Two quotes for the same part, one 18% cheaper, and the cheaper one wins. Three months later the freight invoices, the two expedites, the inspection hours and the requote for the second batch have put the landed cost above the losing bid. Nothing was hidden in the cheaper quote; those costs were never on the comparison.
Total cost of ownership is the discipline of putting them back before the award. The published literature on it is written for category managers with lifecycle models and a finance team. This guide is TCO at the scale of one RFQ round: what a unit price leaves out, a six-line worksheet that uses the numbers already in the quotes, where AI agents change the arithmetic, when TCO becomes an excuse, and when the lowest price is the right decision. The term itself is defined in our answer on what total cost of ownership means in procurement; this post assumes the definition and argues from it. It belongs to the award step of our practitioner's guide to strategic sourcing and to the comparison stage of supplier quote management.
Why is the lowest quote often not the cheapest?
Because a unit price describes one moment of a transaction, and the cost accrues across the whole of it. Lisa Ellram's 1995 analysis, still the reference framework, groups total cost into pre-transaction costs (finding, qualifying and setting up a supplier), transaction costs (the price itself, delivery, inspection, payment) and post-transaction costs (failures, rework, replacement, lost goodwill) (Ellram, International Journal of Physical Distribution & Logistics Management 25(8), 1995). The unit price sits in the middle category and covers part of it.
A buyer who awards on price alone has not avoided the other two categories. They have deferred them to invoices that arrive later under other names: freight, duty, expediting, inspection time, a second RFQ round because the first batch didn't pass. Those invoices are rarely traced back to the award decision, which is why the lowest-price habit survives its own results.
The habit also has a structural cause. Price is the one number every quote contains in a comparable form. Everything else arrives in different units, different terms and different levels of completeness, and making it comparable takes work. When that work is expensive, price wins by default.
What does a unit price leave out?
Six things, and most of them can be read from the quote itself or from the terms around it.
| Cost | Where it appears instead | Readable from the quote? | What tends to trigger it |
|---|---|---|---|
| Freight, duty, insurance | Carrier and broker invoices | Yes, from the incoterm | EXW or FCA quotes compared with DDP ones; tariff changes (over 90% of US- and China-based sourcing networks were hit hard by tariffs in 2025, QIMA) |
| Schedule cost | Expediting fees, buffer stock, line downtime | Partly, from lead time and incoterm | Long or unreliable transit; roughly one in three ocean arrivals ran off schedule in November 2025, late by nearly five days on average (Sea-Intelligence) |
| Quality cost | Inspection hours, rejects, rework, returns | No; from history or a sample | New supplier, new process, loose specification |
| Transaction cost | Buyer hours: clarifications, requotes, chasing | No; from your own process | Incomplete quotes; supplier-side quoting is still about 90% manual by one shop survey (CNCCookbook, n=100, vendor survey) |
| Tooling and switching | One-time charges, qualification, first-article | Yes, if the quote separates them | Tooling amortized in one quote and invoiced separately in another |
| Validity and price basis | Requote at a different metal or currency basis | Yes, from validity terms | Volatile inputs; aluminium casting quotes now often hold for weeks rather than a quarter (AlCircle, 2026) |
Three of the six are in the quote if you look for them: incoterm, tooling treatment, validity. Two need your own history or a sample. One, transaction cost, is your time, and it is the one buyers most consistently leave at zero.
How do you compare quotes on total cost without a finance model?
