Trinetra Revenue logo
TrinetraRevenue
← All insights
Revenue Strategy2026-09-048 min read read

Forecasting and Pace Management for Indian Hotels

How to read booking pace week by week, build a 90-day forecast you actually trust, and act before the rate decision gets expensive.

Most hotels price by feeling. Pace management replaces the feeling with a number: how many room nights are already on the books for a future date, compared with the same point in time last year.

Build the pace view first

For every future date, track three things: rooms on the books, ADR on the books, and the gap to the same lead-in day last year. A 90-day rolling window is enough for city hotels; leisure resorts with wedding and holiday demand need 180 days.

Read the signal, not the noise

  • Ahead on rooms, behind on ADR — you sold too cheap too early. Raise the floor for the remaining inventory.
  • Behind on rooms, ahead on ADR — you are holding rate well but risk empty nights. Open a lower-fenced rate (advance purchase, longer stay) before you cut the public rate.
  • Behind on both — a demand problem, not a pricing problem. Check comp-set rates, event calendars and OTA visibility before discounting.

Forecast in three layers

  1. Base: last year's actuals for the same day of week, adjusted for your current trailing 90-day index.
  2. Events: weddings, conferences, festivals, exam dates, flight capacity changes.
  3. Judgement: known group blocks, renovation, new supply in the comp set.

Write the forecast down before the month starts. A forecast you can compare to actuals teaches you something; one you adjust after the fact teaches you nothing.

The weekly rhythm

A 45-minute pace meeting every Monday is enough: review the next 14 days in detail, the next 90 at summary level, agree three rate actions, and record them. Over a quarter that discipline is worth more than any single clever pricing decision.

What good looks like

Forecast accuracy within 5% at 14 days out, within 10% at 30 days. If you are consistently outside that, the base year data is wrong or the event calendar is incomplete — fix the inputs before blaming the model.