The hidden denominator problem
Digitization is a board-level KPI in banking, healthcare, insurance, and other regulated sectors, but most dashboards still measure only the people who already made it into digital channels. That creates a blind spot: the digital identity gap.
In the source article, that gap means customers who exist on file but never activated or used an online login. The scale is not small. FDIC data says about one third of banked households did not use online banking in 2023. In healthcare, HINTS/JMIR 2025 reports 38.7% of US adults did not access a patient portal in the last 12 months.
This matters because a nice-looking adoption chart can still hide a 10–40% segment that never signed up, never came back, or keeps failing before the backend sees anything.
Why backend login metrics miss the real problem
The most uncomfortable point in the article is simple: over 80% of sign-up and login failures happen client-side and never reach the backend IdP.
That means your server logs can report a healthy success rate while users are dropping off because of:
- missing email verification
- SMS OTP delivery issues
- browser or OS-specific WebAuthn problems
- timed-out prompts
- blocked popups
- password managers fighting the form
For identity teams, this is the difference between seeing request outcomes and seeing the actual user journey. If your telemetry starts at the API boundary, you are blind to most of the funnel.
Not every failure is the same failure
One of the better distinctions in the article is that teams often collapse three different problems into one “adoption” metric:
| Problem type | What it looks like | What usually fixes it |
|---|---|---|
| Conversion problem | User logs in with SMS OTP but never upgrades to passkey | Better upgrade prompts, timing, messaging |
| Funnel problem | User starts sign-up and drops at step three | Better onboarding UX, recovery, delivery reliability |
| Identity problem | Customer has no online profile at all | Assisted enrollment, supervised onboarding |
If those are mixed together, roadmap decisions get distorted. A passkey adoption campaign will not help the customer who never completed sign-up. A better signup form will not help the user who refuses biometric binding and needs a hardware key or PIN-unlocked device credential.
The metrics that actually expose the gap
The article argues for authentication telemetry as a client-plus-server event layer that captures every sign-up and login step, including failures that never hit the backend. That is what makes the gap measurable.
A practical starter set of metrics includes:
- Sign-up completion rate by device and browser
- Login success rate (LSR)
- Authentication error rate (AER) by reason code
- Authentication drop-off rate
- Reach rate by cohort, measured against the full customer base
- Time-to-first-authenticated-action
The key idea is segmentation. An overall sign-up completion rate may look fine, while one browser version or one branch-acquired cohort is failing badly. The article even notes that once enterprises instrument client-side events, reported success rates often turn out to overstate reality by 10–25 percentage points.
Why this is becoming a product-reach issue
This is not only about support costs or password resets. The article connects the digital identity gap to agentic AI digital identity and authentication, embedded finance, shrinking branch footprints, and self-service regulation such as PSD3 and eIDAS 2.0.
Those models assume the customer can authenticate digitally. If they cannot sign up, cannot recover, or cannot use a smartphone in their work context, they are unreachable by the product itself.
That is why the remediation is broader than passkey rollout alone. The article recommends:
- instrument the login funnel first
- fix client-side failure causes before optimizing backend success rates
- match credentials to segment, not ideology
- use supervised onboarding and cross-device credential provisioning for customers who cannot complete enrollment alone
Corbado is a passkey observability and adoption platform for large B2C enterprises.
The big takeaway: if your denominator starts with “active users,” your digitization KPI is already biased. Read the full breakdown.
Top comments (0)