Sofia / Response
The Real Cost of Responding Late to a Hotel Enquiry
A conservative model for valuing enquiries, measuring response and deciding what to automate.
Why speed alone is not enough
A fast but incomplete reply can simply move the work to the next message. An accurate answer that arrives after the guest has chosen cannot recover demand. The useful measure combines time, quality, next action and outcome.
The aim is not to claim that every lost minute equals a lost booking. It is to expose a process currently scattered across email, phone, WhatsApp and platforms.
Define a valid enquiry
Separate availability, quote, group, event, general information, complaint and spam before measuring. Each category has a different urgency, potential value and owner.
A valid enquiry needs a response channel and understandable intent. If dates, people or preferences are missing, the first goal is to obtain the minimum data without repeating resolved questions.
The observable cost
Direct cost includes human time spent reading, looking up information, writing, correcting and following up. Indirect cost includes interruption, duplication and enquiries that nobody owns.
Measure these components with real samples: weekly volume, average minutes by category, escalation rate and the number of steps before a decision.
Opportunity cost
The probability that an enquiry becomes revenue is the sensitive part and must not be shown as certainty. Model conservative, central and favourable scenarios, then replace assumptions with actual data.
The base formula is valid enquiries multiplied by average booking value and conversion probability. Estimated leakage applies the observed difference between enquiries handled inside and outside the property's threshold.
Modeled business scenario
Disclosed example: a property receives 80 valid enquiries per month. Assumed average value is EUR 420 and central conversion is 8%. This is not a Meridia client or a guaranteed result.
Expected enquiry value is EUR 2,688. If 25% have no next action and only half are recoverable, the modeled value to investigate is EUR 336. The decision comes from the formula, not the isolated number.
Expected monthly enquiry value by scenario
How to read these numbers
Disclosed scenario: 80 valid enquiries per month, 420 EUR average value. It does not represent an actual client.Use three thresholds
A simple question may receive a first answer within minutes. A group may need verification. A complaint needs escalation. One universal threshold rewards fast but unsafe responses.
Define time to ownership, time to first useful response and time to decision or escalation. Each threshold needs an owner and a stated service window.
The three operating thresholds
| Threshold | What it measures | Example target |
|---|---|---|
| Taking ownership | The enquiry has an owner and a category. | Within minutes during service hours |
| First useful reply | Complete answer or request for the missing detail. | Within the declared window per category |
| Decision or escalation | Groups, events and complaints reach the right person. | Within the working day |
How to read this data
Each threshold has an owner and a declared service window.
What Sofia can do
Sofia can identify language and intent, retrieve approved knowledge, request missing data, record the opportunity and propose the next action. She can also preserve context across connected messages and channels.
The benefit is not only after-hours response. It is keeping routine demand from consuming the attention needed by exceptions and higher-value enquiries.
What Sofia must not decide
Availability, special prices, refunds, legal commitments, conflicts and safety information need a structured source or authorised person. If the source is missing, the correct response is to state the limit and route the case.
A useful system does not hide uncertainty. It records missing information, risk and who must approve the next step.
Minimum data
Start with four weeks of volume, channel, category, ownership time, first useful response, outcome and value where available. There is no need to capture every word or personal detail.
The record must include legal basis, recipient preference and do-not-contact status where relevant. Measurement does not justify indiscriminate collection.
A fourteen-day test
Measure without changing the process during week one. In week two, introduce classification, approved responses and an escalation queue. Compare time, completeness, follow-up and unowned enquiries, not just message volume.
Stop if human work increases, sources are unreliable or exceptions have no owner. Technology that creates another inbox has not solved the problem.
Reading the outcome
A credible improvement means fewer unowned enquiries, less repetition and more recorded outcomes. Economic conversion needs more time and enough observations.
With low demand, service quality may be the main benefit. With high demand, value may come from concentrating people on complex cases.
Recommendation
Measure before automating, model with visible assumptions and connect each category to a rule. After-hours response matters only when it leads to a verifiable next action.
Meridia screening observes the public journey. The paid audit builds the model from property data and decides whether Response deserves a limited pilot.
Official sources and notes
Sources were verified on the stated date. Scenarios and formulas remain explicitly disclosed.
- Eurostat, Tourism statistics2026-07-11
- Eurostat, E-commerce statistics2026-07-11