For FP&A · Finance Analysts · Controllers

Catch budget variances
before your CFO does.

Upload your P&L, budget report or revenue spreadsheet. ThresholdIQ's 9 ML methods automatically detect revenue anomalies, forecast deviations, seasonal distortions and correlated metric failures — no formulas, no BI tools, no rules to configure.

7-day unlimited trial · No credit card · Data never leaves your browser

$8.3T
lost globally each year to financial fraud and error
9
ML detection methods running simultaneously
<60s
from file upload to full anomaly report
0
threshold rules or formulas required

Finance teams catch anomalies too late

By the time someone spots the wrong number in a P&L, the board deck is already sent, the forecast is already wrong, and the damage is done.

📊

Budget variance surprises

Line items silently drift 15–20% off budget over weeks. Nobody notices until the quarterly review — when it's too late to course-correct or reforecast.

📈

Revenue anomalies buried in noise

A single region's revenue drops 30% but gets averaged out across the portfolio. The signal disappears entirely in the summary row.

🔄

Seasonal false alarms

December always spikes. Q1 always dips. Manual rules can't distinguish a real anomaly from predictable seasonality, so you either over-alert or miss everything.

Correlated metric failures

Revenue is fine, but COGS and headcount both shifted together. No single metric triggered an alert — but the margin is quietly collapsing. Static rules never see this.

From spreadsheet to alerts in 60 seconds

No BI tool to learn. No DAX formulas. No IT tickets. Just upload and get answers.

1

Export your financial data

Download your P&L, budget vs actuals, or revenue report from your ERP or accounting system. ThresholdIQ supports .xlsx, .csv, .json and .xml — or use the Excel file you already have.

2

Upload — data stays in your browser

Drag and drop the file into ThresholdIQ. All ML processing happens locally using Web Workers. Your financial data never touches a server. Zero compliance risk, no security review needed.

3

9 ML methods run automatically

EWMA, SARIMA, Isolation Forest, DBSCAN, correlation deviation, seasonal baselines, trend detection, Z-score and moving-average all evaluate every line item simultaneously. No thresholds to configure.

4

Review anomalies graded by severity

Every flagged anomaly is graded Warning, Critical, or Emergency. See which line items deviated, by how much, and whether the deviation is seasonal or genuinely abnormal. Export as CSV or PDF for your next review meeting.

Works across every financial data type

If your financial data has a date column and numeric metrics, ThresholdIQ finds what's abnormal — whether it's a P&L line item, a cash flow movement, or a forecast deviation.

💰

Budget vs Actuals Monitoring

Upload your monthly budget vs actuals report. Trend detection flags line items drifting off budget across multiple periods before they become a significant variance. Correlation deviation catches when multiple cost centres move together — a structural shift, not noise.

📈

Revenue Anomaly Detection

Multi-regional and multi-product revenue reports. SARIMA seasonal baselines ensure quarter-end spikes and promotional uplift don't trigger false alerts. EWMA detects sudden mid-month revenue drops in individual segments before they average out in the total.

🏦

Cash Flow & AR Aging

Daily cash flow reports and accounts receivable aging exports. Stuck-value detection catches when AR buckets stop moving — a sign of billing system failures or collection breakdowns. Trend detection surfaces gradual DSO creep weeks before it impacts liquidity.

📉

Expense & COGS Monitoring

Travel, software licences, headcount cost and COGS line items. Isolation Forest identifies unusual combinations — e.g. headcount normal but contractor spend doubled and travel spiked in the same period. The multi-metric signal no single-line review catches.

🎯

Forecast Deviation Tracking

Rolling forecast vs actuals over time. Multi-window Z-score escalates from Warning to Emergency as the deviation persists across short, mid and long-term windows — distinguishing a one-week blip from a structural forecast miss requiring replan action.

🔍

Intercompany & Reconciliation

Intercompany balances and monthly reconciliation reports. DBSCAN cluster noise identifies entity-level readings that don't match the normal reconciliation pattern. Correlation deviation flags when two sides of an intercompany transaction diverge simultaneously.

What ThresholdIQ finds in a budget report

This is the kind of anomaly that hides in plain sight in a 200-row P&L — until ThresholdIQ flags it automatically.

Line itemBudgetActualVarianceThresholdIQ
Revenue – APAC$1,240,000$1,195,000-3.6%Normal
Revenue – EMEA$890,000$612,000-31.2%🟠 Critical — EWMA + multi-window
COGS$420,000$398,000-5.2%Normal
Travel & Expenses$85,000$134,000+57.6%⚠️ Warning — trend + Z-score
Headcount cost$560,000$547,000-2.3%Normal
Software licences$48,000$97,000+102%🔴 Emergency — multi-window + isolation

ThresholdIQ detected EMEA revenue collapse, T&E overshoot, and licence cost doubling — three anomalies a simple variance % column would bury in a 200-row report.

How each ML method applies to financial data

All 9 methods run in parallel across every line item and metric in your file. Here's what each one specifically catches in financial data.

Multi-Window Z-Score

Evaluates each line item against 50, 100, 200 and 500-period rolling baselines. A revenue variance that persists across multiple windows escalates from Warning to Emergency — distinguishing a one-month blip from a structural miss.

Primary severity driver
📈
EWMA Spike Detection

Catches sudden mid-month revenue drops, unexpected expense spikes, and instantaneous cash flow anomalies that resolve quickly but are early warning signals of larger issues downstream.

Sudden variance detection
🌀
SARIMA Seasonal Residuals

Models quarter-end seasonality, month-start expense patterns, and annual budget cycles. Ensures a normal December revenue spike or Q1 cost reset doesn't trigger false Critical alerts on healthy data.

Quarter-end aware
🌲
Isolation Forest

Detects globally unusual combinations across all line items simultaneously — e.g. headcount normal but contractor spend, travel and software all elevated together, indicating an unreported resource surge.

Multivariate outlier
🔗
Correlation Deviation

Monitors relationships between revenue, COGS, and operating expenses. When normally-correlated metrics diverge — revenue flat but headcount cost climbing — a structural change is flagged before it hits margin.

Margin compression signal
📍
DBSCAN Cluster Noise

Identifies line items or entities that don't belong to any normal financial pattern cluster. Catches unusual intercompany transactions, one-off accruals that look like posting errors, and entity-level outliers.

Posting error detection
🌙
Seasonal Baseline

Maintains separate normal ranges per month and per quarter. Budget-flush spending in December and Q1 slow periods are learned patterns — alerts only fire on genuinely abnormal deviations.

Context-aware
📉
Trend Detection

Identifies monotonic drift in recurring cost lines, gradual revenue decline by segment, and slowly increasing DSO. Flags the trend at Warning level weeks before any threshold breach.

Early drift warning
🚫
Stuck & Zero Detection

Repeated identical values in a line item indicate a data feed failure or stale extract. A sudden zero in an active revenue stream fires Emergency immediately — catching ERP connectivity issues before month-end close.

Data feed failure

Your financial data already knows
something is off.

Upload your spreadsheet. Nine ML methods run in seconds. Anomalies surface instantly — graded by severity — with zero configuration and zero data leaving your browser.

Just enter your email — no password required

No setup · Works with any accounting system or ERP export