Signal 3: Data-per-Device Compression — Feb 2026
Report date: 2026-03-11
Recent period (MA3): Nov 2025–Jan 2026 (3-month avg
BPD)
Long-term period (MA9): May 2025–Jan 2026 (9-month avg
BPD)
Alert threshold: BPD decline >= 25% (MA3 vs
MA9)
Scope: Top 30 accounts by revenue with >= 50 avg
daily devices
Signal Guide
BPD (MB/dev/day) = total data bytes / days in month / active devices / 1,048,576
A declining BPD means each device is sending/receiving less data than it historically has. This suggests reduced platform engagement — devices may be scaling back telemetry, switching to lighter protocols, or entering maintenance-only mode.
Method: Compare the recent 3-month average BPD against the 9-month long-term average BPD. The 3-month window is a subset of the 9-month window (overlapping MA), which naturally smooths firmware update spikes and seasonal variation.
| Severity | BPD Decline | Meaning | Recommended Action |
|---|---|---|---|
| 🔴 Severe | >= 50% decline | Data per device halved or worse | Escalate — investigate root cause with account team |
| 🟠 Moderate | 25–49% decline | Significant per-device data reduction | Monitor — verify if planned or organic |
| 🟢 OK | < 25% decline | Within normal range | No action needed |
Executive Summary
🚨 5 of 30 qualifying Top-30 accounts show BPD compression of >= 25% (MA3 vs MA9). 🟠 5 Moderate (25–49% decline) — monitor and investigate. ℹ️ 1 account excluded after spike analysis (AVAS) — BPD decline was driven by data usage spikes.
Flagged Accounts — BPD Compression
| Sev | Account | Revenue | Devices MA9 | Devices MA3 | Dev Chg | BPD MA9 (MB/d) | BPD MA3 (MB/d) | BPD Chg |
|---|---|---|---|---|---|---|---|---|
| 🟠 | Deltatrak, Inc † | USD 35,153 | 11,060 | 13,180 | +19.2% | 0.0403 | 0.0259 | -35.7% |
| 🟠 | Vivint Solar | USD 19,703 | 922 | 816 | -11.5% | 0.4390 | 0.2425 | -44.8% |
| 🟠 | LiveFree Emergency Response, Inc. † | USD 89,415 | 16,512 | 16,499 | -0.1% | 0.0257 | 0.0192 | -25.2% |
| 🟠 | PeopleNet Communications Corp | USD 1,083,518 | 270,350 | 243,427 | -10.0% | 21.23 | 14.27 | -32.8% |
| 🟠 | Engie Mobisol GmbH | USD 21,821 | 94,552 | 94,121 | -0.5% | 0.0245 | 0.0172 | -29.8% |
BPD = MB per device per day (daily average). Source:
daily_account_network_summary.
MA3 = 3-month recent average. MA9 = 9-month long-term average. MA3
is a subset of MA9.
Dev Chg shows device count trend for context — BPD Chg is the
primary signal. † BPD MA9, MA3, and Chg adjusted after removing
spike months. See Spike Analysis section below.
Action guide: 🔴 Severe — escalate to account team for root cause analysis.
🟠 Moderate — investigate whether compression is intentional (e.g., protocol optimization) or unintentional.
Spike Analysis Detail
Flagged accounts are checked for data usage spikes in the MA(9) window. A spike is a month where BPD exceeds 2.0× the account’s median BPD. When spikes are found, the MA(9) and MA(3) are recalculated with spike months excluded to determine whether the compression is genuine or spike-driven.
