8 min read
I spent the early part of my career buying direct materials in the automotive industry, and I lived with a daily frustration: the gap between what our internal models said a part should cost and the price the supplier quoted. Sometimes our model was wrong. Sometimes the supplier considered their cost structure proprietary and wouldn't discuss it at all. Either way, the negotiation started from opinion rather than fact.
Should-cost analysis is the discipline that closes that gap. It's a bottom-up method for estimating what a product or part should cost to produce, built from its raw materials, manufacturing processes, labor, overhead, logistics, and a fair profit margin. Instead of anchoring on the supplier's quote or last year's price, procurement builds an independent view of cost and uses the difference as the basis for a fact-based conversation.
Used well, it's one of the most powerful tools in the procurement toolkit. It's also one of the least used. According to research conducted by CADDi, only 28.1 percent of procurement professionals said they know what should-cost analysis is, and just 4.6 percent actually use it. This article covers what should-cost analysis is, what goes into a model, why adoption is so low, and how to put it to work without falling into the trap that ruins most attempts: treating it as a weapon instead of a starting point.
What is should-cost analysis?
Should-cost analysis is a procurement technique that estimates what a product or part should cost by breaking it down into its cost components: materials, labor, machinery and depreciation, overhead, logistics, and profit. The estimate is compared with the supplier's actual price, and the gap between the two becomes the agenda for a data-driven negotiation.
Yushiro Kato, Co-Founder and CEO of CADDi, has built a business around this discipline, and when he joined the Art of Procurement podcast for Getting to Know Should-Cost Analysis, he explained it in refreshingly plain terms:
"Should-cost analysis is a technique in procurement to estimate the cost of a product by breaking it down into details – labor costs, material costs, machinery costs, depreciation, overhead, profit, and sometimes logistics."
The key word is estimate. A should-cost model is not an audit of the supplier's books, and it doesn't need to be. As Yushiro put it, there is always a gap between the estimate and the price being paid, and "that's the starting point of your discussion with your suppliers."
What goes into a should-cost model
A full model can get granular: material grades and scrap rates, operator time per process step, regional wage rates, machine runtimes, equipment depreciation, facility costs, and industry-typical margins. If you're modeling a machined component, you might build up from the raw material weight and price index, add cycle time on a specific machine class, then layer in labor, overhead allocation, and profit.
That level of detail has its place, but it's not the price of entry.
Four components are enough to start
The most practical advice Yushiro offered is also the most freeing: don't let the modeling become the project.
"Even if you have a rough idea of the four components – material cost, process cost, overhead, and profit – that's probably enough to start your conversation."
A simple four-line model that's directionally right will open a better negotiation than a fifty-line model that never gets finished. You can add precision later, ideally with the supplier's help.
Why so few teams use it
If should-cost analysis is this useful, why does CADDi's research show only 4.6 percent of procurement professionals using it? Three reasons come up consistently.
Time goes to transactions, not analysis
Most procurement teams spend the overwhelming majority of their time on transactional work. Yushiro cites data suggesting only about 10 percent of procurement's time goes to strategic activities like cost planning, value analysis, and should-cost modeling. In high-mix, low-volume environments, where a buyer handles thousands of parts, building a model for each one is simply not realistic, which is why part selection matters so much (more on that below).
The knowledge gap is real
Should-cost analysis grew up in automotive and other mass-production industries, where volumes justify deep cost engineering. Professionals who built their careers outside those industries may never have encountered it. That's a solvable problem, and frankly it's an opportunity: if only a quarter of the people who know the technique use it, the practitioners who do use it stand out.
Fear of being wrong
This is the barrier I find most human, and Yushiro named it directly:
"Most people overthink the accuracy of the breakdown. You don't want to look stupid in front of suppliers. But what matters is the result in the end, which comes after the collaboration with your suppliers."
The fear of presenting an imperfect model keeps teams from presenting any model at all. Which brings up the question every first-timer asks.
How accurate does the model need to be?
Less accurate than you think. Yushiro was emphatic on this point:
"Accuracy is actually not that important. What matters is how you leverage it to negotiate with your suppliers, or to have suppliers disclose their own breakdown of the cost and compare. Don't focus too much on the preciseness or accuracy of the should-cost analysis."
The model's job is to change the shape of the conversation. Presenting a rough estimate with honest curiosity ("this is my estimation, it could be wrong, can you show me where?") invites the supplier to correct you with real data. A supplier who engages with that question has just started sharing their cost structure, which is exactly the outcome the model was built to produce. Chasing decimal-point precision before the first conversation gets the sequence backwards.
