Hotel sustainability software, capability by capability

The diagnostic takes about ten minutes and asks 45 questions across eight areas. Each area carries a fixed weight out of 100, and the weighting reflects operational significance for a UK hotel rather than how much attention a topic currently gets: energy and waste carry the most because that is where a typical property's impact and spend actually sit.

Every point on the scale has words attached

Questions are answered 0 to 5, and each of those six values has written wording specific to that question. On the sustainability-policy question, 5 reads “public, approved, and reviewed annually” and 2 reads “draft or informal policy only”. Nobody is asked to rate themselves out of five and guess what a four is supposed to mean, which is what makes two different managers at the same property arrive at the same answer.

Questions that do not apply to you are never asked

Two filters run before you see anything. The first is your property profile — no pool means no pool questions. The second is branching on your own earlier answers: if you say you have no sustainability targets, you are not then asked how you track progress against them. A question filtered out is never shown, never scored, and never counted toward completion, so a hidden question cannot quietly become a zero.

Not Applicable is a real answer with real consequences. It is excluded from both the numerator and the denominator of its area, and if an entire area turns out not to apply, its weight is redistributed proportionally across the areas that remain. The scale still runs 0–100. A property with no grounds is not marked down for having no grounds.

The same answers always produce the same score

Scoring is a pure function: no network calls, no randomness, and no AI anywhere in it. Run it twice on the same answers and you get the same number, on any machine, on any day.

The scoring model is versioned, and each published version is frozen — the files are not edited again. Every stored score records which scoring version and which question-bank version produced it, and recomputing a historical score resolves that exact ruleset rather than today's. Asking for a version that does not exist raises an error instead of quietly falling back, because a wrong-but-plausible recomputation published as reproducible is far worse than a failure. That is what lets a score from last March still mean what it meant in March.

Scores are written by the server, never by your browser

The table that holds scores has no client insert policy at all. The only path to a row in it is a server action that re-reads your raw answers from the database and recomputes the score itself. Nothing a browser sends can become a score — not yours, and not anyone else's using your session. It reads like an implementation detail; it is the reason the number is worth showing to a third party.

What the questions are drawn from

Every question is tagged with the frameworks it evidences: GSTC Hotel Standard, Green Key, Hotel Sustainability Basics, ISO 14001 and UK legal requirement. Where a question reflects a legal obligation rather than good practice, the tag says so, and gaps there are treated differently from maturity gaps everywhere else in the product. Questions also carry the UN Sustainable Development Goals the area contributes to — as context only. The SDGs have no criteria to conform to, so presenting them beside Green Key or GSTC as equivalents would be a claim nobody could substantiate.

The result is placed in one of four maturity bands, and under the current ruleset the boundaries are:

• 90+Leading
• 70+Advanced
• 40+Developing
• 0+Early Stage

Those boundaries are deliberately strict. This is a self-assessment, self-reported instruments skew high, and “Leading” is a label a hotel will put on its website — where an unsubstantiated superlative is precisely the kind of claim the CMA Green Claims Code pursues. A threshold that is harder to reach is one we can stand behind when it is quoted back at us.

The Hotel Sustainability Basics are a published minimum set of hotel sustainability actions, grouped into Efficiency, Planet and People, and split between what is expected in the first year and what is expected by year three. Each criterion is held with its requirement text and the specific assessment questions that evidence it, so a status is always traceable back to an answer you gave.

Four statuses, including one for silence

Every criterion resolves to meets, partially meets, does not meet or not assessed, each with a written rationale saying why. That fourth status matters more than it looks. Where no question in the current bank covers a criterion, the page says so and claims no position — a criterion silently omitted from a gap report reads as a pass, and it is the easiest way to flatter a property by accident.

Each criterion also shows whether any of its supporting questions has a document behind it. That is reported separately from the status, because a criterion can be met on the answers and still be completely unevidenced — and the difference between those two states is the difference between being ready and feeling ready.

Priorities ranked by rules, not by opinion

The top priorities are computed deterministically, in a fixed order of precedence. Minimum-standard gaps come first, because they block readiness regardless of what closing them does to your score. Everything after that is ranked by headroom multiplied by the area's weight — the points genuinely still on the table — and ties are broken by question-bank order so the list does not reshuffle between visits. Each item carries the reason it ranks where it does.

