current “best in class” for new user onboarding.

Check out my video series of new user onboarding breakdowns

• Personalization of Features or Content: Spotify uses user data to curate playlists and recommend music, significantly enhancing user engagement. Check out how in-depth Spotify's personalized playlists can go.

• Testing New Features: Many companies and product teams use products like Launch Darkly to manage feature flags and test with different customer segments inside their products.

Account-Level Analysis

Account-level analysis provides a broader perspective, focusing on users' collective behavior and needs within an organization or group. This data helps you understand the value your product brings to the business, which is the determining factor in whether or not users will keep paying for it.

How to Think About It

Account-level analysis is key when making strategic decisions, understanding organizational behavior, or when decisions impact the client/customer as a whole rather than individual users.

Churn at the account level is more closely related to how people interact with (and within) your product than how active any individual is. Let’s use Figma as a great example here. They started with a B2C setup, trying to get in the hands of as many designers as possible. They always had “multiplayer” on the roadmap, but it was a technical hurdle that took them longer than anticipated to achieve.

Scenarios and Examples for Account-Level Analysis

• Identifying Upsell or Cross-Sell Opportunities: Grammarly does a good job of identifying when multiple people within the same organization are using the product and reaching out to communicate the value of their business plans.

• Predicting and Preventing Churn: You can look at trends in user activity within the account, usage of collaborative features within an account, or even the creation of new “objects” within an account to quantify churn risk. These are just examples of indicators, and your data science team should do the work to connect the dots in terms of finding a true correlation between changes in behavior and churn. A personal example from when I worked at WordPress.com would be that failure to renew a domain for a live site is usually a strong indicator of churn, while a reduction in editing and posting was more of a signal for the overall churn risk going up.

• Understanding Account Growth Potential: HubSpot uses account data to assess the potential for expansion, guiding strategic growth planning. Explore HubSpot’s Growth Strategy.

• Strategic Account Management: The big cloud providers (Google, Microsoft, and AWS) all rely on account managers to help ensure their customers' success and their accounts' growth. They work to understand what their customers are trying to accomplish and introduce them to related solutions within the ecosystem to grow the value of the account.

This probably wasn’t as “sexy” as my post about multi-modal AI product experiences, but I hope you find it useful. It’s important to find a balance between leveraging data and trusting your experience and product sense.

If you found your way here and read this entire post, you should probably go ahead and subscribe.