Meter anomalies cost revenue.
Find them in your export.
Export your meter reads, billing reconciliation, production logs, or distribution data into a spreadsheet. ThresholdIQ's 9 ML methods automatically flag consumption spikes, billing discrepancies, meter faults, and seasonal deviations — no SCADA required, no IT integration, no thresholds to configure.
7-day unlimited trial · No credit card · Data never leaves your browser
Utility data is full of signals nobody has time to find
Meter exports have thousands of rows. Billing reconciliation sheets have hundreds of accounts. Manual review catches the obvious spikes — but the subtle, costly anomalies slip through every cycle.
Meter read anomalies
A meter reports zero consumption for 3 weeks — damaged, tampered, or a data gap? Another spikes 400% after a firmware update. Both look like noise in a 10,000-row export. One is theft; the other is a calibration error. Static rules can't distinguish them.
Billing-to-consumption mismatch
Billed kWh doesn't match metered kWh for 47 accounts. Some are rounding. Some are rate-class errors. Two are revenue losses of $8,000/month each. In a reconciliation spreadsheet, they all look the same until someone manually investigates every row.
Distribution & transmission losses
Technical losses are expected — 4-7% is normal. But when losses creep from 5.2% to 6.8% over six months, is that degrading infrastructure or non-technical loss? The trend is the signal, and monthly reports don't surface trends.
Seasonal pattern violations
Summer demand always peaks. But when a substation's peak is 15% below seasonal expectation, is load migrating, or is a meter miscounting? SARIMA-based detection separates genuine anomalies from expected seasonality — conditional formatting never can.
How utilities use ThresholdIQ on exported data
You don't need a SCADA integration or a data lake. Export to spreadsheet, upload, get results. Here's what each sector monitors.
Meter read validation & theft detection
Export AMI meter reads into a spreadsheet. ThresholdIQ flags meters with zero reads, impossible consumption jumps, rollback patterns, and consumption profiles that diverge from their peer group. Prioritise field investigations by severity instead of reviewing every exception manually.
Billing reconciliation anomalies
Export billed vs metered data. ThresholdIQ detects accounts where billed amounts diverge from consumption patterns — rate misclassifications, estimation errors, and systematic under-billing that static rules configured years ago no longer catch.
Leak detection & non-revenue water
Export DMA (District Metered Area) flow data or customer-level consumption. ThresholdIQ flags zones where night-time minimum flow exceeds seasonal baselines — a leading indicator of leaks. Also catches individual meters with sudden consumption changes that suggest burst pipes or meter failure.
Customer consumption profiling
Upload customer usage data. ThresholdIQ identifies accounts whose consumption is structurally different from their peer group — potential illegal connections, meter tampering, or billing errors. Isolation Forest excels at finding the 0.5% of accounts that don't fit the pattern.
Production & well monitoring
Export daily production logs — flow rates, pressures, water cut percentages. ThresholdIQ detects wells where decline curves deviate from expected trajectories, pressure anomalies that precede equipment failure, and correlated metric shifts across multiple wells in the same field.
Pipeline & facility data checks
Upload SCADA or historian exports of pipeline pressures, flow rates, and temperatures. ThresholdIQ catches slow drift in readings that indicate sensor calibration issues, correlated pressure/flow changes suggesting partial blockages, and seasonal deviations from normal operating profiles.
From data export to anomaly report in 60 seconds
No IT project. No vendor integration. No procurement cycle. Export your data, upload, get answers.
Export from your existing system
AMI/MDM, billing system, SCADA historian, or even a manually assembled spreadsheet. Any structured data in .xlsx, .csv, .json, or .xml format. ThresholdIQ doesn't need a direct integration — it works with whatever you can export.
Upload — all processing stays in your browser
Drag and drop the file. ThresholdIQ's ML engine runs entirely in your browser using Web Workers. Your meter data, customer records, and operational data never leave your machine. No cloud upload, no compliance review, no security questionnaire.
9 ML methods run automatically
EWMA detects slow consumption drift. SARIMA separates real anomalies from seasonal patterns. Isolation Forest finds meters that behave nothing like their peers. Correlation deviation catches when multiple readings shift together. No parameters to set.