There is an AI-written commentary layered on top of this, and it is labelled as guidance. It is strictly commentary: switch it off, run out of budget, or have it fail, and the rankings above are unchanged. The rules are the source; the narrative is a convenience.

Critical gaps are flagged, never deducted

A property can reach a respectable overall score while failing something fundamental — the weighting rewards breadth, and breadth can hide a single hole. Two things get named separately from the number: a UK legal requirement scored 0 or 1, and a minimum-standard criterion scored 0.

They are flagged, not subtracted. A deduction would mean the same answers no longer produce the arithmetic the published weights promise, which is the one property the entire product rests on. The flag sits alongside the score and says what the score cannot.

A list of weaknesses is not a plan. The roadmap turns each real gap into an action with three things attached: when it should happen, who owns it, and why it is where it is in the order.

Effort decides the horizon

Low-effort actions land in the first 30 days, medium-effort in 60, higher-effort in 90. The plan therefore opens with work that can genuinely start this week rather than with the most impressive item on the list, which is the difference between a 30/60/90 plan and a wish list.

Ownership is part of the ranking

Each action is routed to the team that would actually do it — engineering, housekeeping, front office, procurement or management. That routing then feeds the order: work a single operational department can pick up today ranks ahead of work that needs a management decision, a budget or a meeting first. It is precedence rather than exclusion; management actions still appear, they just sit behind what is already actionable.

Impact is the points still on the table

Impact is not “which area is this in”. It is your headroom on that question — five minus your score, over five — multiplied by the area's weight. So a zero in a lightly weighted area can outrank a two in a heavy one, which is the correct ordering and something a weight-only model simply cannot express.

The full precedence order, applied in sequence:

1. Minimum-standard gaps, ahead of everything else.
2. Higher impact — headroom against the area's weight.
3. Clear operational ownership before management decisions.
4. Lower effort first within the same tier.
5. Lower raw score — the bigger gap — to break remaining ties.
6. Question-bank order, so the result is stable.

Every action carries a status you move as the work happens, and completed actions feed both the readiness gate and the delivery figures on your dashboard. As with the gap page, an AI explanation can be generated alongside the plan and is labelled as guidance — the ranking itself never depends on it.

Tasks are assigned to named people with due dates and a status, inside the same tool that produced the action. Nothing has to be re-typed into a spreadsheet that then drifts out of date, which is where most hotel sustainability plans quietly die.

Roles, and who can change what

A property has owners, managers and staff. Owners and managers assign and manage; staff see and complete their own work. Colleagues are invited by email, and where email is not configured the invitation falls back to a link you can copy and send yourself.

Overdue work interrupts you

Overdue tasks are surfaced at the top of the dashboard as blocking priorities, above almost everything else, because slipped work is the most common reason a property misses the timeline it set itself. A task board nobody looks at is a task board that does not exist.

The completion trail is itself useful

Most recognised schemes ask for evidence of continual improvement rather than a single snapshot. A dated record of who did what, and when, is exactly that — so the execution history is not just project management, it is material you will be asked for later.

Certification preparation usually turns into a document hunt three weeks before an assessor arrives. The evidence library exists so that hunt happens continuously and cheaply instead: every question that can be evidenced lists the kinds of document that satisfy it — a policy document, an org chart, a board approval, a KPI dashboard — so you are not guessing what to look for.

Files uploaded for one purpose stay available for every other one. A document attached to an action last quarter is still findable, still labelled, and still counts.

Evidence changes how defensible your score is, never what it is

Uploading a document raises your evidence coverage — verified answers as a proportion of evidence-eligible answers — and that figure is reported beside the score, never inside it.

This is a deliberate constraint, and an unusual one to advertise. If evidence moved the score, the score would gradually become a measure of how much paperwork a property had filed rather than what it does. Keeping them separate is also what allows the score to be recomputed from answers alone by someone who has never seen your documents. As a side effect, attaching a file does not rewrite the timestamp on your score — an earlier version of this product did, and silently corrupted the trend chart the evidence was meant to support.