With six lines per quote, each either a number, a source, or a written assumption. The worksheet below is sized to a single RFQ round and uses your actual order quantity, not the supplier's price-break quantity.
| Line | What goes in it | Where it comes from |
|---|---|---|
| 1. Unit price at your real quantity | The price at the quantity you will actually order, not the break the supplier highlighted | The quote |
| 2. Landed adders | Freight, duty, insurance and brokerage to your dock, per unit | Incoterm on the quote plus a freight estimate |
| 3. One-time charges, amortized | Tooling, setup, first-article, qualification, divided by expected annual usage | The quote, if separated; ask if not |
| 4. Schedule adjustment | Buffer stock carrying cost or expected expediting, implied by lead time and reliability | Lead time on the quote; reliability from history |
| 5. Quality adjustment | Inspection effort plus expected reject and rework cost | History, a sample order, or a stated assumption |
| 6. Transaction adjustment | Buyer hours for clarifications, requotes and chasing, at a loaded rate | Your own record of how this supplier quotes |
The total is a cost per unit delivered and accepted, which is the number the award should be made on. Two quotes 40% apart on line 1 are often within 10% on line 6, because the difference was scope, and the worksheet is what makes scope visible; the mechanics of that normalization are covered for one category in what to include in an aluminum casting RFQ.
The rule that keeps the worksheet honest: a line may be blank, but it may not be vague. "Quality: probably fine" is not an entry. "Quality: no history; assume 2% rejects pending first-article" is. Practical shortcuts for the comparison itself are in quote comparison tips and supplier quote comparison with AI.
Which of those numbers can you actually get?
Fewer than a finance model assumes, and more than most buyers use. The split matters, because it decides which lines are measured and which are assumed.
From the quote itself: the incoterm, which sets line 2; the minimum order and price breaks, which set line 1 at your real quantity; the tooling treatment, which sets line 3; the validity window and price basis, which tell you whether the number will still hold when the order is placed; the lead time, which is half of line 4.
From your own record, if you have one: the supplier's on-time rate and response behaviour, which complete line 4; first-pass completeness of its quotes and how many clarification rounds it took, which set line 6; the defect rate on past deliveries, which sets line 5. This is where most teams find they have nothing, because those facts lived in individual inboxes and were never recorded. The case for putting requests and replies through one channel is made in procurement visibility, and the supplier-side metrics in how to measure supplier reliability.
From nowhere, for a new supplier: expected quality and reliability. The honest entry is an assumption written down, tested by a sample or a small first order, and replaced by a measurement after the first deliveries.
Where do AI agents change the total cost?
Agents don't lower the unit price. They lower the other five lines, and they lower the cost of computing them, which is the change that matters most. When each RFQ round cost hours of a buyer's time, a six-line comparison was affordable on a large award and not on a routine one. When the round costs minutes, it is affordable on both.
The mechanism maps onto Ellram's three categories.
| Cost line | What an agent changes | Control that becomes possible | Practical before? |
|---|---|---|---|
| Pre-transaction: finding and qualifying | Researches candidates against a brief, finds a reachable contact, removes duplicates, at flat effort per bidder | More bidders per request, so competition reaches categories that were single-sourced by default; e-sourcing can cut MRO costs up to 20%, yet only about a third of firms use it (McKinsey, 2025) | No: each extra bidder cost buyer hours |
| Transaction: the request | Drafts every request from the same template | Quotes comparable by construction: same incoterm asked, tooling separated, validity requested | Only through discipline |
| Transaction: follow-up and triage | Chases non-responders, sorts quotes from questions and declines, timestamps everything | Response rate and time to first quote become measured per supplier, feeding line 6 and line 4 | No |
| Transaction: validation | Checks each quote for price, lead time and MOQ before comparison | First-pass completeness as a metric; fewer requote cycles; line 6 falls | No |
| Transaction: normalization | Converts units, currencies, incoterms and breaks into one landed view | Lines 1 to 3 filled from the quotes rather than retyped | Manually, at a cost that limited it to large awards |
| Post-transaction: the record | Every quote, promise and closed request stored with a reason and a date | Lines 4 and 5 move from assumption to history; price drift per supplier is visible across rounds; expiring quotes are flagged | Not without a data team |
| Re-competition | A competitive round costs minutes | Incumbents are re-tested on a schedule instead of when someone gets around to it | No: a round cost hours, so it happened rarely |
The last two rows are where the less obvious value sits. A TCO worksheet built once is an opinion; the same worksheet built on every round, with the record filling lines 4 to 6 from what actually happened, becomes a measurement that improves with use. And a category that was single-sourced because re-competing it cost more than it saved can be re-competed quarterly at negligible cost, which is the structural answer to the incumbent problem discussed below. The mechanism is summarised in how AI agents reduce total cost of ownership.
