Construction ITT evaluation: how UK bids are scored

Under the Procurement Act 2023, contracting authorities must set out their award criteria and how they will be assessed before tenders are submitted, and award to the Most Advantageous Tender. In practice that means the ITT contains the marking scheme: the list of criteria, the weighting of each, the scale, the descriptors, any thresholds, and the price formula.

The framework is normally spread across several documents — the tender notice, the instructions to tenderers, an evaluation methodology annex and the response templates. Reconstruct it as one table before you write, because inconsistencies between documents are common and are worth clarifying formally during the query period.

Framework call-offs — Pagabo, Procure Partnerships, SCF, YPO, NHS SBS, Crown Commercial Service — layer their own mechanics on top: fixed quality/price splits, prescribed templates, sometimes a direct award option. The call-off documentation, not the framework brochure, states the scheme that will be used.

Most UK construction ITTs use one of a small number of scales:

• 0–5 with named bands — the most common. Each mark has a written descriptor, and the jump between bands is qualitative: typically evidence and specificity, not word count.
• 0–4 or 0–10 — the same logic with a different granularity. Fewer bands means a single missing element can cost a larger share of the criterion.
• 0–100 percentage — used where the authority wants finer discrimination; descriptors are usually banded in ranges.
• Pass/fail — for compliance, returnables and some technical gateways.

The descriptors matter more than the numbers. A typical top band asks for a comprehensive, project-specific response with robust, relevant and verifiable evidence and no weaknesses; the band below it usually differs only in that evidence is partial or the response is generic in places. Read the two descriptors either side of your target and write to close the gap between them explicitly.

The standard sequence is: mark against the scale, convert to a proportion of the maximum, apply the weighting, then aggregate. On a 0–5 scale, a criterion weighted 15% and marked 3 contributes 3 ÷ 5 × 15 = 9 percentage points.

Two implications are worth internalising. First, weighting decides where effort pays: one band gained on a 25% criterion is worth five bands gained on a 5% criterion. Second, the projection to a percentage happens per criterion before aggregation — averaging raw marks across differently weighted criteria gives the wrong answer and misleads your own forecasting.

Check whether the ITT rounds, and at which stage. Rounding at criterion level rather than at the total can change the ranking of close bids.

Price is scored by formula, and the formula is published. The most common variants:

• Lowest price = full marks, others scored as lowest ÷ yours × available marks. Steeply punitive: being 10% above the lowest costs roughly 10% of the price marks.
• Difference from lowest — marks reduced by the percentage difference from the lowest price, which penalises more sharply at the margins.
• Difference from mean or median — used to discourage abnormally low tenders; being furthest from the average can lose marks in either direction.

Read the formula before setting the price, then model it. With a 70/30 quality/price split and the lowest-price formula, the price gap you can absorb while still winning on quality is calculable — and it is usually smaller than bid teams assume.

Abnormally low tenders can be investigated and, if unexplained, rejected. Where a submission is deliberately keen, make sure the quality narrative explains how the price is deliverable.

A bid can lose while scoring well. The mechanisms are:

• A minimum score on an individual criterion, below which the tender is excluded.
• A minimum overall quality score before price is considered.
• Pass/fail compliance — insurances, accreditations, financial standing, declarations.
• Non-compliance with the submission mechanics — late, wrong format, exceeding a stated limit, missing returnable.

Only a threshold stated in the tender documents, or a minimum the authority has told you about, can reject a criterion. Competitor comparisons and internal benchmarks are useful for calibration but they are not pass marks — treating a benchmark as a threshold leads teams to over-invest in criteria that were never at risk.

Evaluation is usually individual marking followed by moderation. Each evaluator scores independently against the descriptors and records a rationale; the panel then meets to reconcile differences and agree a consensus mark with a written justification. Some authorities use a chair or procurement lead to moderate rather than to score.

Practical consequences for the writer:

• The rationale must be writable from your answer. If an evaluator cannot quote a sentence that demonstrates the descriptor, the consensus drifts down.
• Evaluators mark one criterion at a time. Evidence you placed under criterion 2 may not be credited under criterion 5 — repeat it where it is needed.
• Technical evaluators are often site or asset staff, not procurement specialists. Specific, verifiable detail lands; marketing language does not.
• Answers are read against limits. Material past the stated cut is commonly disregarded entirely.

The reasons construction answers lose bands are consistent:

• Not answering the published wording. A good answer to a different question scores against the question asked.
• Assertion without evidence. Method described, delivery unproven — reliably mid-band.
• Generic content. No reference to this site, this programme, these constraints. Signals a library answer.
• Missing asked-for elements. No named person, no reporting mechanism, no mitigation for the stated risk.
• Internal contradiction. A programme in one answer that conflicts with the resourcing in another, or a price that contradicts the quality claim.
• Unverifiable claims. Statistics with no source, references that cannot be checked, awards irrelevant to the criterion.

The most reliable pre-submission control is to score your own draft the way the panel will. For each criterion, in order:

1. Read the published wording and the band descriptors, nothing else.
2. Award a band to your answer and write the rationale you would defend in moderation.
3. Quote the sentence in the answer that justifies the band. If you cannot, the mark is not there.
4. Identify the single change that would move it up one band, and make that change.
5. Check the answer sits inside the limit with the evidence before the cut.
6. Apply the weighting and see whether the criterion is worth further effort at all.

Use a reviewer who has not drafted the answer. Authors read what they intended; evaluators read what is on the page.

After award, ask for feedback. You should be able to establish your score per criterion, the winning or average score, and the evaluators' reasoning. The Procurement Act 2023 transparency regime, including assessment summaries for unsuccessful tenderers, gives you a documented basis for that conversation.

Analyse it criterion by criterion, not in the round. For each gap ask whether the cause was substance, evidence, specificity or structure — the remedies are different, and only one of them is "write better". Feed the answer into the library as a correction, not as a new boilerplate paragraph.

What is ITT evaluation?

ITT evaluation is the process an authority follows to score the tenders it receives against the award criteria and weightings it published before the deadline. Under the Procurement Act 2023 the contract is awarded to the Most Advantageous Tender, combining weighted quality scores with the price score. Evaluators mark each quality answer against a stated scale and band descriptors, moderate the marks as a panel, and record a written rationale.

How are construction tender scores calculated?

Each quality criterion is marked on the published scale, the raw mark is converted to a percentage of the maximum, and that percentage is multiplied by the criterion's weighting. Price is scored by a published formula, most commonly the lowest compliant price receiving full marks and others scored in proportion. The weighted quality and price scores are added to give the total, and the highest total wins unless a threshold has been failed.

What is a band descriptor in tender scoring?

A band descriptor is the written definition of each mark on the scoring scale — for example 0 unacceptable, 1 poor, 2 satisfactory, 3 good, 4 very good, 5 excellent, each with a sentence explaining what an answer must demonstrate to earn it. Evaluators are required to mark to these descriptors, so a top-band answer must visibly contain what the top descriptor asks for.

Can a construction bid be rejected on a single criterion?

Yes. Where the ITT states a minimum score or pass threshold on a criterion, an answer below it can eliminate the tender however strong the total score. Mandatory returnables, insurance levels, accreditations and compliance declarations are also usually pass/fail. Only a threshold stated in the tender documents can reject a bid — a competitor benchmark or an internal target cannot.

Why did my tender score lower than expected?

The most common causes are structural rather than stylistic: the answer did not address the published criterion wording, a claim was made without delivered evidence the evaluator could confirm, the evidence sat beyond the word limit, the answer was generic rather than specific to this site and contract, or an asked-for element such as a named person or a reporting mechanism was simply absent. Evaluators can only mark what is written.

What can I learn from a tender debrief?

A debrief should tell you your score per criterion, the winning or average score, and the evaluators' rationale — the characteristics of your answer that limited its band. Read it criterion by criterion against your own submission and identify whether the gap was substance, evidence, specificity or structure. Under the Procurement Act 2023 transparency regime, assessment summaries give unsuccessful bidders a documented basis for this.