Assurance grades, and the one we never award

Reports grade the evidence behind each criterion from A to D, and take the weakest link rather than the average — an average would let a well-documented criterion cover for an undocumented one. Grade A is reserved for independently verifiable documents, and the third-party verification that would justify it does not exist in the product yet, so it is never awarded. Leaving it unreachable is the honest position rather than a limitation to hide.

A person confirms every extracted figure

A model reads the bill and proposes the values — supplier, period, consumption, unit, cost. Every one of those is then presented to someone at the property for confirmation before anything is stored. Nothing a model read ends up in your data unchecked, which is the only version of document extraction that belongs anywhere near a figure you might later publish.

Carbon with a factor you can look up

Carbon is computed from the UK Government GHG Conversion Factors published by DESNZ, and each reading stores exactly which factor row it used: the value, the unit, the publisher, the publication year and the source URL. A carbon figure without its factor cannot be reproduced, and an unreproducible figure is not evidence.

Where no published factor exists for a utility and unit, no figure is reported and none is estimated in its place. Waste carbon is not calculated at all, because there is no single published per-kilogram waste factor — emissions depend on the material and the disposal route. A consumption value in the wrong unit for its factor raises an error rather than being multiplied anyway.

Scope labelling, and its limits

Emissions are labelled to the GHG Protocol: Scope 1 is fuel burnt on site, Scope 2 is purchased energy, Scope 3 is the value chain. Emissions whose factor predates scope labelling are reported separately rather than assumed into Scope 3 — a figure filed under a scope it was not measured against is worse than one openly unclassified.

Scope 3 here covers water and waste only. It excludes purchased goods, guest travel and construction, which are the largest Scope 3 categories for most hotels, so this is not a complete Scope 3 inventory and the product says so wherever the figure appears.

Intensity on the measure benchmarks use

Record your floor areas and emissions are apportioned between guest accommodation and function space, following the Hotel Carbon Measurement Initiative's approach — accommodation reported per occupied room-night, function space per square metre. This is a figure computed the same way from your own data. It is not an HCMI verification: HCMI has its own boundary rules, exclusions and assurance process, and EcoPulseIQ is not an assurance provider.

Record room-nights sold and intensity is computed only across the months where both a utility reading and an occupancy record exist. Dividing a full year of carbon by three months of room-nights would produce a number several times too high that still looks like a benchmark figure — which is considerably worse than showing nothing. Without occupancy, intensity falls back to a per-room measure and is labelled as not comparable to published per-room-night figures.

The dashboard briefing is built by a pure function from your own data, so the same inputs always produce the same briefing. When a manager asks why it says what it says, there is an answer, and it is the same answer tomorrow.

It opens with a headline drawn from whatever is genuinely most pressing — an overdue task, a lapsed certificate, a fallen score — followed by a stack of at most four priorities, each with the reason it is there and a link to the page that resolves it. Below that: twelve months of carbon with a per-utility split, your score trend across assessments, roadmap delivery, evidence coverage, the strongest and weakest impact areas, and which areas moved since your last assessment.

A number we cannot stand behind is not shown as a number

Caveats travel with the figure rather than being dropped for a cleaner-looking card. If some readings were estimated rather than actual, it says so. If your data covers four months, it says this is not an annual footprint. If half your readings have no cost recorded, it says the total is lower than your actual spend. If intensity is normalised by room count because we do not hold your occupancy, it says that figure is not comparable to published per-room-night benchmarks.

Movement below two percent is reported as steady rather than as a trend, because below that it is noise. Distance to the next maturity band is read from the ruleset your score was stamped with, not from today's thresholds — so the dashboard cannot quote a boundary that no longer applies to your score.

Your baseline, not a peer group

A separate progress view charts electricity, gas, water, waste and renewable share over time against your own earlier readings. Against your own, specifically: we do not hold a cohort large enough for peer comparison to be statistically meaningful, and publishing one that was not would be misleading. That limitation is stated in every report rather than left for you to discover.