Agents such as ours normalize and validate quotes, chase replies, and keep the record; the judgement lines, what a reject actually costs you, what a week's delay is worth on this part, stay with the buyer, and the award itself stays human for the reasons set out in what to let an AI agent do in procurement. And the gains are in effort and data quality rather than in a published percentage. The Hackett Group reports that 76% of organizations see AI-driven improvements of 25% or more in key metrics as adoption scales (Hackett, 2026), but the figure that matters to a buyer is their own line 6 before and after. The wider workflow is described in RFQ automation, the completeness check in RFQ validation, and the category where this bites hardest in AI for tail spend.
Is TCO just an excuse to keep the incumbent?
Sometimes, and the tell is easy to spot: a total-cost argument with no figures attached. "The incumbent is cheaper once you count everything" is a conclusion; the six lines are the evidence, and if they are blank the argument is preference dressed as analysis.
If the incumbent wins on total cost, the worksheet's lines show where, and the buyer can defend the award to anyone who asks why the cheaper quote lost. If the incumbent wins only on lines 4 and 5, and those lines are assumptions rather than history, the honest move is a trial order with the challenger, sized so the assumption can be tested rather than defended.
Re-competition changes this dynamic more than any argument does. When a round is cheap to run, the incumbent's total-cost advantage is re-measured every quarter, and an advantage that is real survives while one that was habit does not.
When is lowest price the right decision?
When the other five lines are equal or near zero, which is more often than TCO essays admit.
| Condition | Why price is enough |
|---|---|
| Fully specified commodity or catalog item | Quality is defined by the specification and verifiable at receipt |
| Deep, qualified supplier pool | Switching cost is low; a poor performer is replaceable next round |
| No tooling, qualification or first-article | Line 3 is zero for every bidder |
| Delivered terms (DDP) from every bidder | Line 2 is inside the price; freight risk sits with the supplier |
| Short horizon, one-off purchase | Post-transaction cost has little time to accrue |
| Supplier already measured | Lines 4 to 6 are known and similar across bidders |
In those conditions, insisting on a total-cost exercise costs more than it saves, and a competitive round or a reverse auction is the right instrument; the choice of event is covered in the strategic sourcing guide. The mistake is applying that logic to a made-to-print part with tooling, a new supplier and an EXW quote, where every line is live.
How does this change what you put in the RFQ?
Ask for the inputs the worksheet needs, so the quotes arrive comparable instead of being made comparable afterwards. Most of the transaction cost in line 6 comes from asking for these things one supplier at a time after the quotes are in.
State the incoterm you want quoted, or ask for two. Ask for tooling and one-time charges as separate lines, never amortized into the piece price unless every bidder does the same. Ask for the quote's validity period and the price basis for any volatile input. Ask for lead time in stages, tooling and production separately. Ask for price breaks at your real expected annual usage, not at the supplier's preferred quantities. Templates that carry these fields every time are covered in RFQ templates, and the full package for one made-to-print category in what to include in an aluminum casting RFQ.
The supplier's side of this is worth remembering. A request that specifies incoterm, quantities and tooling treatment is easier to quote accurately, and suppliers triage requests on clarity before deciding whether to quote at all (Production Machining). A worksheet-ready RFQ tends to get more complete quotes back, which lowers line 6 before the comparison starts.
What does total cost look like across a category, not one order?
The same six lines, summed over a year and across suppliers, with award quality replacing unit-price savings as the success measure. In 2026, supply continuity heads the procurement agenda alongside cost reduction for the first time, according to the Hackett Group's Key Issues Study (Hackett, 2026), and continuity is a total-cost concept: the cost of a supplier that cannot deliver is not on any quote.
