Skip to content
Back to Blog

Travel agent resources7 min read

Measuring whether AI improves your travel planning workflow

Measure AI-assisted planning against a fair baseline using full-task effort, quality checks and cautious interpretation of a fictional comparison.

Editorial illustration of travel-planning work: measuring whether ai improves your travel planning workflow

Measure the complete task, including preparation, review, correction and source verification, against a comparable baseline. Keep quality and active effort separate. A faster draft does not establish an improvement if it leaves essential research unfinished or adds a material error that the adviser must resolve later.

Nowah can support conversational research, editable day plans, map/timeline review and flight and hotel searches. To evaluate its place in your agency, define which of those tasks you are changing and what remains in your existing process. Use your own observations; do not turn an illustrative example or a tool’s confident output into evidence of real performance.

Define an improvement that matters to your agency

Choose a specific task and outcome. You might want a coherent route outline with fewer corrections, a clearer hotel comparison, or more consistent identification of unresolved questions. “Use AI more” is an activity, not a service improvement.

Choose a concrete Nowah planning outcome to measure, such as a usable route comparison. Our 20-tab planning problem provides context for investigating fragmented research.

Write an acceptance rule before collecting results. For example: preserve required dates and rooming, complete the same source checks, and reduce active effort on a comparable route task without increasing material errors. Treat unresolved essential suitability as a stop condition rather than a low score that attractive formatting can offset.

Use a narrow scope first. A study of itinerary outlines cannot establish the value of AI-assisted purchasing or in-travel support. If Nowah is used only for the initial outline, count the later review work needed to reach the same deliverable as the baseline. Otherwise, you are comparing different amounts of completed work.

Choose a fair baseline and sample

Use representative fictional briefs or appropriately authorised cases, with consistent detail. Keep the travel dates, party, budget scope, required output and verification obligations comparable. Record complexity rather than mixing easy city breaks with demanding group rooming and assuming their times mean the same thing.

If you repeat the same task, your familiarity can affect the second attempt. Alternate the order where practical, or use matched briefs and record the difference. Keep the protocol clear enough that another adviser can understand what was done. This is an operational evaluation, not a claim of a scientific benchmark.

Do not make an earlier weak process the baseline simply to produce a flattering result. Compare with the method your agency would actually use. Keep any new template, training or scope change visible so you do not attribute every improvement to Nowah when several things changed at once.

Record active effort and completion separately

Track preparation, research, review, correction, supplier checks and transfer into the normal deliverable. Include rework discovered after the first draft. Record waiting for a response or supplier separately from active labour, since both may affect delivery but they are different measures.

Record Nowah research, itinerary editing and human review as separate work stages. The time-tracking guide helps keep the measurement complete.

For quality, record material date or quantity errors, unsupported factual claims, missed essential requirements and unresolved items. Distinguish an error from a legitimate client preference change. A request for a different destination after presentation is not necessarily evidence of poor original work.

Keep the actual evidence that supports each finding. A new Nowah itinerary cannot prove that a historical hotel offer was available or that an earlier adviser should have chosen another activity. Use the original brief, supplier evidence and delivered version to assess the case. Use AI-generated alternatives only as new planning experiments.

Worked hypothetical example: six matched route tasks

Consider an invented training dataset comparing six baseline route tasks with six matched AI-assisted tasks. The same checks and output scope apply: reviewed route outline, night arithmetic, transfer questions and recorded source-verification needs. These numbers are fictional and do not report measured Nowah performance.

Use equivalent fictional briefs for the matched Nowah tasks before comparing the results. The itinerary review rubric supports a consistent quality assessment.

Matched task

Baseline active minutes

AI-assisted active minutes

1

90

85

2

105

90

3

120

110

4

95

90

5

110

105

6

100

95

Total

620

575

The difference is 45 minutes across six tasks, or 7.5 minutes per matched task on average. Baseline average effort is 103.33 minutes; AI-assisted average is 95.83 minutes. Keep the individual rows because the difference is not identical for every case.

