`REPORT-coverage.md`

47 lines
# RENTV CRE dataset — coverage / quality report (2026-08-16)

Analysis only. No emails sent or drafted — RENTV sending is permanently banned
(see `outreach/SEND_BLOCKED`). Source files: `rentv_CRE_worklist.csv`,
`rentv_outreach_READY_verified.csv`, `rentv_outreach_VERIFY_FIRST_inferred.csv`.

## Headline
- CRE worklist: **4,205** contacts.
- Reachability: **92.0%** have an email (3,870), **62.9%** have a phone (2,645).
- Email provenance: **2,839 verified**, 1,026 inferred, 335 none, 5 updated.
- CRE confidence: **3,971 HIGH**, 234 PROBABLE.
- Send-ready split: 2,636 verified vs 977 verify-first (inferred, unsent).

## Warmth
| Bucket | Count |
|---|---|
| 80–100 hot | 208 |
| 60–79 warm | 1,194 |
| 40–59 mild | 2,419 |
| <40 cool | 384 |

## Top firms (verified contacts)
CBRE | Colliers 59 · JLL 57 · CBRE 56 · Newmark 49 · Kidder Mathews 46 ·
Orion 42 · Marcus & Millichap 38 · Matthews 36 · NAI Horizon 32 · SRS 20.

## Top verified email domains
cbre.com 120 · gmail.com 100 · kidder.com 55 · jll.com 54 · nmrk.com 48 ·
matthews.com 44 · marcusmillichap.com 42 · orionprop.com 42.

## ⚠ Data-quality findings (non-send, fixable)
1. ✅ **RESOLVED — pipe-merged company names.** The "CBRE | Colliers" bad merge (59
   contacts) plus the entire pipe-merge bug class were fixed across all 4 worklist
   CSVs in five passes (commits e70a1e4, 2628508, c9a6490, a4557d6, dbbef8f):
   Class 1 cross-firm merges (120 rows), Class 2 `| EMAIL` parse artifacts (370 rows),
   Class 3a same-firm case/typo twins (253 rows), Class 3b evidence-clear variants
   (144 rows), Class 3c domain-clear reserved merges (78 rows). Only intentional
   pipes remain (SVN franchise office names, the C&W|PICOR affiliate, one gmail
   contact whose employer is unknowable).
2. ✅ **RESOLVED — personal-domain emails tagged.** 157 verified emails on personal
   domains (108 gmail, 12 yahoo, 12 cox, plus aol/hotmail/outlook/etc.) are legitimate
   individual agents. Added a non-destructive `Domain Type` column to
   `rentv_CRE_worklist.csv` (personal / firm / blank) so firm-level and domain-level
   rollups can exclude them (commit b6c0043). Tally: 157 personal, 3,713 firm, 335 no-email.
3. **335 CRE contacts (8%) have no email** and **37% have no phone** — the residual
   unreachable tail; expected given the held/obscure firms. Not fixable (no source
   data); documented as-is.