Deltatrak, Inc (adjusted, still flagged)
Median BPD: 0.0522 MB/d | Original: 🔴 -88.4% | Adjusted: 🟠 -35.7% | Spikes: 3
| Month | Devices | BPD (MB/d) | vs Median | Spike | Window |
|---|---|---|---|---|---|
| May 2025 | 6,240 | 1.39 | 26.6× | 🔺 Major | MA9 only |
| Jun 2025 | 7,804 | 0.1501 | 2.9× | ⚠️ Spike | MA9 only |
| Jul 2025 | 8,795 | 0.2383 | 4.6× | 🔺 Major | MA9 only |
| Aug 2025 | 11,268 | 0.0522 | 1.0× | MA9 only | |
| Sep 2025 | 12,418 | 0.0651 | 1.2× | MA9 only | |
| Oct 2025 | 13,478 | 0.0467 | 0.9× | MA9 only | |
| Nov 2025 | 13,229 | 0.0231 | 0.4× | MA3 + MA9 | |
| Dec 2025 | 12,128 | 0.0260 | 0.5× | MA3 + MA9 | |
| Jan 2026 | 14,182 | 0.0286 | 0.5× | MA3 + MA9 |
Verdict: Still flagged after adjustment. Removing 3 spike month(s) reduces BPD change from -88.4% to -35.7% (🔴 → 🟠).
LiveFree Emergency Response, Inc. (adjusted, still flagged)
Median BPD: 0.0274 MB/d | Original: 🟠 -35.1% | Adjusted: 🟠 -25.2% | Spikes: 1
| Month | Devices | BPD (MB/d) | vs Median | Spike | Window |
|---|---|---|---|---|---|
| May 2025 | 17,246 | 0.0225 | 0.8× | MA9 only | |
| Jun 2025 | 16,516 | 0.0277 | 1.0× | MA9 only | |
| Jul 2025 | 16,400 | 0.0274 | 1.0× | MA9 only | |
| Aug 2025 | 15,911 | 0.0315 | 1.1× | MA9 only | |
| Sep 2025 | 16,209 | 0.0609 | 2.2× | ⚠️ Spike | MA9 only |
| Oct 2025 | 16,832 | 0.0388 | 1.4× | MA9 only | |
| Nov 2025 | 17,068 | 0.0198 | 0.7× | MA3 + MA9 | |
| Dec 2025 | 16,476 | 0.0194 | 0.7× | MA3 + MA9 | |
| Jan 2026 | 15,952 | 0.0185 | 0.7× | MA3 + MA9 |
Verdict: Still flagged after adjustment. Removing 1 spike month(s) reduces BPD change from -35.1% to -25.2% (🟠 → 🟠).
AVAS (excluded — false positive)
Median BPD: 0.0088 MB/d | Original: 🟠 -44.6% | Adjusted: 🟢 -6.1% | Spikes: 2
| Month | Devices | BPD (MB/d) | vs Median | Spike | Window |
|---|---|---|---|---|---|
| May 2025 | 40,716 | 0.0085 | 1.0× | MA9 only | |
| Jun 2025 | 40,144 | 0.0221 | 2.5× | ⚠️ Spike | MA9 only |
| Jul 2025 | 39,571 | 0.0088 | 1.0× | MA9 only | |
| Aug 2025 | 39,080 | 0.0501 | 5.7× | 🔺 Major | MA9 only |
| Sep 2025 | 33,402 | 0.0098 | 1.1× | MA9 only | |
| Oct 2025 | 30,913 | 0.0094 | 1.1× | MA9 only | |
| Nov 2025 | 30,620 | 0.0082 | 0.9× | MA3 + MA9 | |
| Dec 2025 | 30,301 | 0.0084 | 1.0× | MA3 + MA9 | |
| Jan 2026 | 30,042 | 0.0080 | 0.9× | MA3 + MA9 |
Verdict: Excluded from flagged accounts. Removing 2 spike month(s) reduces BPD change from -44.6% to -6.1%, which falls below the 25% alert threshold.
