Use case

Website traffic forecasting for capacity and campaign planning

Forecast daily sessions from a CSV, separate weekday patterns from campaign spikes, and use the range to plan infrastructure and content.

Run the sample Put this in production

Traffic forecasts are used for two very different jobs: capacity (will the site stay up?) and planning (how many leads should we expect?). Both need a range. A median alone encourages either over-provisioning or a surprise on launch day.

The sample below is daily sessions with a weekday/weekend pattern and a gradual upward trend. If you have a campaign calendar, forecast the organic baseline first; a campaign spike is better modeled as a separate scenario than as a permanent level shift.

The sample series

The studio below is preloaded with this daily sessions series and a 14-step horizon. It is an illustrative example, not a real company’s data. Replace it with your own CSV to run the same workflow.

Show the CSV

date,sessions
2026-03-01,1180
2026-03-02,1420
2026-03-03,1490
2026-03-04,1510
2026-03-05,1470
2026-03-06,1320
2026-03-07,1100
2026-03-08,1210
2026-03-09,1460
2026-03-10,1530
2026-03-11,1560
2026-03-12,1500
2026-03-13,1360
2026-03-14,1130
2026-03-15,1250
2026-03-16,1500
2026-03-17,1580
2026-03-18,1610
2026-03-19,1550
2026-03-20,1400
2026-03-21,1160
2026-03-22,1280
2026-03-23,1540
2026-03-24,1620
2026-03-25,1650
2026-03-26,1590
2026-03-27,1440
2026-03-28,1190
2026-03-29,1310
2026-03-30,1580
2026-03-31,1660
2026-04-01,1690
2026-04-02,1630
2026-04-03,1470
2026-04-04,1220
2026-04-05,1340

CSV forecast studio

Forecast your website traffic forecasting series.

This studio is preloaded with the daily sessions example above. Replace it with a CSV of your own history, choose a horizon, and run the forecast. The public demo is free while the Hugging Face Space is available; the API is the paid product.

1 · Paste or upload a CSV2 · Pick a horizon3 · Download the 21 quantiles

Steps

1 credit per API call · the web demo is free while the public Space is available

What to watch in the output

Weekday and weekend are different series

A weekly pattern is the strongest signal in most site traffic. If the forecast ignores it, the daily numbers will look alternately too high and too low even when the monthly total is right.

Separate bots and internal traffic

Analytics that include bots or your own team produce a forecast for an audience you cannot convert. Filter known bots and internal IPs before exporting the CSV.

Model campaigns as scenarios

A one-week paid campaign is an intervention, not a new baseline. Forecast the organic series, then add the campaign upside as a separate line so the two assumptions stay visible.

Common mistakes

—Exporting a partial day. The last day of a traffic export is often incomplete; drop it or mark it missing.
—Mixing sessions and users. The definitions differ and the series can diverge after a tracking change.
—Ignoring a tracking outage. A sudden drop to zero is a measurement failure, not a traffic collapse; repair the gap before forecasting.

How to put it to work

1. Start with the sample. Run the preloaded series once so you see the output shape before you debug your own CSV.
2. Replace it with your history. Keep one timestamp column and one or more numeric columns. Mark outages and stockouts as empty cells, not zeros.
3. Backtest before you plan. Hold out the last few periods, compare the forecast with what happened, and check whether the actuals land inside the 80% band.
4. Move it into production. Use the REST API for a nightly job or an internal tool. One credit per successful call; failed calls are refunded.

FAQ

Can I forecast hourly traffic?

Yes, if the CSV has hourly timestamps and enough rows. Hourly traffic usually needs at least a week of history to learn the daily cycle; two weeks is safer.

Will the forecast account for a product launch?

Only if the launch pattern is already in the history. For a new launch, forecast the baseline and model the launch upside separately as a scenario.

What is a good horizon for capacity planning?

Match the horizon to the decision: one to two weeks for staffing and cache capacity, one to three months for content and infrastructure budgeting.

Need this as an API?

The same forecast is available as a metered REST endpoint with JSON and a downloadable quantile CSV.

See the API