Suppose the fictional review also finds two material issues during checking in each condition, all corrected before the defined deliverable is completed. Record the issues and correction effort; do not declare the tools equally reliable from this small count. Check whether the kinds of issues differ and whether all required source work was actually finished.

The agency might decide the narrow workflow deserves a further trial. It cannot claim that every enquiry will save 7.5 minutes, that bookings will increase, or that future client errors are eliminated. The example illustrates interpretation of a small operational sample, not proof of a product benefit.

Copyable measurement register

Use this worksheet in your agency’s normal review system:

Protocol

  • Task and deliverable being compared:
  • Baseline method and AI-assisted method:
  • Fictional or authorised brief reference:
  • Complexity and required checks:
  • Order of attempts and allowed corrections:
  • Acceptance rule and material stop conditions:

Per-task observation

  • Active minutes by stage:
  • Waiting time and cause:
  • Output scope actually completed:
  • Material errors found and correction effort:
  • Unsupported claims or essential gaps remaining:
  • Source evidence used for review:
  • Final status: complete, incomplete or excluded with reason:

Interpretation

  • Matched difference and range:
  • Changes other than tool use:
  • Missing data and limitations:
  • Incremental costs on a consistent basis:
  • Decision, owner and next review date:

Record excluded or incomplete tasks rather than deleting difficult cases from the result. If the baseline or candidate did not finish the same work, explain that before calculating averages. A short table should make the comparison clearer, not conceal unequal scope.

Copyable Nowah task prompt for the trial

Help a travel adviser develop a fictional Portugal route outline for two adults: nine hotel nights,9–18 October 2027, Lisbon four, Coimbra two and Porto three. Origin Toronto; flight travel dates require separate checking. Ground budget EUR 3,200 excluding international flights and personal spending. Priorities: food, culture, moderate walking and light transfer days. Produce a day outline, base check-in/out table, key transfer questions and hotel search criteria. Mark unverified schedules, prices and suitability. No purchase is authorised. I will review night arithmetic and original supplier evidence before treating the outline as client-ready.

Use the same task definition in the baseline method and record what work is done. Review Nowah’s day plan and map/timeline, and include any relevant search activity in the measured effort. The Nowah features page describes supported capabilities; it does not provide your agency’s measured results.

Interpret cost and capacity cautiously

Freed active minutes are a capacity observation, not automatically cash savings or additional revenue. In the hypothetical example,45 minutes is 0.75 hours. At an assumed internal time value of EUR 40 per hour, that represents EUR 30 of time value across the sample. It is not necessarily a reduction in payroll or overhead.

If evaluating cost, include actual incremental tool costs and any additional review effort on the same period and allocation basis. Do not multiply a small sample by all annual enquiries without examining service mix, seasonality and whether those enquiries use the same workflow.

A useful result may be more consistent questions rather than less time. Record that outcome directly instead of forcing every benefit into a revenue claim. Nowah can be worth evaluating for a particular planning stage even when the evidence does not support a broad productivity promise.

Check where effort moved as well as whether its total changed. An outline may take less drafting time but more source verification, or a clearer early brief may prevent later correction. Keep the stage breakdown so those changes are visible rather than crediting the first fast response for work completed elsewhere.

Retain the denominator when reporting quality. Two issues across six tasks is different from two across sixty, and several issues may occur in one task. Record both affected tasks and individual findings where useful. If no client-facing errors were observed in a small trial, report that observation without promising future error-free work. Decide whether another sample or a different service type would change the confidence of your operational decision before expanding the workflow.

Decide whether to keep, adjust or stop

Choose a limited next action from the evidence. Keep the workflow for the task that met your criteria, adjust the prompt or checking process where gaps appeared, or stop using it for a task whose requirements were not met. Assign an owner and a review date.

Continue human checks on actual client work. Supplier availability and conditions govern bookable offers, and the client must approve the specific purchase before payment. A successful draft exercise does not remove those obligations.

Use Nowah as a planning companion within a process you can measure. The strongest conclusion names the task, the observed result and its limits, giving your agency a concrete basis for the next decision.

Share this article