Full Top 30 BPD Overview
25 additional Top 30 accounts are within normal BPD range (> -25% change).
| Sev | Account | Revenue | BPD MA9 (MB/d) | BPD MA3 (MB/d) | BPD Chg |
|---|---|---|---|---|---|
| 🟢 | WQ Technologies | USD 187,038 | 0.0159 | 0.0121 | -23.9% |
| 🟢 | Transpoco | USD 21,203 | 0.3931 | 0.3116 | -20.7% |
| 🟢 | Axon Telematics (Ireland) Ltd. | USD 20,312 | 0.4208 | 0.3416 | -18.8% |
| 🟢 | Bboxx | USD 34,235 | 0.0168 | 0.0139 | -17.3% |
| 🟢 | AddSecure Group Services AB (formerly Verilocation Tracking Solutions) | USD 55,485 | 1.29 | 1.08 | -16.0% |
| 🟢 | Hello Tractor | USD 15,535 | 0.3142 | 0.2659 | -15.4% |
| 🟢 | React Health, formerly 3B Medical, Inc. | USD 281,068 | 0.0071 | 0.0061 | -14.1% |
| 🟢 | SolarEdge | USD 107,514 | 0.2731 | 0.2350 | -14.0% |
| 🟢 | LB Technology Inc (dba LB Telematics) | USD 32,184 | 1.19 | 1.03 | -13.7% |
| 🟢 | SUZOHAPP (formerly known as Coin Acceptors) | USD 28,110 | 0.1540 | 0.1416 | -8.1% |
| 🟢 | Trakm8 Ltd | USD 24,624 | 1.28 | 1.19 | -7.2% |
| 🟢 | AVAS | USD 16,124 | 0.0148 | 0.0082 | -6.1% |
| 🟢 | Visiontrack | USD 110,077 | 17.49 | 16.43 | -6.1% |
| 🟢 | Greenlight Planet Inc (SunKing) | USD 45,971 | 0.0137 | 0.0129 | -5.8% |
| 🟢 | BDLogix | USD 23,818 | 3.62 | 3.42 | -5.4% |
| 🟢 | Trimble Hosting Services | USD 21,989 | 0.0312 | 0.0298 | -4.5% |
| 🟢 | Engie Mobisol GmbH (Formerly Fenix International) | USD 36,314 | 0.0179 | 0.0178 | -0.6% |
| 🟢 | Silicon Controls Pty Ltd | USD 76,774 | 0.0008 | 0.0008 | +0.0% |
| 🟢 | IND Technology Inc. | USD 58,208 | 53.60 | 54.43 | +1.5% |
| 🟢 | Graco Inc. (Applied Fluid Technology) | USD 21,322 | 0.7613 | 0.8138 | +6.9% |
| 🟢 | Tenovi Health | USD 20,821 | 0.0216 | 0.0235 | +8.8% |
| 🟢 | IONX LLC | USD 24,307 | 0.0291 | 0.0318 | +9.3% |
| 🟢 | Spiro Mobility Holdco DMCC | USD 296,468 | 0.8181 | 0.9038 | +10.5% |
| 🟢 | Bobcat | USD 16,100 | 0.0147 | 0.0335 | +127.9% |
| 🟢 | Prospera_Valmont_AgSense | USD 57,917 | 0.3211 | 0.9222 | +187.2% |
Methodology Notes
- Data source:
aeriscom-acpcmn-prod-202006.cmp_sec.daily_account_network_summary— daily device count and data bytes per account - BPD formula:
total_bytes / days_in_month / device_count / 1,048,576→ MB per device per day - Recent period (MA3): Nov 2025–Jan 2026 — 3-month average of monthly BPD
- Long-term period (MA9): May 2025–Jan 2026 — 9-month average of monthly BPD
- Method: Overlapping moving averages — MA(3) is a subset of MA(9). This smooths firmware update spikes and seasonal variation.
- Dampening note: Because MA(3) is inside MA(9), reported declines are ~20-30% smaller than the true underlying change. The 25% threshold compensates.
- Spike detection: Flagged accounts are checked for months where BPD > 2.0× the account median. Spike months are excluded and MA values recalculated. Accounts whose adjusted decline falls below 25% are excluded as false positives.
- Scope: Top 30 revenue accounts with >= 50 avg devices and >= 2 months data in both windows
- Flag threshold: BPD decline >= 25% —
configurable via
BPD_FLAG_THRESHOLDinsignal3.py - Severe threshold: BPD decline >= 50% —
configurable via
BPD_SEVERE_THRESHOLDinsignal3.py - Full methodology: See
s3_data_compression/methodology.md
Generated: 2026-03-11 | Signal 3 v1.1