MEDDICC, scored from the engineer's seat
Score a live deal 0–3 on each MEDDICC letter, move the letter the engineer owns by writing the eval, and hand the one you cannot move to sales today.
On this page
Sales qualifies a deal with MEDDICC; the engineer on site sees the evidence for most of its letters before sales does, because the evidence is in the rooms the engineer sits in. Scoring each letter 0–3 turns a mnemonic into a number two people can disagree about precisely, and into a list of what to do this week. The worked deal below is the claims-triage pilot from the scoping doc: it scores 12 of 21, which is neither good nor bad until you see which letters the points are missing from.
Seven letters, one scale
MEDDICC, as Andy Whyte's MEDDICC sets it out, is the frame that began at PTC as MEDDIC and that most enterprise sales teams run, with Competition added: Metrics, Economic buyer, Decision criteria, Decision process, Identify pain, Champion, Competition. Each letter is a question about the deal, and the same four-point scale answers every one of them:
| Score | What it means |
|---|---|
| 0 | Nobody has asked. We have no evidence either way. |
| 1 | Asserted by us. We believe it; the customer has not said it. |
| 2 | Confirmed by the customer in their own words, but short of what the letter's row below asks for. |
| 3 | What the letter's row below asks for: the number, the name or the date, from the customer's person who owns it. |
The scale is about evidence, not optimism. A 2 means someone at the customer said the thing out loud; a 3 means it carries the figure, the name or the date the letter needs, from the person who owns it, and for most letters that means on paper. The gap between them is where the engineer's week goes, because closing it usually means producing the document. The same discipline the system-design course applies to requirements, numbers instead of adjectives, applies to a deal.
| Letter | A 3 looks like |
|---|---|
| Metrics | the business number, its baseline and its target: median triage 11 minutes, target 7.7 |
| Economic buyer | the person who signs the annual contract, met, and heard saying what they need to see |
| Decision criteria | the test the customer will apply, with the threshold: agreement ≥ 95% on their 200 cases |
| Decision process | the steps from pilot result to signature, with owners and dates |
| Identify pain | what the problem costs today, in their numbers, confirmed by the person who feels it |
| Champion | someone with influence who has put their name on the scoping doc and spends their own time on it |
| Competition | every alternative the buyer is weighing, including doing nothing, named by the buyer |
The worked deal: 12 of 21
Week 2 of the claims pilot. Each letter, with the evidence behind its score.
Metrics, 3. The VP of Claims wants 30% off a measured median of 11 minutes per claim, and the number is in the signed scoping doc with its baseline. Nothing to do here except keep measuring.
Economic buyer, 1. The annual contract is above the VP's signing limit; the COO signs it. We have never met the COO. The VP said "I'll take it up with her", which is a promise to carry the case, not evidence that the case will land. That is a 1, and the 1 is sales's to move.
Decision criteria, 2. The Head of Claims Operations said the pilot has to be "at least as accurate as a senior handler". That is a criterion in the customer's words, with no number. The signed scoping doc carries the definition of done's 95% agreement on a 200-case set, but the set does not exist yet and she still describes the test in her own words; a threshold on cases nobody has labelled is one the customer can still reinterpret.
Decision process, 0. Nobody on either side can say what happens between a passing week-8 result and a signed annual contract. Is there a procurement step, a security review, a budget cycle, a board paper? Which month does the budget open? No one has asked.
Identify pain, 2. Seven of the 11 minutes go on finding policy wording, and the handlers say so; attrition on the floor is high enough that the VP raised it unprompted. Confirmed, in their words, uncosted.
Champion, 3. The Head of Claims Operations got the first week of triage logs released herself, put her name on the scoping doc, and runs the Friday meeting. That is what a 3 looks like, and it is the reason the deal is alive.
Competition, 1. One handler mentioned that the incumbent claims platform "has an AI thing coming". Nobody has asked the VP whether that vendor is being evaluated, or whether "do nothing and hire two more handlers" is on the table. We are asserting that we are the only option.
Total: 3 + 1 + 2 + 0 + 2 + 3 + 1 = 12 of 21. The number alone tells you little. The shape tells you a lot: two letters are strong, two are unknown or near it, and one of the unknowns is the kind nobody can fix in week 7.
