Where the Hours Go: An Anatomy of Booking Operations

Ask a reservations team where their week goes, and the answer is almost always the same word: email. It’s true, and it’s useless. Email is a channel, not a task, and you can’t fix a channel.

So we measured. Over two weeks, a five-person reservations team at a destination management company in the Mountain West logged every working block in ten-minute increments, tagged by task and by what triggered it. One of our operations people shadowed two half-days to check the self-reports against what actually happened at the desk. What follows is the anatomy: anonymized, rounded, and with the method included so you can run it yourself.

The headline: 360 hours, mostly spent agreeing with suppliers

The team logged 360 working hours across the two weeks. In that time they handled 104 bookings at various stages, which together contained 624 supplier services: rooms, guides, transfers, restaurant tables, permits.

Task	Hours	Share
Sending and chasing supplier requests	74	20.6%
Reading and keying confirmations	61	16.9%
Amendments and their knock-on changes	52	14.4%
Building and revising proposals	41	11.4%
Answering pre-trip traveler questions	38	10.6%
Payment reconciliation	27	7.5%
Vouchers and travel documents	22	6.1%
Disruption handling	19	5.3%
Internal coordination	18	5.0%
Other	8	2.2%

The first three rows add up to 52% of the week, and they have one thing in common. Each is about getting someone else’s record of the trip to match yours: a supplier’s spreadsheet, a confirmation PDF, a reservation that moved by a day. Booking operations, it turns out, is mostly the reconciliation of competing versions of the truth.

That reframing matters because it points to a different fix. Typing faster doesn’t help. Reducing the number of places where the truth can diverge does.

The unit that explains everything: touches per confirmed service

Hours tell you where time went. Touches tell you why.

A touch is any human action on a service before it’s confirmed: sending the request, chasing a reply, reading it, keying it in, fixing a mismatch. Across the 624 services, the average was 3.3 touches and about 13 minutes of handling per service. That sits close to the 12-minute default in our ROI calculator, which was reassuring.

The average hides the shape. In this study, 41% of services were confirmed in a single touch. Another 18% needed six or more, and that 18% consumed nearly half of all confirmation time. The long tail is where the hours live.

What makes a service high-touch

The strongest predictor wasn’t supplier size or service type. It was the channel the supplier used to reply.

Supplier reply channel	Share of services	Avg. touches	Avg. minutes per service
API or extranet	22%	1.2	3
Structured email template	31%	2.6	9
Free-text email	34%	4.1	17
Phone, PDF or fax-to-email	13%	6.3	28

The bottom row is small in volume and enormous in effort: a guesthouse that confirms by phone, a guide who sends a photo of a handwritten note, a supplier who still confirms by fax-to-email. These are often the most characterful suppliers on the itinerary. You don’t want to drop them. You want to stop re-keying them.

Beyond the channel, three patterns drove extra touches:

• Partial confirmations. “Three of four rooms confirmed; the fourth on request.” Every partial creates a follow-up.
• Name and date mismatches. A traveler’s passport name differs from the booking name, or a supplier logs the stay against the wrong night.
• Silent changes. A supplier changes a pickup time in a reply without flagging it. Someone has to notice.

The amendment cascade

The team handled 22 amendments in two weeks, at an average of 2.4 hours each. One date change on a nine-day trip touched seven services on average: room nights shift, a guide moves, two transfers are retimed, a dinner reservation moves, a permit is reissued.

Every one of those changes was re-keyed by hand into the booking, sent to the supplier, confirmed and checked. The cheapest amendment is the one your system propagates from a single change.

The 38 hours of traveler questions

Pre-trip questions looked like customer service, but when we read the log, about 60% of them asked for something already written in the itinerary or the booking record: pickup times, what to pack for a cold desert morning, whether breakfast was included, how far the trailhead was from the lodge. Another quarter were requests for changes, which belong in the amendment column. Only the remainder needed real judgment.

That split matters. A question the itinerary already answers is a retrieval problem, and it can be answered instantly, in the traveler’s own channel, from the same facts the team uses. The judgment questions are the ones worth a person’s time.