How to run a should-cost analysis in five steps
1. Pick the right parts. Apply the Pareto principle: roughly 20 percent of parts account for 80 percent of cost, and that core 20 percent is where should-cost modeling belongs. Don't build models for the tail. If you're already segmenting your categories, this analysis slots naturally into the opportunity-identification stage of a structured approach like the 7-step strategic sourcing process.
2. Build the model. Start with the four components: material, process, overhead, and profit. Use material price indices, published wage data, and your own engineering or operations colleagues' knowledge of processes and cycle times. A category profile is a useful home for this cost intelligence, because it keeps the model connected to the supply market context.
3. Compare and locate the gap. Set the estimate against the current price and identify where the difference sits. A gap concentrated in material cost points to a different conversation than a gap in overhead or margin.
4. Open the conversation, not the attack. Share the model with the supplier as a starting point and ask them to challenge it. The purpose is a shared fact base, and how you handle this step determines whether you get one.
5. Set targets over time. The gap rarely closes in one negotiation round. Yushiro described agreeing multi-year trajectories with suppliers: accept a higher cost in year one, with a jointly developed plan to reach the target by year three. That converts the model from a price demand into a shared improvement roadmap.
Negotiate with data, not power
Years ago, on Episode 262 of the podcast, I shared a negotiation principle that should-cost analysis exists to serve: use data, not power. If you're convinced something should cost less, resist playing the power card. Use information, such as cost breakdowns and should-cost models, as the basis of discussion with the supplier.
Power-based negotiation gets a price. Data-based negotiation gets a price and preserves the relationship, because the supplier can see the reasoning rather than just the pressure. Rod Sherkin, Founder and President of ProPurchaser and a former CPO, made the same case on an AOP Live session about cost transparency: by focusing on the primary material and service cost drivers, and building the supplier's overhead and margin into the model, procurement can keep prices in line with supplier costs without sacrificing collaboration.
The mature version of this approach is open-book costing: the supplier shares their actual cost breakdown, and in return procurement guarantees them an agreed profit margin. For strategic suppliers, helping them manage and improve their own cost structure is far more effective than dictating a price and walking away if they can't meet it.
There's one more reason the data-first approach wins, and it's arithmetic. As Yushiro pointed out, a supplier's margin is typically around 10 percent of total cost. Squeezing it is a small, adversarial prize. Working together on the other 90 percent (materials, process efficiency, and overhead) is where the real money sits, and it leaves the supplier healthier rather than weaker.
The fair-profit question
Here's the dimension of should-cost analysis that the software brochures skip: once you can see a supplier's cost structure, you inherit a share of responsibility for what you do with it.
Kelly Barner and I dug into this on Who is responsible for making sure suppliers break even?, prompted by reporting that aggressive procurement behavior was contributing to supplier insolvencies. Two ideas from that conversation belong in any should-cost practice. First, transparency cuts both ways. As Kelly observed, the more steps a supplier takes to prevent procurement from understanding their cost model, the more they release the buyer from responsibility for protecting them. A supplier who wants the benefits of a collaborative cost conversation has to participate in it. Second, "fair profit" is genuinely contested ground: what a buyer considers fair and what a supplier considers fair may never fully align. This is why building an explicit, industry-benchmarked margin into the model, and being willing to discuss it openly, beats leaving it as an unspoken assumption.
A should-cost model that includes a fair margin isn't a concession. It's what makes the rest of the model credible.
Should-cost software: where automation fits
Building models manually limits how many parts you can cover, which is why a category of should-cost software has emerged, spanning spend and cost analytics, CAD-based cost estimation, and platforms that automate should-cost modeling at scale. CADDi, for example, has built its approach around automating should-cost models for low-volume direct material components, precisely the environment where manual modeling is least practical.
The technology solves the coverage problem, but the practice above still decides the outcome. A generated model presented as an ultimatum will fail exactly the way a spreadsheet presented as an ultimatum fails. If you're evaluating tools, the Art of Procurement Provider Directory is a good place to start your research, and the webinar Does-Should-Could: Cost Optimization for World Class Manufacturing is a useful primer on executing at scale.
The bottom line
Should-cost analysis is one of the most powerful and least used tools in procurement, and the barrier is smaller than most teams think. A rough model built from material, process, overhead, and a fair profit margin is enough to change the conversation, because its job is not to audit the supplier but to open a fact-based negotiation.
I suggest you build models for the core 20 percent of parts that drive your spend, present them with curiosity rather than certainty, and treat the gap between estimate and price as a shared agenda. Do that consistently, and you'll negotiate with data instead of power, and your suppliers will be stronger for it.
