Every forecast learns from history. When history carries the wrong segment, the forecast learns the wrong pattern and repeats it with confidence. Revenue managers rarely create these errors, yet they answer for them.
Below: why the forecast inherits coding errors, the three places codes break, a 30-minute audit you can run this week, and how to put the result on the record.
Hotel Forecasting Learns From Coded History
Your forecast model reads segment history as fact, whether it lives in the RMS, in your spreadsheet, or in both. It cannot tell a corporate guest from a leisure guest coded as corporate. It sees a pattern and projects it forward.
Data is never clean in general. It is clean enough for a specific job, or it is not. For the forecast, the job is easy to state: each room night sits in the segment that describes why the guest booked, and it stays there.
Most segment errors start outside the revenue office. They enter at the reservation desk, the front desk, and the sales office, often months before you see them.
Three Places Your Codes Break
Segment codes break in three predictable places.
- Names: sales calls a booking corporate, the PMS codes it BUS, and finance reports it as business transient. Three labels for one guest give three answers to one question. Your segmentation follows the commercial logic of the hotel, and USALI serves its own reporting purpose; the trouble starts when nobody maps one to the other.
- Structure: room type and rate codes differ between hotels in the same group. Packages split room and F&B revenue one way in one hotel and another way in the next, so ADR and TRevPAR move without a single guest changing behavior.
- Input: a default segment fills itself in when a busy agent skips the field. Comp and house-use rooms sit in revenue history. Dummy reservations and manager holds never leave the system. Duplicate company and travel agent profiles split one account into three.
None of these require poor work. They require a busy front desk, a new hire and a rule nobody wrote down.
Run the 30-Minute Code Audit
Export 90 days of reservations from the PMS, with segment, source, rate code, company, and travel agent profile. Count five things and write each one as a share.
|
Check
|
What to count
|
Your share
|
|
Default segment
|
Room nights in "other", "unknown" or the PMS default
|
__% of room nights
|
|
Unmapped rate codes
|
Active rate codes with no segment assigned
|
__% of rate codes
|
|
Conflicting codes
|
Reservations where rate code and segment point to different segments
|
__% of reservations
|
|
Non-revenue in history
|
Comp, house-use, dummy, and hold reservations in revenue history
|
__% of room nights
|
|
Duplicate profiles
|
Companies and agents with two or more profiles
|
__% of profiles
|
Keep the numbers exact. A baseline only helps if you can repeat it next month and compare. Run it per hotel, not per group, because the leaks differ by property.
The audit also tells you how much weight each segment deserves. A segment with clean history earns more of your trust than one where the codes drift.
Fix the Source, Keep the Spreadsheet
Your spreadsheet holds years of corrections: the account you re-coded, the group you moved, the holds you stripped out. That work is valuable. It also protects only the month you work on, because the PMS history underneath stays wrong and teaches the next forecast the same mistake.
Move the corrections upstream. Agree on a segment list with sales and finance, and name one owner for it at each hotel. Remove default segments from the reservation screen so that an empty field prompts a decision. Merge duplicate profiles once, then make profile creation someone's job.
Your judgment stays in the model. The fragility leaves the data.
Put the Audit on the Record
Share the baseline with your GM every month, in the same format, with the same five checks. When a month misses, the record answers the question before anyone asks it: did the forecast move, or did the inputs?
After a few months, you see which checks improve where a hotel has an owner and a rule, and which stay flat where nobody owns them. Your forecast stops carrying errors it never made.
Demand Calendar maps every hotel's segment and rate codes to a single structure at implementation and verifies the data and mapping, so your audit starts from a known baseline.
Until someone checks the codes, every miss lands on the forecast. Run the audit, and the misses land where they belong.
Measure your forecast with the same discipline as your codes.
Download the Forecast Accuracy Scorecard →