How to Build an Incoterm Decision Tool for Your Sourcing Team?
A decision tool needs five inputs — order size, entry capability, cost-control need, delivery address type and whether the classification is confirmed — a weighted score, a decision table that returns a term, and an audit row recording why. One sheet, four blocks. MOQ 500 pcs per style.
Build it in one sheet with four blocks. Inputs: order size, entry capability, cost-control need, delivery address type, and whether the classification is confirmed. Scoring: a weight per input and a score per option. Decision: a table that returns a term and a reason from the score. Audit: a row per decision recording the inputs, the output, the date and who approved it. Four blocks and the tool survives contact with a real team.
Why Does a Team Need a Decision Tool at All?
Because without one the term is chosen by whoever is in the room, and the choice is re-litigated on every order. Two buyers with the same facts reach different answers, the answers are not written down, and a year later nobody can say why the programme is on the term it is on. A tool fixes the decision to the inputs rather than to the person.
There is a third reason that appears once a team grows past two or three buyers: onboarding. A new buyer faced with eleven rules and no guidance will ask a colleague, and the answer they get is the habit of that colleague rather than the policy of the business. A tool is the fastest way to make a new buyer as good as an experienced one on this specific decision.
The second reason is that the inputs change. A programme that started at 500 pcs with no broker is not the same programme two years later at a full container with an entry team. Without a tool the term never gets revisited, and the programme keeps paying for a service it no longer needs.
What Inputs Does the Decision Actually Need?
Five, and they are the ones that actually move the answer. Order size, expressed in cartons and cubic metres rather than in units, because volume buys the freight. Entry capability at destination, meaning whether an entity can be the importer of record. Cost-control need, meaning whether the buyer needs to see and benchmark the freight and duty lines.
Two inputs are deliberately absent from the list and the absence is worth defending. The unit price is not an input, because the same bag has a different unit cost under each term and comparing prices across terms is comparing different things. And the relationship with a particular forwarder is not an input, because it changes the execution rather than the decision.
Delivery address type, because a fulfilment centre or a residential address changes who has to solve the last mile. And whether the classification is confirmed, because an unconfirmed classification is an unquantified duty and that is an argument for buying a number that includes it.
| Input | How to express it | Why it moves the answer | Where it comes from |
|---|---|---|---|
| Order size | Cartons and cubic metres | Volume buys the freight, not unit count | The packing plan |
| Entry capability | None, broker, or in-house | Decides whether buyer-side terms are available at all | The finance or compliance function |
| Cost-control need | Low, medium, high | Decides whether the freight line must be visible | The commercial function |
| Address type | Dock, residential, appointment | Decides who books and solves the last mile | The destination record |
| Classification | Confirmed or not | An unconfirmed duty is an unquantified cost | The HS code work |
How Do You Weight the Inputs?
Entry capability outweighs everything else, because it is a gate rather than a preference: if no entity can be the importer of record, the buyer-side terms are unavailable whatever the other inputs say. Give it a weight high enough that it decides the outcome on its own in the extreme case.
One caution about weighting: resist the temptation to add inputs. A model with nine inputs and nine weights is harder to argue with and no more accurate, because the five that matter are doing all the work and the other four are noise that occasionally flips the answer. If an input cannot change the outcome on its own, it is commentary.
Cost-control need comes second, because it decides between the terms that remain. Order size and address type come third, and they mostly decide the mode and the delivery structure rather than the term. Whether the classification is confirmed is a modifier: it does not change the term so much as strengthen whichever recommendation the other inputs produced.
- Entry capability
- Weight 5 — a gate, not a preference
- Cost-control need
- Weight 3 — decides between the terms that remain
- Order size in cubic metres
- Weight 2 — decides mode and structure
- Delivery address type
- Weight 2 — decides the last mile
- Classification confirmed
- Weight 1 — a modifier on the recommendation
What Should the Decision Table Contain?
Rows are buyer profiles built from the inputs; columns are the candidate terms; the cell holds the recommendation and one line of reasoning. Keep it to six or seven rows, because a table with twenty rows is a table nobody reads, and the value of the tool is that it is short enough to be used.
A second design point matters as much as the reason: the table should be able to return "re-quote and compare" as an answer. Some input combinations genuinely do not resolve — a programme sitting exactly on the consolidation-to-container boundary with medium cost-control need is one — and a tool that forces a term in those cases is less useful than one that says ask for both.
