Pinpointing the Onboarding Question That Stalled 40% of New Users
A leading European SaaS company had a high onboarding call drop-off rate but no tooling to analyze what happened inside those conversations. Sherlock found the bottleneck in hours.
40%
Fewer drop-offs
Real-time
Monitoring
2x
Activation rate
TL;DR — What Sherlock found
- 1
A leading European SaaS company had a high onboarding call drop-off rate but no tooling to analyze what happened inside those conversations.
- 2
Sherlock analyzed onboarding transcripts at scale, revealing that 40% of calls stalled at a product setup question users couldn't answer — consistently, across thousands of calls.
- 3
The team restructured the script based on Sherlock's evidence. Drop-offs fell 40% and the user activation rate doubled within two weeks — no engineering work required to reach the insight.
The Problem
New user onboarding calls had a high drop-off rate, but the team had no tooling to analyze conversation flow or identify where users got stuck.
The Process
Sherlock analyzed onboarding call transcripts at scale, revealing that 40% of calls stalled at a confusing product setup question that users couldn't answer.
The Solution
The team restructured the onboarding script based on Sherlock's insights. Drop-offs fell by 40% and the activation rate doubled within two weeks.
Results
- ✓40% fewer onboarding drop-offs
- ✓Real-time conversation flow alerts
- ✓2x user activation rate
Use Cases
Questions the team asked Sherlock
- SC“Where in the onboarding call are users dropping off?”
- SC“What question causes the most confusion during onboarding?”
- SC“Show me all onboarding calls where the user paused for more than 10 seconds”
- SC“Compare drop-off rates across this week's onboarding calls”
- SC“Which onboarding agents have the highest completion rates?”
Deep Dive
The full story
For a leading European SaaS company with an active voice-assisted onboarding program, a high call drop-off rate was a significant growth barrier. The team knew users were dropping off — they could see it in activation metrics — but without visibility into what was happening inside those conversations, every proposed fix was a guess.
Sherlock analyzed onboarding call transcripts at scale, mapping user responses and silence patterns against conversation stage. The finding was precise: across thousands of calls, 40% of users who didn't complete onboarding stalled at the same moment — when asked to locate and enter their product configuration key. Most users didn't have the key readily accessible, and the prompt didn't explain where to find it, leading to extended silence followed by disconnect.
The onboarding script was restructured based on Sherlock's transcript evidence — the configuration key step was moved later in the flow and preceded by guidance on where to find it. Within two weeks, onboarding drop-offs fell by 40% and the user activation rate doubled. The insight that drove this change was available in the existing call data; what was missing was a way to query it.
Real-time monitoring was the second outcome: with Sherlock's alerts configured, the team now receives notifications when drop-off rates at any conversation stage exceed normal thresholds — giving them the ability to detect script problems within hours rather than weeks, and respond before large numbers of users are affected.
FAQ
Frequently asked questions
How does Sherlock find conversation bottlenecks in onboarding calls?
- Sherlock transcribes all calls and maps drop-off events against conversation stages — identifying which prompt, question, or step has the highest abandonment rate. You can ask 'where in the onboarding call are users dropping off' from Slack and Sherlock returns a breakdown by conversation stage, with the highest-drop-off moments highlighted and example transcript excerpts attached.
What does transcript analysis look like at scale?
- At scale, transcript analysis means processing thousands of conversations simultaneously — not sampling. Sherlock indexes every transcript and makes it queryable in real time. When you ask a pattern-based question like 'what question causes the most confusion during onboarding,' Sherlock scans all transcripts for that pattern and returns an aggregated answer, not a manually reviewed sample.
How fast can Sherlock identify an onboarding bottleneck?
- Typically within minutes to hours, depending on the size of your call dataset. The initial query — 'show me drop-off rates by conversation stage' — returns a ranked list in seconds. Follow-up queries to examine specific moments in detail add a few minutes. The full investigation for the SaaS case study took less than a working day from first query to confirmed root cause.
How do I track activation rate changes after a script update?
- Sherlock tracks completion rates by conversation version or time period. After a script update, you can compare 'onboarding completion rates this week vs two weeks ago' in Slack and get a direct comparison with percentage change. Sherlock can also segment by agent, region, or other dimensions if your onboarding uses multiple configurations.
Are real-time alerts available for onboarding drop-off spikes?
- Yes. You can configure Sherlock to alert you when the drop-off rate at any specific conversation stage exceeds a threshold — for example, 'alert me if more than 20% of users drop off at the configuration key step in any hour.' Alerts are posted to Slack as soon as the condition is met, with the call data attached for immediate investigation.
What makes an onboarding script question likely to cause drop-offs?
- The most common causes are: asking users for information they don't have readily available (like account IDs or configuration keys), using technical jargon without explanation, presenting too many options simultaneously, and poorly timed questions that require the user to take an action before they're ready. Sherlock identifies these patterns by correlating silence durations, user responses, and call termination events across thousands of transcripts.
Ready to stop guessing?
Let Sherlock investigate your voice calls. Find failures, cut costs, and get answers — all from Slack.