Tinker Observability & error analysis

by Kondasamy Jayaraman · drawn with Colorfigures by April Zhu

Bins, front left to rightSearch missed the passage: 6 · Answered from an old passage: 5 · Looked up the wrong order: 3 · Ignored what the tool returned: 2 · Promised a refund it can’t give: 2

Chapter 5 of 7

Which ones matter

Five failure modes, counted. The team can fix one this week.

This chapter: Which failure mode should Larkspur fix first?

You will learn

  • Why the most common failure isn’t always the first to fix
  • Weighing how often against how much harm
  • Picking one fix, with a reason
Words you’ll meet (1)
Harm
How much a failure costs the customer or the business when it happens. Here, the team’s rating from 1 to 3.

The picture on the left is live. Each step below asks you to do one thing in it; the story continues here as you go.

Step 1 of 3

The team can fix one failure mode this week. The table counts each mode over yesterday’s 18 failed traces, and gives the team’s rating of how much harm one failure does, from 1 to 3, with the reason. Before you choose:

Yesterday’s failure modes
ModeCountHarm
Search missed the passageThe customer gets no answer and asks again or calls; annoying, but nothing they act on is wrong.61
Answered from an old passageA confident, cited, wrong answer: customers act on last year’s prices and rules.53
Looked up the wrong orderA wrong date or status, and sometimes another customer’s order details.32
Ignored what the tool returnedA wrong delivery status, though the right one was in hand.22
Promised a refund it can’t giveA promise Larkspur must break or pay for, to its biggest accounts.23
“Search missed the passage” happened 6 times, more than any other mode. Is it the one to fix first?