Should-Cost Modelling: What Practitioners Told Us — and What the Wider Research Confirms

Should-cost modelling sits in an odd place in procurement. Everyone agrees it matters — it underpins supplier negotiation, cost challenge and spend transparency. Yet it is rarely taught with any rigour, and practitioners largely learn it on the job, if they learn it at all. Inside the Collaborative Procurement Operating Model it is the centrepiece of Domain 03 — True Cost — the discipline that decides whether a procurement team is a price taker or a market maker.

To understand where the profession stands, we ran six LinkedIn polls covering training, usage, accuracy, purpose, methodology and data sources, gathering 171 responses from procurement professionals. The results are revealing on their own. Read alongside the current body of consultancy and academic research on cost modelling — from CIPS, Deloitte, McKinsey, BCG and others — they tell a consistent story: should-cost modelling is a proven, high-value capability that the profession has not yet learned to scale.

This analysis follows the same structure as our earlier pieces on Spend Intelligence (Domain 01) and Stakeholder Engagement (Domain 02). Each of those found a poll finding that turned out to be an industry-wide condition rather than a local quirk — savings pipelines leaking before they reach the P&L, and stakeholder alignment being the binding constraint teams refuse to train for. The should-cost picture rhymes with both: a capability the profession already knows is valuable, structurally under-invested in, and about to be reshaped by AI whether teams are ready or not.

1. The training gap is the profession's biggest problem

78% of procurement professionals either learned should-cost modelling through experience alone (46%) or have never been taught it at all (32%). Only 14% received it as a core part of their formal procurement training.

This finding lines up closely with what CIPS and other bodies see across the profession more broadly. CIPS' own cost-analysis training and skills-benchmarking services exist precisely because organisations routinely find, on assessment, that cost modelling capability is unevenly distributed and rarely benchmarked in a structured way.

Deloitte's 2025 Global Chief Procurement Officer Survey, based on responses from more than 260 CPOs, found the talent gap ranks among the top four barriers preventing procurement functions from delivering value — alongside siloed ways of working (57%), competing priorities (46%), and weak organisational or technology capability (40%). Digital skills and consulting skills were flagged as the two most prominent competency gaps CPOs are trying to close.

The implication is consistent across both data sets: capability is unevenly distributed, junior practitioners have no structured pathway, and the quality of models — and confidence to use them — varies enormously by individual rather than by organisational design. Professional bodies, employers and training providers have a clear mandate to treat should-cost modelling as a formally assessed, core competency rather than an accidental skill picked up mid-career.

2. Awareness isn't converting to practice

48% of respondents say they use should-cost models regularly. But 28% describe themselves as familiar with should-cost but not actively using it — a significant pool of practitioners who understand the value but aren't acting on it.

Industry research on should-cost adoption points to structural reasons for this gap, not just individual hesitation. Building an effective model requires category expertise and real time to construct and back-test — time that strategic sourcing teams under productivity pressure often can't spare against competing priorities like market research and supplier relationship management. As a result, most procurement organisations can only afford to build should-cost models for their highest-spend categories, typically the top 10 to 20 out of hundreds, meaning the practice is applied by a few people, on a small fraction of spend, even where the wider team is broadly supportive of the method.

High upfront investment compounds the problem: dedicated cost-modelling capability, clean data infrastructure and category-specific engineering knowledge all carry a cost that is easy to defer, especially when the return on that investment is difficult to quantify in advance.

Combined with the training findings above, the picture is consistent: the barrier isn't awareness or motivation, it's capability, time and confidence. This is the same shape of problem the Stakeholder Engagement piece surfaced in Domain 02 — teams recognise the value but do not resource the capability. Organisations that invest in structured enablement — even one or two practical workshops, or a defined onboarding path for cost modelling — are likely to see meaningful returns from practitioners who are already convinced but feel under-equipped.

3. The accuracy debate has been resolved

None of our respondents said they worry about the accuracy of their cost models. 55% take the view that estimates are most valuable as conversation-starters; 39% believe models need to be precise. No one is paralysed by the fear of being wrong.

This pragmatic stance matches the direction current commentary on should-cost methodology is taking. Industry guidance consistently frames the tool's purpose as giving buyers a fact-based starting position, not a definitive answer — the value lies in shifting the burden of justification onto the supplier, not in achieving perfect precision. Recent strategic commentary describes this explicitly as "precision meets pragmatism": directionally sound models that open a negotiation are worth more than perfect models that take too long to build to be useful.

This also matters for how the profession talks about should-cost internally. Framing accuracy as the goal, rather than credibility, is one of the more common reasons practitioners hesitate to build or present a model at all. The message for anyone hesitating: good enough to challenge is good enough to use.

4. The purpose framing is predominantly defensive

61% of respondents see should-cost modelling primarily as a way to validate whether a supplier quote is fair. 26% use it to start a cost conversation. Only 13% use it proactively to win price reductions.

