Evidence and approach

How we reach a conclusion, and how you can check it

Our work is analytical. We assemble evidence, separate what has been demonstrated from what has been assumed, and show the reasoning so a decision-maker can interrogate it rather than take it on trust. This page sets out how we weigh evidence and includes three worked examples of the reasoning.

The evidence hierarchy

Not all evidence carries the same weight. Where a conclusion rests on a weaker tier, we say so, because that is what tells a reader how much the conclusion can bear.

  1. Measured on the operation

    Data from your site, plant or records — meter readings, production figures, test results, purchase and dispatch data. Strongest, because it describes the actual case rather than an analogous one.

  2. Contracted and documented

    Signed agreements, supplier specifications, quotations, capacity offers and consents. Evidence of what has been committed, which is what a project can actually rely on.

  3. Published official sources

    Regulation, official guidance, national statistics and authority publications. Reliable for the framework and the general position; it does not establish what applies to your specific case.

  4. Sector and market reference

    Industry reporting, trade data and comparable published figures. Useful for orientation and sanity-checking, never a substitute for your own measurement.

  5. Management estimate and assumption

    The view of people who know the operation. Legitimate as an input when it is labelled as an assumption, so its influence on the conclusion can be seen and tested.

Desk research and verified client results are different things

The analysis published here is original desk research: it draws on official sources and works through reasoning, but it reports no client engagement and no measured client outcome. Where we describe how an engagement works, we describe the method. Where we show numbers, they are labelled assumptions with their arithmetic shown, not results. Verified figures only ever come from a specific engagement, and they are not published here.

Three worked examples

Each example shows the reasoning and the arithmetic. Every figure is an assumed input with its basis stated; none is a measured result and none describes a client.

Illustrative analysis — not client work

Example 1 — Investment sensitivity: which assumption decides the case

The question. A project is close to the hurdle. Three inputs are uncertain: saleable output, the price achieved, and the operating cost. Which one actually decides whether the case clears its threshold?

Assumed inputs. Planned saleable output 8,100 tonnes a year; assumed price 400 per tonne; assumed operating cost 2.40 million a year; assumed capital charge to be covered 0.60 million a year.

Baseline arithmetic. Revenue is 8,100 × 400 = 3,240,000. Contribution after the capital charge is 3,240,000 − 2,400,000 − 600,000 = 240,000. The case clears the charge with 240,000 to spare — a margin of about 7% of revenue.

Testing each input.

Illustrative sensitivity of the surplus to a 10% adverse change in each of three assumed inputs
AssumptionChangeSurplus after the changeEffect
Saleable output−10% (to 7,290 t)240,000 − 324,000 = −84,000Shortfall — fails
Price achieved−10% (to 360/t)240,000 − 324,000 = −84,000Shortfall — fails
Operating cost+10% (to 2.64m)240,000 − 240,000 = 0Breaks even

Reading it. A 10% fall in output or in price wipes the surplus out by the same 324,000, because both scale revenue directly. A 10% rise in operating cost removes exactly the 240,000 surplus. Output and price are equally decisive, cost slightly less so — which tells the project where to spend its effort: on confirming volume and price, not on shaving cost.

Caveats. The inputs are assumed, not measured. The comparison treats each change in isolation, whereas in practice a weak market may move volume and price together. No discounting is applied, and the capital charge is treated as a single annual figure. A real assessment would model these jointly and over time.

Decision implication. Before committing, establish volume and price to a tighter range — those are the two inputs whose uncertainty the case cannot absorb.

Example 2 — Hotel resource intensity: why the denominator matters

The question. A hotel wants to know whether it is improving its energy performance. It looks at energy per occupied room. In a quieter month the figure worsens. Has performance slipped?

Assumed inputs. A fixed baseload of 60,000 kWh a month that does not vary with occupancy, plus 12 kWh per occupied room-night.

Two months.

Illustrative energy per occupied room at two occupancy levels, with a fixed baseload
MonthOccupied nightsTotal energyPer occupied room
Busier3,00096,000 kWh32.0 kWh
Quieter2,00084,000 kWh42.0 kWh

The arithmetic. Busier month: 60,000 + (3,000 × 12) = 96,000 kWh, and 96,000 ÷ 3,000 = 32.0 kWh per occupied room. Quieter month: 60,000 + (2,000 × 12) = 84,000 kWh, and 84,000 ÷ 2,000 = 42.0 kWh per occupied room.

Reading it. Absolute energy fell by 12,000 kWh — a reduction in consumption; spend also depends on tariffs and fixed charges. The per-room figure rose by 10 kWh, about 31%. Performance did not change at all: the same baseload and the same per-room consumption applied in both months. Only the denominator moved.

Caveats. The baseload and per-room figures are assumed. Real properties have several services — restaurants, conference space, laundry, pools — sharing meters, so a per-room figure bundles in consumption that has nothing to do with rooms. The split can be estimated from a property's own records but not from this illustration.

Decision implication. Read absolute use alongside the ratio, and give each service a denominator that matches its own meter boundary. A ratio alone will penalise a quiet month for reasons nobody can act on.

Example 3 — Supplier evidence register: what a claim actually rests on

The question. A business states that a material it sells contains 30% recycled content. A customer asks for the basis. What would actually demonstrate the claim, and what is missing?

Method. The claim is broken into the evidence requirements it depends on, and each is assessed against what exists. This is a structure, not a score: each line is answered so the gaps are visible.

Illustrative supplier evidence register for a recycled-content claim
Evidence requirementWhat would demonstrate itStatus in this example
Recycled input by massSupplier declaration of recycled fraction, per batchHeld for some batches, not all
Chain of custodyRecords linking the input to the originating sourcePartial — origin known, not per batch
CompositionBill-of-materials at component levelHeld, but last confirmed over a year ago
CalculationHow the 30% is derived, and on what basisNot documented
Change controlNotice of any change to material or supplierNot agreed with the supplier

Reading it. The claim is plausible but not yet demonstrable. Two gaps matter most: the calculation is not documented, so the 30% cannot be shown to be derived correctly, and there is no change control, so a supplier substitution could silently alter the material. Traceability exists in part, but traceability is not quality — it says where material came from, not what it can be shown to be.

Caveats. The statuses are illustrative and describe no real supplier. The requirements would differ with the market, the product and the specific claim being made — a recycled-content claim for one application is not governed by the same evidence as another. Nothing here determines how any material should be classified or what any rule requires.

Decision implication. Two first actions are to document how the percentage is calculated and to agree with the supplier what changes must be notified. These alone do not substantiate the claim: the missing batch evidence, the chain of custody and the current composition record must also be resolved before the 30% can be demonstrated to a customer.

The examples above are illustrative analyses prepared to show our reasoning. They are not client engagements, they describe no completed project, and the figures in them are assumptions rather than results. Real work begins with your own data.

What our work is not

  • Not engineering design, technical sign-off or specification of plant and equipment.
  • Not legal opinions, and not a determination of how any material or product should be classified.
  • Not certification, accreditation or formal assurance of any kind.
  • Not food-safety or chemical-safety determinations.
  • Not a guarantee of funding, finance or any investment outcome.
  • Not EPC contracting or delivery of physical works.

Our work is analytical: studies, requirement and evidence registers, decision models and implementation roadmaps. Where a project needs engineering, legal, testing, assurance or delivery specialists, those inputs are scoped separately and sit outside our engagement.