Each scored assessment produces a report that prints: your score with each area's arithmetic traced back to its inputs, conformity against the minimum standards, the evidence attached to each criterion with its assurance grade, and an explicit list of everything that could not be substantiated. That last list is the part most reports omit, and the part an assessor reads first.

The reproducibility export

Every report ships with a structured export containing each input, rule and formula behind it. The full ruleset is embedded — every weight, every band threshold — rather than referenced by version number, because an export that only says which version was used is useless to a consultant who cannot look that version up.

Alongside the ruleset it carries:

• your raw answers, exactly as submitted;
• each area's calculation written out as an expression beside its result, so it can be checked line by line;
• how Not Applicable is handled and how weight redistribution works, in words;
• every emission factor used, with its publisher, publication year and source URL;
• each carbon calculation as an expression, and a note wherever no factor existed;
• a numbered procedure for verifying the whole thing by hand.

It also carries its own limitations, which is the part that is usually left out: answers are self-reported; this is a readiness assessment and the bands do not correspond to any scheme's award levels; there is no peer benchmarking; waste carbon is not calculated. The point of the export is to turn “trust our number” into “check our number”, and a document that hides its own weaknesses cannot do that.

What AI did, and what it did not

Every report carries the same disclosure: no AI system participated in calculating the score or in determining conformity with any criterion. AI is used in exactly two places — reading values from uploaded documents, where a person confirms every value before it is stored, and writing plain-language explanations of results that have already been calculated. AI-written narrative is labelled as guidance wherever it appears.

A gate a manager can click past is not a gate, it is a suggestion. This one refuses, and it refuses for reasons it names.

Four things block readiness:

• having no submitted, scored assessment at all;
• any minimum-standard criterion that is partially met or not met;
• any roadmap action addressing a minimum standard that is not complete;
• any evidence-eligible answer with no supporting document behind it.

Criteria the question bank cannot evaluate are raised as recommendations rather than blockers. Blocking a property on something we are not in a position to assess would make the gate unpassable through no fault of theirs — but an assessor may still ask about it, so it is shown and flagged as something to be ready to speak to.

The completion percentage counts mandatory items only, so 100% is reachable and means something when you get there. The verdict is recomputed every time the page is opened rather than read back from the last snapshot: a stale pass sitting above a button that then refuses would be worse than showing nothing.

What passing the gate does not mean

It does not mean a property will be certified. Certification is awarded by an independent assessor applying their own judgement, and we neither make nor influence that decision. Passing the gate means we have nothing left to find — which is the only thing we are ever in a position to assert, and the only thing this page claims.

Monitoring tracks the three things in a sustainability programme that have dates on them: certifications and their renewal dates, evidence documents and their review dates, and public sustainability claims and when each is next due to be re-checked. Each is bucketed by how close it is to lapsing.

Items that have already lapsed are surfaced on the dashboard ahead of readiness blockers, deliberately. A readiness blocker is work not yet done; an expired certificate or a live claim with nothing behind it is a problem you already have. A scheduled job raises alerts for overdue work and missing data, and a weekly digest summarises what changed.

Related, and free to anyone: the green claims checker screens your published wording against phrasing that UK regulators have been most active on under the CMA Green Claims Code and the DMCC Act 2024. It runs in your browser, needs no account, and is pattern-matching against known-risky phrases — useful as a first pass, and not legal advice.

Not a certification body. This assessment benchmarks your practices and helps you prepare for recognised frameworks. It does not constitute certification, and the maturity bands do not correspond to any scheme's award levels.

Not a substitute for reading the criteria. Scheme criteria belong to the schemes, and they get revised. We map questions to their themes rather than reproducing their requirement sets, and we point you at the publisher.

Not an audit, and not assurance. Answers are self-reported. The evidence coverage figure tells you how much of a score is supported by documents rather than by assertion — it does not verify those documents, and we are not an assurance provider.

Not a benchmark against other hotels. Every trend in the product is measured against your own baseline. A peer comparison drawn from a cohort too small to be statistically meaningful would be worse than none.

Not a promise about outcomes. Nothing here guarantees a result, a saving or an award. Certification decisions are made by independent assessors applying their own judgement.