Cheapest fixes for the long tail

Some of the biggest gains in this study needed no new software at all. In order of cost:

1. Give free-text suppliers a reply format. A one-line reply pattern (booking reference, dates, guest count, rate, notes) in every request turned a good share of free-text replies into something a person or a parser can check in seconds.
2. Put one booking reference in every subject line. Most “which booking is this?” touches disappeared once every message carried the same reference.
3. Offer the phone-and-fax suppliers a one-tap confirmation. A text message with three options (confirm, change, decline) meets a small guesthouse owner where they already are.
4. Decide which record wins. Your booking record is the source of truth; a supplier’s change arrives as a proposal someone accepts, never as a silent overwrite.

None of these replaces automation. They make the automation work, because they turn the long tail into inputs a system can actually read.

Sort the hours: automate, assist or keep

Before buying anything, sort every task into one of three bins.

Automate. The machine does the work and a person sees only the exceptions. Drafting supplier requests from the booking record. Parsing structured confirmations and matching them to booking lines. Generating vouchers. Matching payments to bookings.

Assist. The machine drafts and a person decides. Free-text confirmations that contain discrepancies. Amendments, where the system proposes the full cascade and a person approves it. Traveler questions, answered from the itinerary’s own facts. Disruption re-plans.

Keep human. The work that is the relationship. Negotiating with suppliers. Unusual requests. Upset travelers. Judgment calls on exceptions.

Task	Bin	How it works
Supplier requests and chasing	Automate	Requests drafted from the booking record; chasers scheduled
Confirmations	Automate, exceptions assisted	Parse, match, flag mismatches for review
Amendments	Assist	System proposes the cascade; a person approves
Proposals	Assist	Drafted from traveler intent; the advisor edits
Traveler questions	Assist	Answers grounded in itinerary data
Reconciliation	Automate	Exceptions routed to finance
Vouchers	Automate	Issued from confirmed records
Disruptions	Assist	Re-plan options drafted; a person decides
Supplier relationships	Keep human	Where margin and reputation live

Our ROI calculator uses conservative reductions, set below what we have measured in pilots: 70% for confirmation handling, 55% for proposal work and 40% for disruption handling. Applied to this team’s two weeks, that’s roughly 95 hours from supplier requests and confirmations, 23 from proposals and 8 from disruptions: around 125 hours, or about 35% of the time logged. We deliberately leave amendments and traveler questions out of that estimate, even though both improve, because the variation between teams is too wide to model honestly.

The confirmation work maps onto booking workflow automation: requests, parsing, matching, vouchers and reconciliation in one place, with exceptions lifted aside for a person. The fax-to-email suppliers usually need some data engineering and integration first, so their replies land as structured records instead of attachments. And the disruption hours respond best to written playbooks before any automation; that’s the starting point for disruption response automation.

What the hours become

The team in this study didn’t want fewer people. They wanted to stop being the slowest part of their own sales process. When we asked what they would do with the time, the answers were specific: turn quotes around the same day; call travelers before departure instead of after a complaint; visit suppliers in person during the quiet months.

Those are revenue activities hiding inside an operations problem. Name them before you start, so the recovered hours go somewhere deliberate rather than being absorbed into more email. For operators weighing this at the business level, our page for tour operators and DMCs walks through where that time typically goes next.

Run your own study in ten working days

You don’t need software for this. You need a shared spreadsheet and two weeks of discipline.

1. Fix the categories up front. Use the ten above, or your own. If “other” grows past 5%, your categories are wrong.
2. Log in ten-minute blocks, with the task and its trigger: new booking, amendment, supplier reply, traveler question or disruption.
3. Tag every supplier service with its reply channel, and count its touches until it’s confirmed.
4. Shadow two half-days. Self-reports drift; observation keeps them honest. People also work differently in the first day or two of being measured, so treat day one as practice and leave it out of the totals.
5. Count amendments and how many services each one touched.
6. Compute five numbers: hours by category, touches per service, minutes per service by channel, the 20 suppliers who cost the most minutes, and hours per amendment.
7. Sort every task into automate, assist or keep human.
8. Pick one target, usually the highest-minute supplier channel, and fix that first.

The study is cheap, slightly tedious and almost always surprising. At Nova Adventure it’s the first thing we ask for when a team tells us it’s drowning, because it turns “email” into a list of problems that each have a fix.