Fixed-Scope Engagement

Review the record. Diagnose. Treat. Only the first step needs read access.

Every organisation that moves a workforce by air holds a record of what actually happened: who booked, who flew, who cancelled, what was refunded, what was claimed, what was posted. The record already knows where the money leaks. What it lacks is someone to read it end to end, reconcile it to the payments and postings it should match, and rank what it finds.

That is the review. It asks for read access and nothing else, and returns the ten largest recoverable opportunities, ranked by money. The findings are yours whether or not you go further. We diagnose because we can treat: the same team that reads the record runs the platform that books, rations and posts it.

Talk to us about a review

Three steps, in order

  1. Review. Read the movement and claim records as they are. Reconcile what the systems believe to what the payment gateway and the accounting system actually did, and map every booking and payment path, including the traffic current reports cannot see. Nothing is changed, no user is retrained, no process is touched.
  2. Diagnose. Rank what the reconciliation finds by money. Ten findings, each with the rows that prove it.
  3. Treat. If you choose, the findings are treated in priority order: each becomes a rule the engine executes, a control at the booking or claim chokepoint, or a decision queue for the operations team, in the same system that holds the record. Dated, logged, reversible; the change log is available to an evaluator on request.

The named reviews

Most engagements start with one and widen once the first ranked list is in hand.

  • Unmanaged travel spend. Flights booked by phone, hotels arranged by email, charters organised by an operations desk: travel that never touches a booking tool never triggers budget approval, and surfaces in the accounts after the money is spent, if at all. Managed-spend reports also value leased travel at zero, because a leased seat has no fare. The review quantifies the fraction that bypasses approval and where it enters.
  • Cancellations and refunds. When staff pay by card for seats on company aircraft, the organisation is the airline and holds a real liability. The review reconciles refund state to the payment gateway, ranks the drift, and designs the decision queue that should come before any automation. The guide sets out the model. Airline revenue-management and revenue-accounting experience sits on this review.
  • Seat utilisation. Leased and company-operated fleets run at 40% utilisation on average; policy-governed allocation lifts it to 90%. The review measures where you are and which levers move it. The benchmarks explain the levers.
  • No-shows and empty seats. A no-show the night before costs the operation a seat; one a month out costs nothing. The review finds which no-shows cost seats and what penalties, quotas, and go-show refills would recover. The economics are set out separately.

Why the record, not a model of you

Process-mining tools reconstruct a process from other systems' event logs and reason about the reconstruction; the comparison is on its own page.

Where the operation already runs on UnityTrip, every movement passes through one booking number and every dollar through one claim number or gateway reference, the two chokepoints, so the record is complete and policy-annotated before the review begins and there is no reconstruction step. Where it does not, the review reconciles the records you hold, and the gaps between them are usually the first findings.

We do one domain. The review is drawn from running booking, approvals and expense claims in production in energy and resources.

Enterprise AI is credible at the end that can act on the record; the platform page sets out why.

AI proposes. Experience disposes.

The review uses AI to read the record and to rank candidates. It does not use AI to decide what is real. Every candidate finding is checked by people who have run workforce aviation operations, with a revenue-accounting view, before it reaches the ranked list. A confident wrong recommendation is the known failure mode of the tool; the method is built around catching it, and the record shows both the proposal and the decision.

The request is rarely the need

Most engagements begin with a request that sounds complete: automate the refunds, add a report, tighten a rule. The record shows what happened. It does not show what was wanted.

One finding, anonymised

An operator sells spare seats on its own aircraft to staff who pay by card. The request was to automate refunds on cancellation.

The review found the request assumed a refund per booking, while the record showed refunds happen per passenger and per segment, so one traveller cancelling the return leg of a party of four is a different sum from the booking total. It also found the refund ledger had drifted from the card gateway in three ways: an event recorded with no amount, a refund whose status never advanced after the money had moved, and a refund made outside the system and never confirmed back.

