Which result is the hotel trying to measure?
Define the claim before choosing the baseline. The project might need to measure a particular equipment boundary, evaluate a whole-property change, or understand billing effects. Those questions can require different data and methods.
Identify the measure, affected equipment, relevant fuels, and service requirements. A retrofit described only as energy efficiency leaves the comparison boundary unclear. Write down what the result should establish and what it will not be able to isolate.
The efficiency hub connects project choices. This guide’s deliverable is a versioned measurement plan that survives the equipment change and lets another reviewer reconstruct the calculation, including its assumptions and limits.
What does the measurement guidance establish?
DOE’s measurement and verification guide, Version 5.0, dated September 2024, discusses measurement boundaries, baseline methods, and adjustments. Its federal performance-project context is useful technical guidance, not an automatic private-hotel contractual requirement.
Apply the principles to the actual decision. The measurement plan should identify independent variables and relevant static factors rather than assume every change can be explained by one percentage. Qualified reviewers should evaluate whether the method and data support the intended claim.
DOE’s energy management information system capabilities distinguishes analytics, measurements, and models. A dashboard’s calculated result is not necessarily direct measurement. Retain the underlying method so the hotel can tell whether it is reviewing a meter total, estimated load, or adjusted baseline.
What should the baseline plan record?
| Plan field | Evidence or decision | Reason to preserve it |
|---|---|---|
| Measurement boundary | Equipment, meters, and fuels included | Avoid changing the comparison denominator |
| Reference period | Actual dates and service intervals | Explain the operating context |
| Variables and static factors | Model structure and supporting records | Identify relevant changes |
| Adjustments | Approved method and evidence | Make the result reconstructable |
| Concurrent activity | Other projects, outages, and operating changes | Limit unsupported attribution |
| Data quality | Missing values, clocks, units, and estimates | State uncertainty clearly |
Record responsibility for collecting each item before work begins. The project can destroy access to useful baseline evidence if measurements are deferred until after equipment removal. Identify that dependency in the construction schedule.
How should weather and occupancy be treated?
Choose a model that fits the actual system and available evidence. Weather can affect cooling or heating, while occupancy and activity can affect service demand. Neither should be included merely because a spreadsheet has a convenient column.
Occupancy treatment depends on the plan. It may be a modeled independent variable, or a material unexpected change in an assumed static condition may require non-routine treatment. Do not label every occupancy difference routine or non-routine without examining the defined method.
Keep the operational record beside the numerical data. Hotel room closures, event patterns, changed hours, or major service changes can explain differences that a simple occupancy total misses. The plan should identify what evidence is needed and how material changes will be reviewed.
How should baseline preparation be sequenced?
- Define the intended claim and measurement boundary before project alteration.
- Select a reference period and assess available data quality.
- Document variables, static assumptions, and relevant operating records.
- Approve adjustment methods and treatment of concurrent changes.
- Preserve raw data and collect missing evidence before access is lost.
- Version calculations and review the final result against the approved plan.
The submetering priorities guide helps when the existing boundary cannot answer the question. Additional measurement should resolve a named uncertainty, not merely produce more dashboard points.
Include the acceptance criteria for the data itself. Verify units, clock alignment, service-period boundaries, and unexplained discontinuities. A large dataset is not a valid baseline if its fields are inconsistent or the reviewer cannot tell which periods were estimated.
How can a worked comparison remain properly qualified?
Assume, only for illustration, a modeled baseline adjusted to the reporting conditions is 120,000 kWh, while recorded consumption on the same fictional boundary is 110,000 kWh. The arithmetic difference is 10,000 kWh.
Those invented values demonstrate subtraction, not a real hotel result. Whether the difference can be attributed to the project depends on the approved model, data quality, concurrent changes, and review. The calculation alone does not establish annual persistence or a financial outcome.
Do not multiply that difference by an undated generic rate and call the result verified bill savings. Price and demand effects need their own actual account terms and comparison. Keep the physical result, modeled attribution, and financial scenario in separate fields with their evidence.
What can a hotel pilot example contribute?
The DOE-hosted Marriott measurement and verification case describes a historical public process for testing technologies with hotel operating context. It supports considering representative conditions and service requirements when designing a pilot.
It is not this site’s client result and does not establish transferable savings. The hotel should use its own equipment boundary, climate, operating records, and measurement plan. Avoid importing a public case’s outcome as an assumed benefit in a different property.
A pilot can also reveal that a proposed measurement method is inadequate. If the result is dominated by an unrecorded operational change, preserve that finding and revise the plan. A credible review can conclude that attribution remains unresolved rather than manufacturing a favorable project percentage.
How should concurrent work and outages be recorded?
Create an event log identifying dates, affected boundaries, and available evidence. Equipment replacements, room closures, construction interruptions, or other changes may alter the relationship between the reference and reporting periods. Do not exclude them without a documented reason.
The solar and roof coordination guide illustrates why availability and construction records matter. Lower generation during an interruption is a different explanation from poorer equipment performance.
The sensor and schedule guide addresses another measurement hazard: changing inputs or controls while assuming the data series is unchanged. Record those corrections and have the reviewer assess their effect on comparability before merging periods.
What should calculation review check?
Have another reviewer trace selected results back to the raw record and defined method. Check the sign of differences, units, date ranges, estimated values, and any exclusions. A formula can be arithmetically correct while using the wrong boundary or period.
Preserve the original calculation and a clear reason for each revision. A model update should not erase earlier assumptions. Record whether the update responds to new evidence, corrected data, or a changed analytical method.
If supplier or broker analysis is involved in the financial layer, disclose compensation and service scope. An adviser may help interpret bills without validating the physical model. Keep that responsibility visible so procurement commentary does not become mistaken for engineering verification.
What makes the baseline and final result reviewable?
The final file should include the measurement plan, raw data location, variable definitions, adjustments, event log, calculation versions, and review decisions. State which quantities were measured and which were estimated. Preserve unresolved evidence gaps.
Report results only within the demonstrated boundary and period. A short evaluation does not establish future performance, and a whole-property difference does not automatically isolate one measure. Explain the practical limits in the same record as the result.
A useful baseline enables a bounded decision: the project effect is supported under the defined method, more measurement is needed, or attribution remains unresolved. It should not turn a lower invoice, an attractive chart, or invented arithmetic into a claim that every observed difference came from the hotel project.