Collecting KYC/Docs in Chat

November 07, 2025•4 min read

Collecting KYC/Docs in Chat

POPIA-aware upload flows, validation, and secure handover to your CRM.

When customers are ready to buy, asking them to find a scanner or fill a long form is a conversion killer. The fix is KYC in chat: collect IDs, proof of address, or company docs right in WhatsApp or your website widget, validate them automatically, and hand them off to your CRM with a clean audit trail. Done right, it’s as convenient as chat—and as safe as back-office.

Below is a practical blueprint we deploy using your WhatsApp stack, AI WhatsApp Chatbots, and InOne CRM.

The KYC-in-Chat flow (copy this pattern)

Start in the right lane
Add a high-intent launcher on pricing or application pages with WhatsApp Website Integration. Prefill the first message (“Start verification”) so the bot knows to open the KYC flow.
Consent + purpose (POPIA basics)
The bot clearly states what’s needed and why, links to policy, and records opt-in + timestamp. Customers can opt out or switch to email at any time (logged to the contact).
Smart requests, not a laundry list
Ask for exactly what’s required—e.g., ID front, ID back, and utility bill. Use Quick Reply chips so users pick the document type before uploading; this tags metadata for the CRM.
Inline validation
As each file lands, the bot checks:
- Type/size (JPG/PNG/PDF under your limits)
- Legibility & crop (face/fields visible, glare detection)
- Expiry & recency (ID not expired; proof of address < 3 months)
- Duplicate pages (catch re-uploads early)
  If something fails, the user gets an instant, friendly fix (“Your bill photo is a bit dark—try a brighter shot or send the PDF”).
Optional redaction & field capture
For bank statements or utility bills, the bot can auto-redact irrelevant lines before handover and OCR key fields (name, address, account) to pre-fill the CRM record.
Secure handover to CRM
Files and structured fields save directly to the contact/deal in InOne CRM with versioning, uploader, and time. Staff get a tidy checklist: ✅ ID, ✅ POA, ✅ Company reg, ❌ Missing signature.
Outcome + next step
If everything passes, the bot confirms “Docs received” and offers Book a call or Pay deposit. If a manual review is needed, it sets expectations and opens a support ticket with warm context so humans reply inside the same thread.

Guardrails that protect trust (and your number)

Data minimisation: Only request what’s necessary for the transaction. Avoid collecting sensitive data in free text.
Template discipline: Outside the 24-hour window, use approved WhatsApp templates for reminders (“Please upload proof of address”) with a clear opt-out. See the FAQ for policy basics.
Access control: Agents see thumbnails and extracted fields in InOne CRM; raw files follow your role permissions.
Audit trail: Every message, file hash, decision (pass/fail), and human action gets a timestamp—painless audits later.

Interesting AI fact: Lightweight models now score document quality (blur, glare, crop) and extracted-field confidence in real time. If confidence is low, the bot asks for a better image before humans ever see it—saving time and back-and-forth.

Human-in-the-loop (without the “stuck with a bot” feeling)

Escalate when any of these fire: low OCR confidence, mismatched names, expired IDs, or user frustration. The hand-off carries the last 5–10 turns, thumbnails, and a mini-summary so a person can reply with context in seconds. If the customer prefers email for the final signature, the agent can send the link while keeping WhatsApp as the status lane.

UX copy that reduces drop-off

“We’ll only use these documents to verify your account. You can stop at any time.”
“Crisp photo tip: place the ID on a dark table and avoid flash glare.”
“Prefer a PDF? Just attach it here—no need to scan.”
“All set. Want to book a quick call or go straight to the quote?”

Short, helpful lines beat legal-ese and keep momentum high.

What to measure (and improve weekly)

Track these in your CRM dashboards:

Start → complete rate for KYC flows
Average time to complete (by segment/channel)
Auto-pass vs manual-review ratio
Re-upload rate (signals clarity or quality issues)
Revenue per verified conversation (the north-star)

Why partner with us (not DIY)

Collecting docs in chat is easy to demo and hard to run at scale. The pitfalls: template rejections, messy file storage, missing consent logs, and frustrated users who get stuck in loops. We design the flows, validation, guardrails, and hand-offs end-to-end so your team gets clean docs and clean data—without the plumbing headaches.

Useful links to implement the stack: