The Death of "Trust Me, Guv": Why PA23 Demands Hard Evidence for Your Procurement Decisions

The Procurement Act 2023 has fundamentally changed the game for public sector procurement, and many authorities still haven't quite grasped what this means in practice. At its heart is a seemingly simple shift: from selecting the "Most Economically Advantageous Tender" to choosing the "Most Advantageous Tender." But don't be fooled by the brevity of that change, it's created a requirement that's proving rather uncomfortable for procurement teams used to more, shall we say, flexible approaches.

Here's the thing. Under the old regime, you could get away with a bit of hand-waving. Your evaluation might say something like "Bidder A scored 7/10 for quality based on the panel's assessment of their methodology." Challenge that, and you'd explain it was a matter of professional judgement, perhaps reference your extensive experience, and generally hope the courts would defer to your expertise. Often, they did.

Not anymore.

The Golden Thread That Won't Break

PA23 introduces what's called the "golden thread" running from Section 19 (where you publish your evaluation methodology) right through to Section 50 (where you explain to unsuccessful bidders why they lost). And here's the kicker, that thread needs to be made of steel, not gossamer. Every decision, every calculation, every trade-off must be traceable back to published criteria and demonstrable facts.

The legislation essentially says: "You must explain, transparently and completely, why Tender A was more advantageous than Tender B." Not "we felt it was better" or "our expert panel preferred it." But why, specifically and factually, it delivered greater advantage across the dimensions you said mattered.

This creates a rather significant problem for the traditional weighted scoring approach that's been the backbone of public procurement evaluation for decades. You know the one; you assign weights to criteria (30% for technical approach, 20% for experience, etc.), score each tender out of 10, multiply, add up, and Bob's your uncle. Except Bob's not your uncle under PA23. He's more like a very stern auditor asking awkward questions.

The Complexity Problem Nobody Wants to Talk About

Contemporary procurement doesn't just ask you to balance quality against price anymore. Oh no. You're now expected to systematically assess capability and competence (following BS 8670 if you're sensible), fraud prevention procedures (hello, Economic Crime and Corporate Transparency Act 2023), quality management systems, productivity commitments, whole-life costs, social value outcomes, and environmental sustainability. That's seven major value dimensions, each with multiple sub-criteria.

Try holding all that in your head whilst making consistent, defensible judgements across four, five or even 10 tenders. Research suggests humans can reliably handle about four chunks of information simultaneously. You're being asked to process something in the region of 25+ criteria. It's cognitive overload, frankly, and it shows in the results, studies find that different evaluators assessing identical tenders can disagree on rankings by 30-40%. That's not a trivial margin of error.

So What Counts as "Proof"?

This is where it gets interesting (and, for some authorities, rather uncomfortable). Proof under PA23 means:

Published methodology before tender submission. You need to specify upfront, in your Section 19 notice, exactly how you'll assess each dimension. Not vague statements like "we'll consider relevant experience" but specific criteria with defined evidence requirements.

Objective calculations, not subjective scores. The emerging best practice is monetising everything you can, whole-life costs, social value (using the National TOMs framework), environmental benefits (carbon at £241/tonne), productivity improvements. For things that can't be directly monetised, like capability or fraud prevention, you quantify risk exposure using empirical data from past projects.

Complete audit trails. Every number in your final calculation must trace back through documented steps to source evidence in tender submissions. "We scored them 7/10" won't cut it. "Their proposed approach delivers £2.1M additional social value compared to baseline, calculated as follows..." might.

Empirical grounding for risk assessments. If you're saying Bidder A carries higher capability risk than Bidder B, you need to base that on evidence, historical data showing that organisations with similar evidence profiles experienced cost overruns X% of the time.

The Uncomfortable Truth

Here's what many authorities are discovering: meeting PA23's transparency requirements for genuinely multidimensional assessment probably requires computational support. We're talking about AI systems that can extract evidence systematically, verify claims against databases, calculate risk probabilities from historical patterns, and generate complete documentation automatically.

This isn't about replacing human judgement, quite the opposite. It's about freeing humans from the impossible task of processing complexity that exceeds our cognitive architecture, so we can focus on the genuinely human parts: deciding what matters, applying contextual understanding, and taking accountability for outcomes.

The old approach; a panel sitting round a table, discussing each tender, reaching a consensus view, simply cannot produce the level of factual proof that PA23 demands when you're assessing seven dimensions across multiple tenders. Something has to change. Either we simplify assessment (abandoning legitimate policy objectives), or we embrace systematic, evidence-based approaches supported by technology.

Because "trust me, it's the right decision" isn't compliant with PA23. And sooner or later, the courts are going to make that abundantly clear.