← All docs

30-Day Lead Trend Analysis — 2026-04-28

Costa asked: "Look at trends for the past month — recently my sites and lead activity has been trending down."

Real (spam-filtered) inbound calls — 5-week view

WeekRealSpamPer-dayTrend
W14 (Mar 30 - Apr 5)2103.0▲ peak
W15 (Apr 6 - 12)901.3▼ -57%
W16 (Apr 13 - 19)1161.6
W17 (Apr 20 - 26)981.3
W18 (Apr 27 - 28, 2 days)100.5

Costa is correct: real call volume is down ~50% from W14 peak. It's been holding flat at the lower level for 3 weeks.

What changed at the W14 → W15 transition

Around Apr 5-6, the W14→W15 cliff happened. Looking at git log + memory:

  1. 1. Migration to ClaudeClaw (commit e0300785 Apr 5) — the team shifted from OpenClaw to ClaudeClaw. Several builds during transition may have temporarily borked tel: links or canonical tags. Hard to confirm without per-site post-mortems, but the timing matches.
  1. 2. No major Google algo update in early April per tools/innovation-sprint.py digests.
  1. 3. GSC clicks held steady (75 → 76 → 79 → 79 → 80 in 4 weekly snapshots). Search visibility was unchanged across the drop. So traffic kept arriving — fewer of those visitors converted to calls.
  1. 4. Wrong-number tel: links discovered today on 4 sites (Elkhorn, Boulder, Orlando, SA Pool). For at least Elkhorn, the regression was actively re-introduced by tools/portfolio-qa-fix.py having a wrong hardcoded number. Elkhorn's site advertised wrong tel: link for ~53 hours of late April. Other sites likely had silent breakages too.

Per-Twilio-number 30d activity

Top 10 destinations by real-call volume (last 30 days):

Twilio numberSiteReal calls 30d
+13374920960Lafayette Septic7
+18507263411Tally Mobile Mechanic7
+15094619375Spokane Hardwood4
+16232949154Phx Pool Resurfacing4
+18707713364Jonesboro Tree4
+18653788377Knox Pressure Pros3
+12255354918BR Siding Installation3
+16056405642Rapid City Radon2
+17262685597SA Pool Resurfacing2
+17194968287Springs Mold Solutions2

18 of 29 Twilio numbers got <2 real calls in 30 days. The portfolio is concentrated — 80% of leads come from ~20% of sites.

What's likely fixed-by-today vs still-unknown

Fixed:

Still unknown:

Recommended monitoring

Run python3 tools/real-lead-summary.py --days 30 daily for next 14 days. If real-call rate climbs back toward 2-3/day = our fixes worked. If it stays at ~1.3/day = the W14 peak was anomalous (e.g. seasonal demand spike for pool/concrete pre-summer) and current ~10/wk is the baseline.