Comparison
EchoDepth vs Manual QA
for vulnerable customer identification.
Manual QA is a sampling process. FCA Consumer Duty requires proactive identification. These two things are not compatible — and the FCA knows it.
of interactions reviewed by
manual QA processes
of interactions assessed
by EchoDepth
How do you move from 2–3% QA sampling to 100% Consumer Duty coverage?
A contact centre handling 50,000 calls a month with five QA reviewers can assess about 1,500 of them — roughly 3%. The remaining 48,500 interactions are never reviewed for vulnerability signals, which is why sampling cannot evidence FCA Consumer Duty outcomes: the regulator's proactive identification requirement applies to every customer, not to a sample. Moving to 100% coverage means separating screening from review. Instead of reviewers scanning recordings to find the interactions that matter, an automated layer assesses every interaction against consistent vulnerability thresholds and flags the subset that warrants human attention — typically 1–2% of volume. The same five reviewers then spend their time on roughly 800 genuinely high-risk interactions rather than 1,500 randomly chosen ones, and every interaction carries an auditable record whether it was flagged or not. EchoDepth performs that screening layer for financial services firms, scoring facial Action Units, voice prosody and text emotion in real time.
Head to head
| Capability | Manual QA | EchoDepth |
|---|---|---|
| Interaction coverage | 2–3% (sample) | 100% |
| Vulnerability flag timing | Hours or days post-interaction | Real-time, during interaction |
| Detects suppressed distress | Rarely — reviewer sees surface | Yes — involuntary signal |
| Consistency | Variable — reviewer dependent | Consistent signal thresholds |
| FCA audit trail | Incomplete (97% unreviewed) | Full interaction-level log |
| Outcomes monitoring MI | Manual compilation | Automated, continuous |
| Scales with interaction volume | Requires proportional headcount | Linear API cost only |
| FCA proactive ID requirement | Does not satisfy | Satisfies |
EchoDepth doesn't replace compliance staff. It changes what they do.
Manual QA reviewers are skilled professionals. The problem isn't their capability — it's the coverage constraint. A team of five reviewers cannot listen to 50,000 calls a month. They review 1,500 and call it QA.
EchoDepth screens all 50,000. It flags the 800 that show vulnerability signals requiring review. Your five reviewers now spend their time on the interactions that matter — with full context from the signal data — rather than sampling randomly and hoping the vulnerable interactions are in the 3%.
The compliance team gets better, more targeted work. Coverage goes to 100%. The FCA gets a complete audit trail. The outcome is better for everyone — including the customers who were in the 97% that manual QA never reached.
Why is manual QA insufficient for FCA Consumer Duty vulnerable customer identification?
Manual QA typically reviews 2–3% of customer interactions, selected by sampling rather than systematic risk criteria. This means 97–98% of interactions are never reviewed for vulnerability signals. The FCA's Consumer Duty requirement for proactive identification applies across all interactions — a 2–3% QA sample cannot satisfy a 100% proactive identification requirement. EchoDepth analyses every interaction submitted, generating vulnerability flags in real time without human reviewer bandwidth as a constraint.