Three steps, in order
- 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. Nothing is changed, no user is retrained, no process is touched.
- Diagnose. Rank what the reconciliation finds by money: the unmanaged fraction, the refund drift, the seats flown empty, the no-shows that cost a seat. Ten findings, each with its evidence, each checked by people who have run the operation before it reaches the list.
- Treat. If you choose, each finding 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.
Review and diagnose are how we observe and orient; the treatment design is where we decide, with you; the platform is how we act. The three steps are the engagement. The platform is what makes the advice credible.
What you get
- The ranked ten. The largest recoverable opportunities in your operation, ordered by money, with the rows that prove each one.
- A reconciliation. Where the platform, the gateway, and the ledger disagree, by how much, and why: the number your organisation has never measured.
- A flow map of every booking and payment path, including the offline, phone, and email traffic that current reports cannot see, and the leased and chartered capacity they value at zero.
- A treatment design for each finding: the rule, the control, or the queue, and where it belongs. Enforced end to end, the control is mechanical: for business bookings, no payment executes without an approved claim number.
- An implementation path, if you choose to take it, grounded in what has already run in production rather than a proposal deck.
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
Some tools promise to find and fix problems across a whole enterprise. They reconstruct each process from the event logs of systems built for other purposes, then ask a probabilistic model to reason about the reconstruction. Both steps guess. A digital twin is only as good as the input it receives, and the input was never designed to describe the process.
Workforce travel is different, and that is the whole point of choosing it. 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, wherever they sit, and the gaps between them are usually the first findings. Either way, treatment executes in the platform that then holds the record, so a finding does not have to be handed to another product to become a fix.
We do one domain. UnityTrip runs booking, approvals, and expense claims at enterprise scale in energy and resources, where travel once considered too complex to automate now runs as policy-governed workflow. Workforce aviation logistics is the beachhead, the hardest case, deliberately chosen. The review is drawn from that production experience, not from a claim to understand thirty industries from the same twin.
Enterprise AI is credible at the end that can act on the record, not at the crowded end that only diagnoses; 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: it looks right, it reads well, and acted on unchecked, it costs real money. 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, and the two are often different.
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.
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; nothing is changed, and 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, and it has to be worth it for every side.
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. Perfection is not the aim: part of the craft is balancing the operational and the financial for outcomes that are imperfect and better.
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. The pace is set by how quickly your organisation can grant access, not by the analysis.
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 have implemented comparable controls in production quickly, and 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.
Is this process mining?
No. Process mining reconstructs a process from event logs in systems built for other purposes, then reasons about the reconstruction. A UnityTrip review reads the travel records directly. On the platform, every movement passes through one booking number and every dollar through one claim number or gateway reference, so the record is complete before the review begins; elsewhere, the review reconciles the records you hold rather than inferring a process from logs. The treatment executes in the same system that holds the record.
Which reviews are available?
Four are named: unmanaged travel spend (how much bypasses budget approval and where it enters), cancellations and refunds (reconciling refund state to the payment gateway and ranking the drift), seat utilisation on leased and chartered capacity, and no-shows and empty seats. Most engagements start with one and widen once the first ranked list is in hand.
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; the largest unmanaged category is usually the one nobody counts. 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, bounded engagement quoted against the scope agreed at the start. It is priced so that it is a purchase decision with a deliverable rather than a favour with a liability: a defined ask, a defined output, and no obligation to continue.
What happens after the review?
The findings are yours either way. If you proceed, each finding becomes platform configuration or a queue in the operations tool, in priority order, with a dated change log an evaluator can see on request. UnityTrip onboarding is configuration, not a migration project.
What if we decide not to proceed after the review?
Then you have a ranked list of your ten largest recoverable opportunities, with the evidence, and no obligation. The decision to treat is yours, and deciding not to is a legitimate outcome of a review. Most organisations act on some findings and not others; that is what the ranking is for.