Why compset math leaves 3–8% on the table

PriceLabs is the best compset-based pricing engine in vacation rentals. It watches your competitors, it watches your market, it watches the pace of the booking window, and it prices accordingly. For most nights, in most markets, that's enough.

But "most nights in most markets" isn't how a multi-market operator makes revenue. Revenue is made in the exceptions — the gap night that never books, the peak weekend you got pulled off the market because of a maintenance hold nobody told pricing about, the owner arrival that ate your highest-rate Friday, the pacing alert in Austin that should have triggered a minimum-stay drop four days earlier.

Compset-only pricing leaves 3–8% RevPAR on the table across a portfolio — not because the algorithm is wrong, but because the algorithm can only see the compset. It can't see your goals, your operations, your owner schedule, or your unit-level constraints. That context exists inside your business. It's just not flowing into PriceLabs.

When it does, the 3–8% stops being a leak.

The signals that change pricing decisions

There are roughly a dozen context signals that most portfolios have but never pipe into their pricing engine. Four matter more than the rest.

Maintenance holds that weren't communicated

A unit gets pulled off the market because the water heater failed. Your ops team creates a hold in Breezeway. Your housekeeping coordinator blocks the calendar. But PriceLabs is still pricing the nights around that hold as if the full inventory is live — so the adjacent weekend prices as a gap-night deal when it should be pricing as last-remaining-inventory.

The fix isn't a policy document telling ops to remember to update pricing. The fix is a feed — Revenue agent watches the hold event fire in your PMS, infers the revenue implication, and pushes the adjusted constraint to PriceLabs before the next sync. It's a ten-second loop that closes a revenue leak you can't close with a standup meeting.

Owner arrivals on the calendar

Owner stays show up in your PMS as blocks, not bookings. PriceLabs sees a block and treats the adjacent inventory as normal. But an owner arriving Thursday afternoon changes your dynamics: the Thursday night pricing should already be dropping, the Wednesday check-out window should be tighter, the Sunday departure should align with the turn crew's schedule.

None of that happens without context. Revenue agent carries the owner-stay metadata forward — "this is a 3-night owner stay, returning Sunday, no cleaning fee cost recovery" — so the rates around it reflect the constraint, not a generic block.

Quarterly RevPAR goals set in planning

You set a Q3 RevPAR target during your quarterly planning cycle. It's in your Goals agent as a KR. But your pricing engine doesn't know it exists.

This is the gap that's hardest to see because it's the most invisible. You can be pacing -12% against your Q3 target in week 5 and PriceLabs will still be doing the best job it knows how to do on compset math — even though what you actually need is a portfolio-level pricing pivot to catch up. The pacing watcher has to see the pacing gap, the drift detector has to flag it, and Revenue agent has to translate the gap into a pricing move (minimum-stay relaxations, open-window discounting, channel mix shift) before the quarter runs out.

Gap nights between high-rate bookings

A 2-night gap between a Friday arrival and a Monday checkout is a different problem than a 4-night gap mid-week. Compset pricing treats them the same. An operator-aware pricing loop treats the first as a check-in-only discount candidate and the second as a minimum-stay relaxation candidate.

Gap.closer — a sub-agent of Revenue — looks at each orphan window in the booking curve and proposes the shape of the fix, not just a price change. Sometimes the right answer is a $45 credit to a past guest to convert the window. Sometimes the right answer is simply dropping the minimum-stay from 3 to 2. PriceLabs can do the rate move, but it can't make the category-of-fix call.

How Revenue agent feeds context into PriceLabs

The architecture is simple on purpose. PriceLabs stays. It's good at what it does. Revenue agent is a context layer, not a replacement.

The loop runs like this:

An event fires in your PMS — a hold, an owner stay, a pacing variance, a new gap.
Revenue agent catches the event and asks: does this change the pricing posture for any affected nights?
If yes, it computes the delta — a minimum-stay change, a floor adjustment, a window override, a constraint tag — and writes it to the PriceLabs unit configuration.
PriceLabs does the actual pricing work, but now with the context baked in.

No replacement. No re-platforming. The team that learned PriceLabs doesn't re-learn anything. The rates just get smarter.

What stays in PriceLabs

Everything pricing. Compset analysis, demand curves, DOW adjustments, seasonality, the minimum-stay ladder, the channel rate stacking, the promotion logic. PriceLabs is the rate engine and it stays the rate engine. That's important to say explicitly, because the common misread is that "AI-powered revenue management" means swapping pricing tools. It doesn't. You already paid for the right tool. The gap is that it's been priced without the context of your business.

Revenue agent doesn't move rates. It moves constraints, goals, and holds — the inputs that change what "good pricing" means for a given night.

A real example — 8 listings pacing -18% in Austin

A scenario we see regularly: at 9:04 AM on a Tuesday, Pacing.watcher flags Austin pacing -18% against goal for the next 30 days. Eight listings are implicated. The 30-day revenue gap is $11,420 if nothing changes.

Without context feeding, the next move happens at a standup three days later — if it happens at all. Someone pulls a report, someone debates whether it's a market softness or a unit-mix issue, someone eventually asks pricing to "do something."

With Revenue agent in the loop, three things happen before 9:20 AM:

9:04 AM — Pacing.watcher posts the variance. 8 listings, $11,420 gap, 14-day booking window.
9:16 AM — Guests agent's Recovery.planner fires a past-guest flash campaign targeted to the 212 past guests in the Austin market. Subject lines and offers pre-drafted, sitting in your queue for one-click review.
9:20 AM — Revenue agent writes the PriceLabs configuration update: minimum-stay relaxed from 3 to 2 on the affected nights, floor rates dropped 8%, check-in-only pricing enabled for 2-night orphan windows. Align agent logs the pacing variance against the quarterly RevPAR KR so the trend stays visible in your planning view.

The human decision — do you approve the flash campaign, do you approve the pricing move — happens at 9:25 AM instead of next week. That difference is the 3–8% RevPAR lift. Not algorithmic magic. Operational speed.

What this looks like on your portfolio

You don't swap PriceLabs. You don't retrain the team. You add the context layer, and the rates PriceLabs was already producing start reflecting the holds, the owner arrivals, the goals, and the gaps. The 3–8% that was leaking stops leaking.

Ready to see Revenue agent on your portfolio? Request access →