The critical design point is that every row returns a reason as well as a term. A tool that returns "DDP" is a black box; a tool that returns "DDP because no entity can be importer of record and the classification is unconfirmed" is a decision somebody can defend to a finance director.
| Buyer profile | Entry capability | Cost control | Recommended term | Reason returned |
|---|---|---|---|---|
| New importer, no entity, no broker | None | Low | DDP to door | Nobody can file the entry and the duty is unquantified |
| New importer, broker appointed | Broker | Medium | FOB Xiamen | Entry capability exists and the freight line should be visible |
| Marketplace seller, appointment needed | Agent only | Low | DDP to the fulfilment address | An appointment is required and there is no broker |
| Established importer, new category | In-house | High | FOB Xiamen | The routine exists; only the classification is new |
| Established importer, consolidation | In-house | High | EXW with a nominated consolidator | Pickup and consolidation are already owned |
| Programme with a landed-cost target | In-house or broker | Very high | FOB Xiamen, freight bought separately | The lines have to be separable to be modelled |
| Multi-country programme | Varies | High | A term per destination | One term rarely fits every destination |
How Do You Build It in a Spreadsheet?
Four blocks, top to bottom, on one sheet. Block one is the input block: five labelled cells with drop-down lists rather than free text, because free text produces two spellings of the same answer and a lookup that fails. Block two is the scoring block: a weight per input and a score per option.
Add a fifth block if the programme runs more than one destination: a destination tab holding the entry capability and address type per country. The decision block then looks up the destination first and returns a term per country, which is how a multi-country programme actually works and what a single-country tool gets wrong.
Block three is the decision block: a lookup that returns the row matching the scored inputs, and a cell that concatenates the term and the reason. Block four is the audit block: one row per decision with the inputs, the output, the date and the approver. Four blocks and no macros.
What Outputs Should It Produce?
Four, and they are all short. The recommended term. The named place, because a term without a place is not a term. The reason, in one sentence. And the review trigger — the condition under which the answer should be revisited.
A fifth output is worth adding because it prevents the most common misuse of the tool: the document clause. The output should show the sentence to put in the contract — the term, the named place and the version — so the decision and the paperwork are produced by the same step rather than by two steps that can disagree.
The review trigger is the output teams forget and the one that keeps the tool honest. "Review when the order exceeds 20 cubic metres, when a second destination is added, or when entry is brought in-house" is a sentence that prevents the tool becoming a permanent answer to a temporary situation.
How Do You Stop It Becoming a Rubber Stamp?
Three mechanisms. Make the reason mandatory, so a decision cannot be recorded without one. Make the review trigger a dated item rather than a note, so it comes back. And review the table itself once a year against the decisions actually taken, because a table that is routinely overridden is a table that is wrong.
The second mechanism is the one that works where the first is unavailable: publish the override rate. A team that can see that the tool was right in nine decisions out of ten stops treating it as an obstacle, and a team that can see it was right in six out of ten has a concrete reason to fix the table. Either way the number does the work that an argument would not.
The second of those is the one that works. A tool with a review date produces a calendar entry; a tool without one produces a decision that is still in force five years after the circumstances that produced it have changed.
How Do You Keep It Current?
Tie it to the two things that change: the rules and the programme. The rules change when the International Chamber of Commerce revises them, which is roughly once a decade, and the check is whether the version reference in the tool still says Incoterms 2020. The programme changes much more often.
Put the review on the same cycle as the annual sourcing review, and put the trigger conditions in the tool itself so they are checked rather than remembered. A one-hour review once a year is enough for most programmes.
What Should It Record for Audit?
Per decision: the date, the five inputs, the output term and place, the reason, the approver, and the review date. Eight fields, one row, and the whole history of the programme term decisions fits on one page.
One caution about the audit block itself: record the inputs as they were, not as they turned out. A row that says the classification was confirmed when it was not is worse than no row, because it makes a wrong decision look reasonable in hindsight and removes the one signal that would have prompted a review.
One practical note on the audit block: keep it on the same sheet rather than in a separate log. A separate log is a separate task, and a separate task is the one that stops being done in the second quarter. One sheet, four blocks, and the audit row is written at the same moment as the decision.
That record has a use beyond audit. When a supplier or a forwarder asks why the programme uses the term it does, the answer is a row in a sheet rather than an opinion, and when the term needs to change the argument is already written down.
Six design rules for the tool
- Inputs as drop-down lists, not free text, so the lookup cannot fail on spelling.
- Entry capability weighted as a gate, because it decides availability not preference.
