Skip to content

Customer renewal discount

Four roles, each right, on one commercial decision

A renewal worth $400K ARR plus a $150K expansion closes in three weeks, and the customer's executive sponsor says they are evaluating other options. The account lead wants to offer 25% off.

Four agents hold the decision, and each defends a position that is correct from where it sits:

Agent Position What it is optimising
Finance 12% flat, and not above 18% A low headline discount
Customer Success 25%, with engagement attached Value delivered to the customer
Deal Desk 15%, but only on 24 months with committed volume Term and volume protection
Revenue Ops An honest close inside 21 days Speed, and the expansion

The criteria clash. None of the four is wrong, and no one of them can concede without giving up the thing it exists to protect.

Customer renewal discount

Watch the recording  ·  720p · 2 min 27 · 57 MB

The four runs

The recording puts the same decision through four configurations. Taken together they are the clearest demonstration of why the two protocols are separate, and why neither substitutes for the other.

Run What happens Outcome
0:10 No protocol The moderator only relays. Finance opens at 12% flat and will not go above 18%; Deal Desk wants 15% but only on 24 months with volume. Each agent defends its own number Stalled after 46 exchanges, with no package
0:33 ACCORD alone The agents align before negotiating. Convene: each states who it is and what it can judge. Frame: the goal becomes short, checkable statements — objectives and constraints Aligned in 8 exchanges — and still stalled. Aligning intent is not deciding
1:03 CONCORD alone Calibrate: each agent sets its own scale from unacceptable to ideal, so scores are comparable. Score: each proposes a package and everyone scores everyone, without sharing private evidence. Repair: the least satisfied agent proposes a fix and everyone re-scores Certified at 10% — a real decision, but only a number
1:36 ACCORD + CONCORD Ground: each agent defines its terms — what headline % and term mean to it — so the words mean the same thing. Then Score: Customer Success rates a bare 12% at 28, below its floor, so the objection is explicit. Repair: Deal Desk repairs the package with a 24-month term, committed volume and a named remedy for the credits Certified: 12% on 24 months, with committed volume and expansion, a health checkpoint, and $60K credits with a named remedy — closing inside 21 days

What the four runs establish

Without a protocol they argue and stall. Forty-six exchanges produce no package, because relaying positions does not reconcile them.

Aligned intent alone is still a stall. ACCORD gets the agents to agree on the goal and on what the terms mean, in 8 exchanges rather than 46 — but agreement on the question is not an answer to it.

A negotiation alone produces a number. CONCORD reaches a certified outcome without ACCORD, and every agent genuinely clears its floor. But 10% is a headline figure, not a set of terms anyone can sign.

Aligned intent first, then a certified decision, produces the deal. The final run commits a complete package — discount, term, volume, expansion, health checkpoint and credits with a named remedy — and certifies that all four roles accept it.

Why the two protocols are separate

Runs 2 and 3 are the argument. ACCORD without CONCORD aligns and does not decide; CONCORD without ACCORD decides something thinner than the decision the business needed. The composition is what produces a signable outcome.

Handoff between the protocols

What to watch for

  • Satisfaction stated as a number and exchanged, rather than argued.
  • Customer Success scoring a bare 12% below its floor, which is what makes the objection actionable instead of rhetorical.
  • The repair coming from the agent that was least satisfied, and adding terms rather than moving the headline.
  • A floor that does not move when it binds.

Every step is visible in the shared space the agents work in, so the record of how the package was reached is available afterwards rather than reconstructed.

Results across the four decisions