Why North Star metrics often get you lost

A North Star is a singular guiding light that aligns teams, simplifies decision-making, and keeps everyone focused on delivering value. Managers love them. They gained traction from companies like Airbnb (nights booked), Facebook (monthly active users), and Spotify (time spent listening). The idea is simple, instead of drowning in data you have a single, carefully chosen metric keeps teams aligned on what matters most. Here’s the catch, a single metric with unclear assumptions can distort reality.

Context is king

Numbers, like language, only make sense within the context of the assumptions behind them. A single metric, just like a single word, can mean vastly different things depending on how it’s used. North Stars often oversimplify complex systems. Customer behaviour is rarely dictated by a single variable. Relying on a single metric often hides the truth rather than revealing it. I encountered this in a recent role when analysing customer contact rate, a key metric in our customer support experience. Our North Star for support performance was simple:

📈 Contact rate = total contacts / total orders placed

If contact rate is rising, it suggests an increase in customer issues. If it’s falling, things must be improving, right? Well, no. For weeks, our contact rate was climbing. In November, it had surged to over 94% (a 26% increase). Alarms went off and crisis meetings were called. It was assumed something was going badly wrong in the customers experience.

When we dug deeper, we realised something was off. Our contact rate assumed that contacts were directly correlated to orders but order volume fluctuated week-to-week. If order volume went up our contact rate would drop. If order volume dropped, then contact rate would rise. Our North Star could go up or down regardless of any material change in the customer experience.

Unravelling assumptions

The assumption in the metric was that the number of placed orders was a reliable denominator, an anchor that would give us a meaningful ratio. To test this assumption we averaged out orders to normalise the denominator. Quickly a different story emerged:

• At the beginning of November the contact rate seemed to be rising sharply but the real issue was that orders had dropped. A lower number of orders artificially inflated the contact rate. It made it look like customer issues were increasing when, in reality, contact volumes remained stable.
• At the beginning of October, it looked like contact rate has been on the decline, but in reality, it was rising. The reason we missed it? Orders increased, which diluted the contact rate. It made it appear like things were improving when the proportional number of support contacts had actually risen significantly.
• At the end of October, we saw the largest increase in absolute contact volume, but it was masked by a simultaneous increase in orders. This meant that we missed an important shift in customer behaviour simply because our metric was anchored to orders.

The timing of contacts also challenged our assumptions. Only 40% of contacts occurred within a week of the referenced order, customer inquiries usually happened two weeks after the order was delivered. This further broke the assumed link between orders and support volume.

We had adopted an e-commerce metric for a subscription business. Most of our contacts weren’t about the order itself, but about the consumption of the product after it had been delivered. The pattern went deeper, contacts weren’t evenly distributed across the customer base. A small percentage of customers were responsible for multiple contacts, meaning our overall contact rate was being skewed by a vocal minority. By questioning the assumptions baked into our North Star metric, we realised the metric was an illusion, one that made us think we understood customer behaviour when, in reality, we didn’t.

Layer your metrics

North Star Metrics are a useful tool, but they can’t stand alone. They must be paired with supporting metrics to ensure they reflect reality. They are a signal, not a conclusion, often noisy and influenced by multiple inputs. Just because an industry-standard metric works for others doesn’t mean it works for you. In our case, the assumed correlation between orders and contacts didn’t hold up.

The goal of a North Star Metric is to provide focus, not to serve as an unquestionable truth. A metric without context is just a number. A metric with context is a story. If you only follow one number, you might think you’re headed north, when in reality you’re lost.