There is nothing wrong with validation as a purpose — knowing whether you're being overcharged is genuinely valuable. But the most sophisticated procurement teams use cost models earlier in the process: to shape a negotiating position before a quote arrives, to inform make-versus-buy decisions and to drive supplier conversations about cost structure and design-to-cost opportunities.

This is where the return-on-investment case for proactive use becomes concrete. McKinsey's Cleansheet should-costing practice has been credited with helping a US discount retailer identify roughly $500 million in savings across apparel and footwear spend through parametric should-cost modelling used proactively in specification and design decisions, not just quote validation. Separately, a specialty chemical company's should-cost centre of excellence, built on materials, labour and market data, is reported to have delivered 13% savings on raw materials by shaping sourcing strategy ahead of supplier engagement.

The gap between the profession's dominant "check the quote" framing and the scale of value available from proactive use suggests should-cost modelling remains under-leveraged as a commercial strategy tool, even where the underlying capability exists. It is the same pattern our Domain 01 piece found in savings identification: the levers exist; the operating discipline to use them proactively does not.

5. Methodology and data are maturing — and AI is accelerating it

Among active practitioners, both methods and data sources are increasingly sophisticated. 44% adapt their methodology to the situation; factor/statistical approaches and bottom-up builds tied at 22% each; parametric regression was used by 11%.

Together, structured sources account for 88% of data inputs among active practitioners — the days of should-cost as a gut-feel exercise are largely over for this group. That is the same story Domain 01 told about spend analytics: the leading edge of the profession has already moved to structured, data-driven practice; the middle of the distribution has not yet followed.

This trend toward structured, data-driven methodology is the same one consultancies are now actively accelerating with AI. McKinsey's Cleansheet platform uses parametric modelling to calculate should-cost for thousands of items automatically, supporting fact-based negotiation at a scale manual modelling cannot reach; the firm estimates AI copilots and task-level tools can lift procurement productivity by 25 to 40%. BCG's 2025 research on generative AI in procurement similarly finds AI can streamline manual work in key sourcing processes by up to 30% and reduce overall category costs by roughly 15 to 45%, depending on the category and starting point.

Reported case examples reinforce the direction of travel: one pharmaceutical manufacturer's use of AI-driven should-cost models alongside digital negotiation is reported to have delivered a 10% average reduction in spend and a large increase in negotiation savings won. The practical significance for practitioners is less about any single tool and more about what AI resolves structurally — should-cost models built manually go stale quickly (a model reflecting January's commodity prices is already dated by March, and can be unreliable within weeks in volatile categories like energy, chemicals or semiconductors); AI-assisted refresh cycles are what make should-cost viable across more than the top 10 to 20 highest-spend categories.

For most organisations, the realistic near-term opportunity is not full automation but AI-assisted modelling: faster refresh of existing models, wider category coverage, and a lower expertise threshold to build a first, directionally sound model — directly addressing the training and coverage gaps raised earlier in this analysis.

What this means for Domain 03 — True Cost

Both our poll data and the wider research point to the same conclusion: the profession has largely accepted the value of should-cost modelling, but has not yet systematically invested in building the capability at scale. Within the Collaborative Procurement Operating Model, that is the exact remit of Domain 03 — moving procurement from being a price taker to a market maker by grounding every commercial position in an independent view of what a product or service should cost. The practical implications are five-fold:

Close the training gap — treat should-cost as a core, formally taught and assessed competency, not an accidental skill picked up mid-career. The 78% informal-learning figure is not sustainable, and it mirrors the talent and skills gaps CPOs themselves are naming as top barriers.

Convert the aware-but-inactive — a focused enablement push (workshops, structured onboarding, mentoring) will unlock latent value from the 28% who understand the tool but haven't used it.

Reframe the accuracy concern — communicate clearly that "good enough to challenge" is the standard. Perfection-paralysis holds practitioners back unnecessarily, and the profession has, in practice, already resolved this debate.

Use models earlier — shift from validation to proactive cost shaping. The largest documented value — from McKinsey's and BCG's client work — comes from using should-cost before the quote arrives, not after.

Build structured data assets, and treat AI as an accelerant, not a replacement — internal cost libraries and benchmarks compound in value over time, and AI-assisted refresh is what will make should-cost viable beyond the top 10 to 20 highest-spend categories. This is also where Domain 01 (Spend Intelligence) and Domain 03 (True Cost) reinforce each other: clean, structured spend data is what any AI-assisted cost model needs to work at all.

Should-cost modelling is not a complex or esoteric skill. It is a structured, learnable capability that every procurement professional should have in their toolkit. Both the practitioner data and the wider consultancy and academic research agree on this point — the next step for the profession is acting on it. A dedicated True Cost Diagnostic, aligned to Domain 03, will follow shortly to help teams score their own current capability against the same dimensions surfaced here.