Browse the book & resources

Supporting material

Desk-researched interview 1: The customer outcome waiting for release

Desk research, not original interviews. These question sets synthesize published accounts; no practitioner was interviewed for this book.

Question: Which customer outcome was last delayed despite engineers feeling busy? Where did the work actually wait?

Method: This is a desk-researched interview with published engineering accounts, not a conversation with Spotify or Etsy. Research snapshot: November 2026. The accounts describe delivery constraints; none measures how busy individual engineers felt.

What the published accounts say

  • Spotify engineers Jacob Vesterlund and Katie Walker describe instrumenting the main app's release process every five minutes. The largest measured time sinks were testing/fixing release-blocking bugs and waiting for App Store approval. Manual advancement could add up to 12 hours when the previous stage finished outside working hours. Their release-conductor automation reduced the average release cycle by about eight hours. These are release-cycle figures, not a measured increase in customer adoption. Spotify Engineering, 2026.
  • The release still has necessary gates: regression sign-off, build verification, crash thresholds and app-store review. Some waiting is a safety control or external dependency, not waste to eliminate. The published account does not identify the customer feature whose arrival was last delayed. Spotify Engineering.
  • Thoughtworks and Etsy report that product teams previously took about 50 days to validate an idea with customers; pilot teams reduced that “time to learning” to five days after changes to prototyping and product delivery. Etsy CTO Mike Fisher cautions that it is hard to isolate a process tweak across hundreds of engineers. This is an example of customer learning delayed by the product discovery/delivery system, not evidence of a particular customer's delayed purchase or of AI causing the improvement. Thoughtworks/Etsy case study.
  • A separate, named customer-impact example appears in Spotify's June 24, 2026 incident report: video podcast episodes that normally became available within minutes took hours to publish, and creators reported missing episodes. A saturated transcoding queue, competing batch work, increased per-episode processing cost and a scheduler bug combined to delay publication. This was a production-capacity incident, not the app-release wait described above. Spotify, Content Ingestion & Podcast Video Incident Report.

Synthesis, not testimony

A plausible answer is: a customer's access to a working release, or a team's ability to validate a customer need, can wait after implementation appears finished. Spotify's app-release account documents release gates, external approval and manually triggered transitions; Etsy describes a longer path to customer validation. The June 24 Spotify podcast incident instead identifies a specific creator outcome delayed by processing and detection queues. These are distinct workflows, not a single incident. The primary outcomes at risk are effective (value or learning reaches the customer) and efficient (elapsed time); reliable release checks must remain intact.

To find the last delayed outcome in a real team, follow one named customer commitment backward from delivery or validation. Timestamp ready-for-review, test start/end, sign-offs, submission, approval, rollout and first customer use. Separate active fixing, avoidable queue time and necessary safety/external waits; ask customers or product managers what was missed rather than inferring impact from a release timestamp.

Questions to ask a practitioner

  1. Name the last customer outcome that missed its promised date. Who was affected, and by how much?
  2. On that change's timeline, how long was it actively worked versus waiting, and who owned each handoff?
  3. Which gate protected users, which wait was external, and which transition could safely be automated?

Follow-up desk-researched responses

1. Last missed customer outcome, who and how much? The closest documented case is video podcast publication on June 24, 2026: creators' episodes reached listeners hours later than the normal minutes. Spotify does not report the number of affected creators/listeners, an individual episode's exact delay, or a promised contractual deadline; “last” is not established across all of Spotify's work. Incident report.

2. Active work versus waiting; owner of handoffs? The public timeline records first internal alerts at 13:30 UTC, a batch job stopped at 16:35, incident response at 17:34, a scheduler fix deployed at 20:49, extra capacity online at 00:14 the next day, and queues cleared at 01:02. The episodes waited in the transcoding queue; Spotify itself identifies a roughly four-hour recognition gap between first alerts and formal response. It does not provide per-episode timestamps or engineer work logs, so active versus waiting hours and responsibility for each handoff cannot be computed. Incident report.

3. Protective gate, external wait, automatable transition? In the different app-release workflow, regression tests, crash thresholds and build verification protect users; App Store review is external; Spotify automated previously manual, condition-checked state transitions and reports about eight hours off the average release cycle. None of those facts establishes which release gate held up a particular customer's feature. Release-process account.

Evidence gap: The incident gives a real delayed creator outcome and timestamps, but not a missed customer promise, engineer utilization, individual queue dwell times or the last delay across Spotify. Eight hours of faster app release is not eight hours of measured customer benefit.