- Every recommendation returns a reason, not only a term.
- The named place is an output, because a term without a place is not a term.
- A review trigger with a date, so the decision comes back.
- One audit row per decision, with the approver.
How Do You Roll It Out to a Team?
Run it in parallel for one quarter rather than switching it on. Let buyers make the decision as they normally would, run the tool on the same facts, and compare. Where the two agree, the tool is confirmation. Where they disagree, the disagreement is the training material, and it usually reveals an input the tool was missing.
One last measure keeps the tool alive: name an owner. A tool with a named owner gets reviewed when the rules change and when the programme changes; a tool without one is a spreadsheet somebody inherited, and it will still be returning the same answer long after the business has moved on. Put the name in the sheet itself.
The rollout has one further step that is worth building in from the start: publish the override reasons. Where three buyers have overridden the tool for the same reason, the table has a gap, and a quarterly look at the override column is the simplest way to find it before it becomes a habit.
The comparison phase has a second benefit that is worth designing for: it tests the inputs. Where the tool and the buyer disagree consistently on one profile, the input set is usually missing something real — a destination quirk, a customer requirement, a cash-flow constraint — and finding it in a quarter of parallel running costs much less than finding it after rollout.
Then make it the default and keep the override. A buyer who disagrees with the tool should be able to record a different decision with a reason, because a tool that cannot be overridden is a tool that will be worked around. The override rate is then the measure of whether the table is right.
Freight bands and transit windows in this cluster are planning ranges drawn from quoted lots, not published tariff rates. They move with bunker cost, vessel rotation, port congestion and the week you book; ask for a live quotation against your sailing before you commit the order to a calendar.
Ask for the decision inputs in writing
Tell us the destination, the order size and whether you have a broker, and we will run the same five inputs and return a recommended term with the reasoning written down.
Frequently Asked Questions
What inputs does an Incoterm decision need?
Five: order size in cartons and cubic metres, entry capability at destination, cost-control need, delivery address type, and whether the classification is confirmed.
Which input matters most?
Entry capability, because it is a gate rather than a preference. If no entity can be the importer of record, the buyer-side terms are unavailable whatever the other inputs say.
Should the tool return only a term?
No. It should return the term, the named place, the reason and a review trigger. A term alone is a black box nobody can defend.
How many rows should the decision table have?
Six or seven. A table with twenty rows is a table nobody reads, and short enough to be used is the point of the tool.
How do I stop it becoming a rubber stamp?
Make the reason mandatory, make the review trigger a dated item, and review the table itself once a year against the decisions actually taken.
How often should the tool be reviewed?
The rules change roughly once a decade, so check the version reference. The programme changes much more often, so put the review on the annual sourcing cycle.
Should buyers be able to override it?
Yes, with a recorded reason. A tool that cannot be overridden will be worked around, and the override rate is the measure of whether the table is right.
What should the audit row contain?
The date, the five inputs, the output term and place, the reason, the approver and the review date. Eight fields, one row.
Can one term cover several destinations?
Rarely. Entry capability and address type differ by country, so a multi-country programme usually needs a term per destination.
Does the tool need macros?
No. Four blocks — inputs, scoring, decision and audit — with lookups and one concatenated reason cell is enough.
How do I roll it out?
Run it in parallel for a quarter, compare the tool answer with the human answer, and treat the disagreements as training material rather than as errors.
What does the tool do about the rules version?
It should state it. Incoterms 2020 is the current set, and the tool should name the version so the recommendation is traceable to a rule rather than to a habit.
People Also Ask
What inputs does an Incoterm decision need?
Order size in cartons and cubic metres, entry capability, cost-control need, address type and whether the classification is confirmed.
Which input matters most?
Entry capability — it is a gate rather than a preference, because buyer-side terms are unavailable without it.
Should the tool return only a term?
No. It should return the term, the named place, the reason and a review trigger.
How do I stop it becoming a rubber stamp?
Require a reason, date the review trigger, and review the table against decisions actually taken.
Should buyers be able to override it?
Yes, with a recorded reason. The override rate is the measure of whether the table is right.
Sources and further reading
Every figure on this page is either our own programme data or attributed below. Open the originals before you write them into a specification.
- Incoterms 2020 — ICC rules for the use of domestic and international trade termsInternational Chamber of Commerce
- ISO 2859-1:1999 — Sampling procedures for inspection by attributesInternational Organization for Standardization
- ISO 9001 — Quality management systems: requirementsInternational Organization for Standardization