Latest / The Bootstrapped Tech Founder with Fexingo: Profitable Software Companies Without VC / Bootstrapped Founders Use Customer Health Scores
Transcript
- Lucas: If you run a bootstrapped SaaS company, you probably know the feeling of losing a customer you thought was happy. They paid on time, they never complained, and then one day they cancel. The data was telling a different story — you just weren't looking at it the right way. Luna: That's the classic silent churn. And in a bootstrapped business, every customer matters so much more than it does for a vc backed company that can buy growth. Lucas: Exactly. So today I want to talk about customer health scores — a way to systematically predict churn before it happens. Not with a data science team, not with machine learning, but with a simple composite score you can build in a spreadsheet. Luna: I love this because it's a concept that sounds complex but is actually pretty straightforward. Let's start with the basics — what goes into a health score? Lucas: Right. The key is to pick three to five leading indicators that correlate with retention in your specific product. The most common ones are login frequency, feature adoption, support ticket sentiment, and payment consistency. Luna: Login frequency is usually the easiest to track. If a customer who used to log in five times a week drops to once a week, that's a red flag. Lucas: Exactly. But you need to normalize it — a power user might log in ten times a day, while a casual user might log in twice a week. The health score should track deviation from that user's own baseline, not an absolute number. Luna: That makes sense. So for each customer, you calculate a 'login health' subscore based on how much their current usage differs from their historical average. Lucas: Right. Then feature adoption — are they using the core features that drive stickiness? If you have a reporting tool that reduces churn by 40 percent, you want to know which customers haven't tried it yet. Luna: And support ticket sentiment — that's the qualitative piece. I've seen companies tag tickets as positive, neutral, or negative, and then track the ratio over time. Lucas: Yes. A customer who opens a ticket with 'this is broken' language is different from one who asks 'how do I do this?' One signals frustration, the other signals engagement. Sentiment analysis can be done manually with a simple rubric. Luna: Payment consistency is the last one — are they paying on time, or are they hitting payment failures? Involuntary churn from a declined credit card is often a sign they're not engaged enough to update their payment info. Lucas: Exactly. So you take these four signals, score each one from one to five, and then combine them into a composite health score. The weighting matters — for most B2B SaaS, feature adoption and login frequency are stronger predictors than support sentiment. Luna: How do you determine the weights? Is there a standard formula? Lucas: You can start with equal weights and then adjust based on your own data. Look at the customers who churned last quarter — what pattern did their scores show? If login frequency dropped early but feature adoption stayed flat, then login frequency might deserve a higher weight. Luna: So it's iterative. You build a version, test it against historical churn, and refine. Lucas: Exactly. And once you have a working model, you set thresholds. A score above 4.0 means green — healthy. Between 3.0 and 4.0 is yellow — monitor. Below 3.0 is red — immediate outreach. Luna: What does that outreach look like? I imagine a red score triggers a phone call, not just an automated email. Lucas: Absolutely. For a bootstrapped company, you can't afford to lose a red-score customer to automation. The founder or a customer success person calls them, asks what's going on, and offers help. Sometimes it's as simple as a feature they didn't know existed. Luna: I've seen a bootstrapped project management tool that used health scores to reduce churn by 25 percent in six months. They had a single metric — days since last login — and when it hit 14 days, the founder sent a personal video walkthrough. Lucas: That's a great example. Simple, human, and directly tied to the data. And it doesn't require any fancy tech — just a script that checks the database and sends a Slack notification. Luna: What are the common mistakes founders make when they first try this? Lucas: The biggest one is over-complicating the model. I've seen people try to include 20 different metrics, and then they can't figure out why a score dropped. Three to five well-chosen metrics outperform a dozen noisy ones every time. Luna: Another mistake is acting on false positives — customers whose score drops temporarily because of a holiday or a product update that changed their workflow. Lucas: Right. That's why you want to look at trends over a two-week or one-month window, not a single day. A one-day dip might be noise, but a consistent downward trend over two weeks is real. Luna: And then there's the opposite problem — ignoring yellow-zone customers because they're not red yet. But by the time they turn red, it's often too late. Lucas: Exactly. Yellow is the sweet spot for proactive outreach. You send a check-in email, offer a free training session, or share a case study that shows how another customer got more value. That can prevent the slide into red. Luna: So if someone listening wants to implement this tomorrow, what's the minimum viable health score? Lucas: Start with two metrics: login frequency and a single core feature adoption. If your product has a key action — like creating a report or sending an invoice — track whether they've done it in the last 30 days. Score each from one to five, average them, and start calling the bottom third. Luna: That's simple enough to set up in a day. And it's a huge step up from the gut-feel approach most bootstrapped founders rely on. Lucas: You know, this kind of practical, data-informed approach is exactly why we do these shows. We deliberately keep them ad-free so we can focus on what actually works for operators and builders, not on selling you something. If you find value in that, you can support the show at buy me a coffee dot com slash fexingo. Luna: Yeah, it's a small way to keep this independent and focused on real tactics instead of sponsored content. We appreciate everyone who chips in. Lucas: Alright, back to health scores. Once you have the two-metric version working, you can layer in support sentiment by looking at the ratio of positive to negative tickets. Most help desk tools let you tag tickets with a simple dropdown. Luna: And payment data — you get that from your billing system. Stripe, for example, has webhooks for failed payments. Combine that with the login data and you've got a pretty robust early warning system. Lucas: The beauty is that all of this data is already in your database. You're not building a data warehouse, you're just writing a couple of SQL queries or using a tool like Metabase to create a dashboard. Luna: One thing I want to emphasize — health scores should be used to help customers, not to punish them. It's not about churning low-score customers, it's about re-engaging them. Lucas: Exactly. The goal is to increase net revenue retention — keeping the customers you have and growing their lifetime value. A customer who goes from red to green is worth more than a new customer you acquire at high cost. Luna: Speaking of cost, bootstrapped founders often worry about the time investment of manual outreach. But a health score system actually saves time because you prioritize the right people. Lucas: Right. Instead of randomly calling customers or only handling complaints, you have a systematic process. You know exactly who needs attention this week. That's the operational efficiency that makes bootstrapped companies profitable. Luna: So to wrap up, the key takeaways: pick three to five leading indicators, weight them based on your historical data, set green yellow red thresholds, and take action on yellow before red. Lucas: And iterate. Your first model won't be perfect. But a 70 percent accurate model that you actually use is infinitely better than a perfect model you never build. Start simple, learn from the data, and refine over time. Luna: I think this is one of those 'drop everything and implement' ideas. If you're a bootstrapped founder, you can build this tomorrow and start seeing results in weeks. Lucas: Totally agree. Next time we'll talk about how to use health scores not just for retention, but for upsells and expansions — because a healthy customer is also your best candidate for a higher plan. Luna: Looking forward to it. Until then, keep building.