Review anomalies graded by severity
Every flagged item is graded Warning, Critical, or Emergency. See which meters, accounts, or zones deviated, by how much, and which detection method flagged it. Export the anomaly report as CSV or PDF to share with field teams or management.
What ThresholdIQ finds in meter data exports
Exported from a billing system — 6 accounts out of 4,200. A manual review would scan the kWh column and maybe catch the zero-read. ThresholdIQ catches all four anomalies.
| Meter ID | Account | kWh (period) | kWh (prev) | Billed ($) | ThresholdIQ |
|---|---|---|---|---|---|
| MTR-4201 | Residential | 842 | 810 | $126.30 | Normal |
| MTR-4202 | Commercial | 0 | 12,400 | $0.00 | 🔴 Emergency — zero read |
| MTR-4203 | Residential | 780 | 795 | $117.00 | Normal |
| MTR-4204 | Industrial | 48,200 | 31,600 | $4,338.00 | 🟡 Warning — +52% spike |
| MTR-4205 | Residential | 920 | 3,100 | $138.00 | 🔴 Critical — 70% drop |
| MTR-4206 | Commercial | 5,400 | 5,200 | $242.00 | 🔴 Critical — billing mismatch |
MTR-4202: possible theft or fault. MTR-4204: legitimate demand increase or meter error. MTR-4205: meter rollback or tampering. MTR-4206: billed at wrong rate class (residential rate on commercial meter).
How each ML method applies to utility data
All 9 methods run in parallel across every meter and metric in your file. Here's what each one specifically catches in utility data.
Evaluates each meter against rolling baselines at 50, 100, 200 and 500 readings. A consumption deviation that persists across multiple windows escalates from Warning to Emergency — separating a one-period spike from a sustained theft or fault pattern.
Primary severity driverCatches sudden consumption spikes, instantaneous pressure anomalies, and rapid flow-rate changes. Detects the sudden event before it compounds across billing periods — flagging it within the same reporting window it occurs.
Instantaneous spikeModels summer irrigation peaks, winter heating demand, and industrial production cycles. A predictable summer consumption increase won't fire alerts. A summer peak that exceeds the seasonal expectation by 40% will — distinguishing weather from anomaly.
Seasonal demand awareDetects globally unusual combinations across all meter metrics simultaneously — e.g. consumption normal but pressure, flow and power factor all abnormal together, indicating a distribution network fault rather than meter-level theft.
Network fault detectionMonitors relationships between meters on the same transformer or distribution segment. When one meter's consumption drops while all peers stay constant, the deviation is flagged as potential bypass. When all meters in a DMA drop simultaneously, a main break is suspected.
Tamper & main breakIdentifies meters that behave unlike any normal consumption cluster. Catches reverse-wiring patterns that produce negative consumption, meters that consistently read in a pattern inconsistent with their tariff class, and meters with unusual load profiles suggesting bypass.
Bypass & wiring faultsMaintains separate normal ranges per hour-of-day and day-of-week for each meter class. Industrial meters at 3am and residential meters on Sunday mornings have very different expected baselines — alerts only fire when a meter deviates from its own time-specific normal.
Meter-class awareIdentifies monotonic drift in consumption — a meter that reads 2% below average this period, 3.5% the next, 5% the next. Slow tamper or gradual meter register drift accumulates into significant billing loss before any threshold-based rule fires.
Slow tamper & driftRepeated identical meter readings or sudden zero consumption from an active account immediately signals meter freeze, communication failure, or complete service disconnection. Fires Emergency — catching AMI communication faults before an entire billing cycle is estimated.
Meter & AMI failureYour utility data never leaves your browser
ThresholdIQ processes everything client-side using Web Workers. No meter data, customer information, or operational records are transmitted to any server. This means zero compliance burden — no security questionnaire, no data processing agreement, no vendor risk assessment. Your IT team doesn't need to be involved. Export, upload, analyse, done.
Your meter data already contains the warnings.
ThresholdIQ surfaces them.
Upload any utility data export. Nine ML methods run in seconds. Meter anomalies, billing errors, and distribution losses surface instantly — graded by severity.
Just enter your email — no password required
Works with any AMI, MDM, SCADA, or billing system export