Why Intake Data Quality Eats Your Routing
Routing/follow-ups only work as well as the data that comes in. In law firms, the problem is rarely "not enough AI", but:
- Required fields missing,
- Phone/email inconsistent,
- Duplicates create chaos,
- Special cases land in wrong status.
Principle: Validation at the Source (Not "Clean Up Later")
If a field is important, it must be captured cleanly at the form – otherwise you pay later every time.
Minimal Required Fields (For Most Law Firms)
- Name
- Email or phone (at least one, validated)
- Concern category (dropdown, not free text)
- Deadline/urgency (if relevant)
Artifact: Validation Rules (Copy/Paste)
| Field | Rule | Error Message | Notes |
|---|---|---|---|
| RFC check + lowercase | "Please enter valid email" | Alias/plus ok | |
| Phone | E.164 normalize | "Please enter phone number" | Country code DACH |
| Category | Whitelist only | "Please select category" | Reduces free text |
| ZIP | 5-digit DE / 4 AT / 4 CH | "Please check ZIP" | Optional |
Dedup: As Simple as Possible, As Strict as Necessary
Goal: Recognize "same person/same case" without losing real new inquiries.
Dedup Strategy (Recommended)
- Primary key: Email (normalized)
- Fallback: Phone (normalized)
- Fallback: Name + ZIP (fuzzy)
Rule: If match → don't block, but route to a "Duplicate Review" status.
Edge Cases You Should Model Explicitly
- Client writes from different email
- Third party asks on behalf of client
- Multiple concerns in short time
- "Please urgent" without deadline date
Solution: Status model + owner decision (don't let AI guess).
KPI Block (So You See If It's Getting Better)
Validation error rate (form): Target < 5%
Duplicate review rate: Target < 10%
Routing corrections (manually rerouted): Target < 10%
Guide: Lead Intake Automation
Deep dive: Intake Form: 7 Fields That Really Help
Deep dive: Status Model in Law Firm Intake (2 Variants + SLAs)
Next Step
If you send me 3 example inquiries (anonymized) + your current form, I can tell you which 5–10 rules have the biggest impact.