At category level the worksheet turns into a scorecard. Line 4 becomes on-time performance, line 5 becomes defect rate, line 6 becomes response rate and first-pass completeness, and a qualified alternate for every strategic part becomes a line item rather than contingency paperwork. The cycle that produces those numbers is the seven-step process in the strategic sourcing guide, and the reason most teams can't fill them in is the visibility problem described in procurement visibility. For suppliers sourced abroad, where lines 2 and 4 dominate, the operating detail is in managing overseas suppliers.
What's the one-hour version?
Six lines per quote is the whole method. Fill lines 1 to 3 from the quotes themselves, which takes minutes once the incoterm and tooling treatment are located. Fill lines 4 to 6 from history where you have it and from a written assumption where you don't. Award on the total per unit delivered and accepted, and record the decision with its reasons, so that when the same category comes round, the assumptions of this round have become the measurements of the next.
The suppliers who lost will rarely learn why, which is a problem on their side of the table too; see why win/loss data is blank. On the buyer's side, the record is what turns one afternoon's worksheet into a category's total cost over time.
FAQ
Why is the lowest quote often not the cheapest?
Because a unit price covers one part of the transaction. Ellram's framework groups total cost into pre-transaction costs (finding and qualifying), transaction costs (price, delivery, inspection) and post-transaction costs (failures, rework, goodwill). Freight, duty, schedule, quality, buyer time and tooling accrue outside the price and arrive later on other invoices.
What costs does a unit price leave out?
Freight, duty and insurance to your dock; schedule cost from lead time and unreliability; quality cost from inspection and rejects; transaction cost in buyer hours for clarifications and requotes; tooling and switching costs; and exposure to a short validity window or a volatile price basis. Three of these can be read from the quote's terms; the rest come from history or a stated assumption.
How do you compare quotes on total cost without a finance model?
Use six lines per quote: unit price at your real quantity, landed adders, one-time charges amortized over annual usage, a schedule adjustment, a quality adjustment and a transaction adjustment. Each line is a number, a source or a written assumption. Award on the total per unit delivered and accepted.
When is lowest price the right decision?
When the other lines are equal or near zero: a fully specified commodity, a deep qualified pool, no tooling, delivered terms from every bidder, a short horizon, and suppliers whose performance is already measured. In those conditions a competitive round on price is the right instrument.
How do AI agents reduce total cost of ownership?
By lowering the cost of running the comparison and by generating the record it needs. Agents find and qualify more bidders at flat effort, enforce a consistent request, validate and normalize quotes, chase replies, and store every outcome with a date and reason. That makes a total-cost comparison affordable on routine purchases and turns assumed schedule and quality lines into measured ones over time. The award stays with the buyer.
Key takeaways
- Awarding on the lowest unit price defers costs rather than avoiding them. Ellram's 1995 framework places two of three cost categories outside the quote.
- A unit price leaves out freight and duty, schedule, quality, buyer time, tooling and switching, and price validity. Three of the six are readable from the quote's terms.
- Total cost can be compared within one RFQ round using six lines per quote, each a number, a source or a written assumption, totalled per unit delivered and accepted.
- Most teams can't fill the schedule and quality lines because those facts were never recorded. Running requests through one channel is what creates the history.
- AI agents lower total cost by lowering the cost of measuring it: more bidders at flat effort, consistent requests, validated and normalized quotes, and a record that turns assumptions into measurements. Re-competing incumbents becomes cheap enough to do on a schedule.
- A total-cost argument without figures is a preference. The worksheet is what distinguishes a real incumbent advantage from a habit.
- Lowest price is the right answer for specified commodities with deep pools, no tooling, delivered terms and known suppliers. The worksheet earns its time on made-to-print parts, new suppliers and EXW quotes.