Analytics5 min

Forecast

Forecast is the time-series intelligence module of the Analytics track. Map a date column and a numeric metric, pick a horizon, and it projects the next 30, 60, or 90 days with trend and seasonality detection, anomaly (spike) analysis, and confidence intervals — so you can plan revenue, demand, traffic, or cost with an honest view of uncertainty.

At a glance

Level
Beginner
Duration
5 min
Product
Intelligence Navigator
Prerequisites
A dataset with a date column and a numeric metric
Audience
Owners, founders, firms, technical & non-technical teams
Subjects
Time-series forecastingtrend & seasonalityspike detectionconfidence intervalsAI insights

Map two columns, get a forecast

Forecast needs just two things: a date column (transaction_date, CloseDate, order_date) and a numeric metric to project (revenue, sales, demand, traffic, cost). Auto-detection suggests the mapping; you confirm it and choose how much history the model sees. At least 30+ rows is recommended for accuracy.

  • Choose a forecast horizon — 30 days for immediate planning, 60 for balanced accuracy, 90 for strategic planning
  • Control the historical range the model learns from — all data, last 30/90/180 days, last 12 months, or a custom range
  • Works on any numeric metric, not just revenue — demand, traffic, cost, inventory

See Forecast in action

Watch: map a date and a metric, run the forecast, and read the chart, drivers, and AI insights.

More signal when you have it

Two optional inputs sharpen the projection. Exogenous variables let you add other numeric columns that may influence the metric — marketing spend, price, headcount — and a breakdown dimension segments the forecast by customer, region, or product so you see the top drivers behind the totals.

  • Exogenous regressors improve accuracy when an outside factor drives your metric
  • Breakdown by segment reveals which customers, regions, or products move the number
  • The model automatically detects trend, seasonality, holidays, and cyclical behavior

Read the results

The results open with headline KPIs — total forecast, average daily, growth vs historical, and peak daily — then the forecast overview chart with upper and lower confidence bounds for risk-aware planning. A forecast-drivers view explains what's behind the curve, volatility and trend direction flag how steady the outlook is, and Claude-powered AI insights turn it into strategic priorities.

  • Confidence intervals give an honest upper/lower range, not just a single line
  • Spike analysis flags anomalies and unusual events in your history
  • Export the forecast — dates, values, and confidence bounds — to Excel, or save any chart to a Dashboard, Presentation, or Report

Best practices

  • Feed the model at least 30+ rows — more history means a more reliable trend and seasonality read
  • Add exogenous variables when you know a driver (e.g. marketing spend) moves your metric
  • Use the 90-day horizon for strategic planning, but trust the confidence bounds more than the point line
  • Re-run the forecast as new data lands so projections stay current

Next steps

Once you know where the number is headed, KPI Analytics tracks the metrics that matter — next in the Analytics track.