The treatment, in order: a value line frozen per passenger per segment at booking, so every later refund is arithmetic on a known figure; a decision queue in which a person settles each refund the rule cannot, with a non-binding recommendation; and only then the automated rule. Five days to establish that, working with the people who run the operation; one to deliver it.

Built as first asked, the work would have been quick and wrong. No event log held the answer. It came out of conversation.

So the diagnosis is done with your operations staff, not only about them. They know why the workaround exists, which rule gets gamed, who gambles on a seat coming free, and what the joke about the system is really saying. All of it is evidence.

Where the review came from is told in The capacity to surge.

Agreement before treatment

A change to how people travel lands in one of two ways. Some changes are a management decision: the organisation has decided, and the work is to carry the decision through cleanly. Others only hold if the staff who run the operation accept them, because people block what they did not agree to. Part of the diagnosis is knowing which kind of change is in front of you.

Where agreement is needed, it has to be informed. The people affected see the evidence, join the analysis and the testing, and understand what will change for them before it does. That takes longer, and it is why the change holds.

A confidentiality agreement comes first. The review works from read access and from extracts you approve; what we hold is returned or deleted at the end, as the agreement sets out. The security page and the privacy policy describe how the platform and the website handle data.

What we hold to

One north star: the travellers, and the staff who support them. The commercial arrangement follows from that and from the conversation.

We agree the why and the what with you. Where the record shows a real problem for travellers or support staff, we fix it, whether or not the fix is funded, provided it is worth it for every side. When we carry the cost, we decide the how. When you want to set the how, you fund the change and carry its risk. Most work sits somewhere between, by agreement. We choose clients who can work this way.

Common questions

What is the Travel Operations Review?

A fixed-scope, priced engagement in three steps. Review: UnityTrip reads your movement and claim records, read access only, and reconciles what your systems believe to what actually happened. Diagnose: it returns the ten largest recoverable opportunities ranked by the money involved, each with the evidence behind it. Treat: if you choose to go further, each finding becomes an executable rule, a control, or a queue in the platform, with a dated change log. The findings are yours whether or not you proceed.

What data does the review need?

Read access to the records that already exist: bookings, movements, cancellations, claims, and the payment or posting records they reconcile to. For an operation not yet on UnityTrip, that usually means exports from the booking tool or agency, the charter or fleet schedule and manifests, the claims or expense system, and the card gateway and ledger; the review reconciles them to each other. Nothing is changed, no user is retrained, and no process is altered during the review.

How long does a review take?

It depends on your scope and on how quickly your organisation can grant read access to the records that need to be seen. In our experience the pace is set by the organisation, not by the analysis; we would rather do what is needed than pretend it is simple. The proportion in one recent review was five days to establish what was needed and one to deliver it.

Does the review use AI?

Yes, to read the record and rank candidates; no, to decide what is real. AI proposes and domain experience disposes: every candidate finding is checked by people who have run workforce aviation operations, with a revenue-accounting view, before it reaches the ranked list. A confident wrong recommendation is the known failure mode of the tool, and the method is built around catching it.

How can finance see travel on our own aircraft when the seat has no fare and nothing goes through the booking tool?

Not from spend reports. A seat on transport the organisation already pays for has no fare and no reimbursement, so spend reports value it at zero, and a charter organised by an operations desk never triggers budget approval. Visibility comes from the record. When no traveller can use a travel service without a booking number and no provider is paid without an approved claim number, every movement passes through one booking number and every dollar through one claim number or gateway reference, and approved claims post to the finance system. The unmanaged-spend review measures the fraction that bypasses approval, including the travel valued at zero. Definitions: unmanaged travel and zero-dollar travel.

How is the review priced?

As a fixed-scope engagement, priced from US$15,000 against the scope agreed at the start. It is a defined ask, a defined output, and no obligation to continue.

Start With the Number

Before choosing a platform, read the record.

Talk to us Read the cancellations and refunds guide →