The letter you can move this week
Decision criteria sits at 2 because the test exists only as a sentence. The engineer turns it into a 3 by building the thing the sentence describes: sample 200 claims stratified by line and value band, have two senior handlers label them, measure their agreement with each other, and have the QA lead sign the labelled set, with the measured reviewer agreement beside the 95% line. That is a week of work the pilot needs anyway, and it changes the deal: a buyer who has agreed in writing what "good" means cannot later decide the pilot failed on a criterion nobody stated.
Identify pain moves the same way, from 2 to 3, with arithmetic rather than a document. 1,200 claims a day at 11 minutes is 13,200 handler-minutes, 220 handler-hours, more than the 210 productive hours 28 handlers have at 450 minutes each, which is why claims queue overnight. Thirty percent of 11 minutes is 3.3 minutes; across 1,200 claims that is 66 handler-hours a day, nearly nine of the 28 handlers' productive time. Assume a loaded cost of $30 an hour and the pain is $1,980 a day, about $495k a year at 250 working days. Put that on one page, have the champion check the inputs, and the pain is costed in the customer's numbers. What the deal depends on shows the same sum becoming the renewal case.
These two letters are the engineer's because the evidence for them is produced by engineering work. No one in sales can label 200 claims.
The letter you cannot move
Decision process is a 0, and the engineer cannot change it by working harder. Finding out whether there is a security review, how long procurement takes, and when the budget opens is a conversation with the VP and the COO's office that sales owns. What the engineer owns is telling sales today, in one sentence: "Decision process is a zero; nobody knows the path from week 8 to signature, and I need you to find it before week 4."
The cost of not saying it is specific. Suppose the pilot passes in week 8, the budget cycle closed in week 6, and the next one opens five months later. The pilot then sits in a queue, the champion changes roles, the handlers' habits revert, and the renewal case goes stale. A 0 in decision process is how a passed pilot fails to become a contract, and it is the most common way.
Economic buyer and competition are also sales's letters, but the engineer still reports what the site tells them: the handler's remark about the incumbent's roadmap goes to sales the day it is heard, not at the quarterly review.
What the score is for
The total is not a forecast. A 12 with a 0 in it is a worse deal than a 10 with no zeros, because any single 0 can stop the contract on its own, and the total hides that. Read the scorecard as a to-do list sorted by who can act: the engineer's letters this week, sales's letters today, and the ones that are done. Re-score every two weeks, in the same meeting as the eval numbers, so that the commercial state of the deal and the technical state of the pilot are never discussed separately. The stakeholder lesson covers who hears which score.
In an interview
The case round gives you a deal description and asks what you would do first. What is being tested is whether you can separate the letters evidence supports from the ones optimism supports, and whether you know which of them an engineer can move. Score it out loud, letter by letter, with the evidence for each score in a sentence; then name the letter you move this week and the one you hand to sales, and say why the split falls that way.
The mistake that loses points is reporting the total as if it were a probability: "it's a 14 out of 21, so it looks decent". A total with a 0 in it is not decent, and an interviewer who hears the total without the zero concludes you would carry a doomed deal to week 8 without telling anyone.
Check yourself
1. The customer's COO says, on a call you were on, "if the pilot shows 25% I'll sign it in Q3". Which letters change, and to what?
Economic buyer goes to 3: the signer is met, and has said what she needs to see. Decision process goes from 0 to 2: a date exists in her words, but the steps between week 8 and Q3 are still unnamed. Metrics does not change; the target in the scoping doc is 30%, and now there is a second, lower number from a different person. Get the two reconciled in writing this week, because a pilot that lands at 27% must not be arguable either way.
2. Decision criteria is at 2 in week 3 and the eval set will take until week 5. Do you build the eval or ship the retrieval improvement the handlers asked for?
The eval. Without the written criterion, the retrieval improvement is unmeasurable, and the week-8 result can be argued down to a 1 by anyone who would rather not sign. Two weeks on the eval is also two weeks the handlers spend labelling, which is where the pain gets costed. Ship the retrieval improvement in week 5 against a criterion that exists.
3. Everything scores 2 or 3 except competition, which is 0. How much does it matter?
Enough to ask one question. The alternative that most often wins is "do nothing and hire": at $495k a year of costed pain, two more handlers at the assumed $30 an hour cost about $120k a year and remove a visible backlog with no security review. Ask the sponsor what else is on the table; if hiring is, the renewal case has to beat it on numbers, not on the demo.