What is Resolvability?
When triaging errors in Noibu, product owners consider two data points: How impactful is the error, and how much developer effort will it take to resolve the error? This second point is difficult to estimate. A developer can spend hours investigating an error, only to discover the problem has no easy fix or can’t be fixed at all.
We want your time in Noibu to be as productive and impactful as possible, and we're taking steps toward that goal with a new prioritization metric: Resolvability. We've developed a predictive machine learning (ML) model that examines our historical data and predicts whether an issue can be resolved based on what has been fixed in the past. If an issue has similar characteristics to issues other Noibu users have Closed-Fixed, it will likely be resolved again. This gives you another aspect to consider when triaging and prioritizing issues.
We've added a column to the Funnel and All Issues tables for resolvability insights. A green wrench next to an issue's title means Noibu has historical data of clients resolving similar issues.
Keep in mind that these are only predictions based on previous data. If an issue does not have a green wrench, it does not mean it's impossible to resolve. It just means we don't have sufficient data to make that claim.
This is just the beginning. Like any ML model, our predictive model learns over time as users resolve more and more issues. Every time you resolve an issue and mark it as Closed-Fixed in Noibu, the model improves, and predictions become more nuanced.
Filtering & Sorting
We’ve added a new option to the filters menu to filter the All Issues or Funnel table to only show resolvable issues. When this filter is applied and the page is sorted, issues are listed by highest-to-lowest resolvability. This means issues at the top of the list have the most data to support their resolvability.
How is Resolvability Calculated?
First, we collected training data by examining issues marked as Closed-Fixed, which became our ML model's positive class. We also examined issues that were opened but never fixed–the model's negative class. We used this training data to build a model that maximizes the precision of predicting Closed-Fixed issues.
How does Resolvability Help Me?
Resolvability offers another metric to consider while triaging issues. For example, you might prioritize an issue that meets the following criteria:
- The issue has a high Annualized Revenue Loss
- The issue includes a Stacktrace
- The issue is a First-Party error
- The issue is marked resolvable–it is marked with a green wrench
Check out the video below for a walkthrough of Noibu's Resolvability feature!