What the model cannot know
A rate recommendation system cannot see information that never enters a data feed. TurnStay's model uses lead-time pickup, competitor rates, local event calendars, historical baselines, and occupancy position. For a small independent property, that is a strong set of signals. It is still incomplete. Some demand clues exist only in your own knowledge, and those are the moments when overriding the recommendation makes sense.
TurnStay includes an override function because pricing should remain open to operator review, not become a black box. The aim is to give you a stronger starting point each morning, not hand decision authority to software. Knowing when the model is likely right, and when local judgment deserves priority, is part of sound revenue management with a system like ours.
When to trust the recommendation
The model is strongest when market behaviour stays within its historical range. Recurring weeknight and weekend patterns, seasonal cycles, and compression events that return each year are structured variations it is calibrated to recognise. When conditions fit that pattern, its recommendations are generally reliable.
Trust the recommendation when you have no new local demand information, your property is trading steadily, no unusual outside factor is present, and the rate fits your own market read. Reviewing a recommendation and agreeing with it is active confirmation, not passive acceptance. The model supplies the starting number, while your review remains part of the process.
During the first month after connecting to TurnStay, operators often override frequently, and many of those decisions seem reasonable at the time. Yet at 30 days, most find that the model would have delivered better outcomes on most of them. Its signal set is wider than it feels, and it does not share the anchoring and loss aversion that can shape decisions based on memory and instinct.
Situation 1: Private events or group business outside the model
A private event missing from public calendars, a corporate group inquiry not yet converted into a confirmed booking, or a local conference known through community channels before aggregated data records it are all examples of demand information you have and the model lacks.
In these cases, override upward and set the floor above the recommendation for the target nights. The model will adjust as bookings appear over the following days. Until then, your local knowledge is the fuller input. TurnStay lets you set rates for specific dates and use them instead of the recommendation, while continuing to track pickup and alert you if demand weakens unexpectedly.
Situation 2: Changes in competitor supply
A nearby property reducing availability during renovation, a new property using deep introductory discounts to gather early reviews, or a key competitor temporarily leaving a distribution channel can all change the local supply picture. The model's competitor feed may not capture these shifts fully, depending on its comparison set and how often the feeds refresh.
If a key competitor has cut inventory this week, pushing rates above the recommendation has a credible basis. If a new entrant fills the comparison set with discounts that misrepresent the market and will normalise in 60 days, holding above the model's temporarily low recommendation may be wiser than matching a price that will disappear before bookings close.
Situation 3: Operational limits and brand policy
A lower rate can sometimes be the right override even when demand has not changed. If you are short-staffed, a full weekend could reduce service quality. If one room type has a maintenance problem, it may be a weaker product than usual. These are operational constraints, not demand signals, and they can justify protecting service delivery ahead of RevPAR.
Brand floor is another case. For a premium property, a slow week's recommendation may fall below the lowest rate the brand should show, regardless of occupancy. A hard floor in TurnStay applies that policy over the demand signal. You are not disputing the model's read, but adding a business constraint, which is a valid use of the override function.
When overrides usually hurt
The costliest common mistake is raising rates above the recommendation on already-strong demand nights out of habit rather than new information. When the model has made a significant lift and occupancy is strong, it is already responding correctly. Another increase can price out bookings that the lower recommended rate would have captured.
The reverse mistake is cutting rates on nights that look slow simply because the week feels quiet. The model's view includes your position on the pickup curve at the current lead time. Looking only at occupancy, without that context, makes operators overread softness and reduce rates that would otherwise have held.
Use this rule: override when you know something the model cannot see. Keep its recommendation when you are relying on instinct or pattern matching. Review overrides at 30-day intervals. That habit, more than any single decision, supports better rates across a full season.
Ready to optimise your rates and screen your bookings?
Start a 14-day free trial. No credit card required.
