A scenario
The margin arrives in the first month. Nothing else does. Contracted customers keep their old price until their term ends, competitors answer somewhere in the middle of the year, and the account that leaves over it is often one nobody was watching. The hard part is not the arithmetic. It is the order the consequences arrive in.
Run this on your own numbersAll scenarios
Ships as a starter scenario: one lever, price_pct at 20, landing in month one.
Mechanics
A scenario is an ordered list of levers with a month attached to each. Inside a month the order is fixed: the lever first, then your own executives, then competitors, then suppliers, then customers, then sales, then the books, then the people, then the watchers.
The lever multiplies two separate indexes by 1.2: the list price and the price quoted to new business. Both move together. Everything else in that month happens after it, so nobody, including your own finance lead, gets to react in the same tick. If a regulator has already opened a review, the rise is held at the cap instead and an event says so.
Each segment closes the gap between what it actually pays and the new list price by one over the average contract months times its own contract multiplier, held between 2 and 100 percent a month. The multipliers are 0.4 price led, 1.0 mainstream, 1.6 anchored. On a fourteen month average that is about eighteen percent of the rise a month for the price led and about four and a half for the anchored. Revenue does not step up in month one. It walks.
Each competitor sets a target: its reaction strength times your move times a factor of 0.7 plus 0.6 of its aggression. It queues that move to land at its lag times 1.3 minus its aggression, rounded to whole months, and then marks the target answered. It matches a share of your move once, not a share of it every month. An aggressive rival in a manufacturing twin answers roughly nine tenths of a 20 percent rise inside two months. A cautious one answers about a third, four months later, and only after a monthly coin flip decides it has noticed at all.
Churn is the churn from your own record multiplied by a deviation, not replaced by one. The deviation is the log of your price against the strength weighted competitor price, damped by switching cost, scaled by 40 percent of the elasticity, with relative quality and the change in satisfaction in the same exponent. Then it is damped again by the lock: a contracted segment can only act on one minus its locked share times its contract multiplier times 0.85.
Every account from your customer list carries a clock drawn per replication between 40 and 130 percent of its contract term. When it reaches zero that account jumps to the full new price in one step, and a logistic weighs price against relative quality, its own satisfaction and switching cost. It leaves or it renews. If the chance of leaving was above 30 percent and it stayed anyway, it renews at 3 percent less, which is what a renegotiation looks like from outside the room.
Above a 10 percent price index each partner has a 22 percent chance every month of starting to lead with somebody else. When it happens, up to 35 percent of that partner share drifts away, capped at 40 percent of revenue in total. Nothing announces it. It shows up as revenue that did not arrive.
If churn runs more than half again above baseline for two months running, the chief executive puts about 1.2 percent of revenue a month into retention and cannot do it again for four months. If runway falls below a threshold set by the finance lead's own risk appetite, headcount and marketing come out. None of that is in the scenario. It is what this company does when its numbers move, and it is usually where the real answer lives.
The ecosystem
Every agent has objectives with weights on them. A price rise is interesting because it pushes on several objectives at once, in different directions, on different clocks.
Objective: pay as little as the job allows, at 1.7 times the base elasticity. Contract multiplier 0.4 and switching multiplier 0.5, so they see the rise fastest and have the least reason to absorb it. By default they are 45 percent of accounts and 22 percent of revenue.
Objective: get the outcome they bought, weight 0.35, against avoiding the cost of changing supplier at 1.8 times the base. Contract multiplier 1.6. They pick up the rise slowly and argue rather than leave. By default 15 percent of accounts and 35 percent of revenue.
Objective: renew at a price they can defend internally, weight 0.35. One decision each, at their own renewal month, at the full new price. This is where concentration turns from a ratio into a month.
Objective: take share when you give them an opening, weight 0.45, held against not starting a price war they lose. Aggression sets both how much of your move they follow and how fast.
Objective: sell whatever is easiest to sell, weight 0.3. Channel loyalty is a margin number. When your price rises and nobody else's does, the easy sale moves.
Objectives: grow the company at 0.4 and keep it solvent at 0.35 for one, protect margin at 0.45 and cash at 0.4 for the other. Both are on cooldowns, so they cannot pull the same lever twice in a quarter.
The ledger
| Assumption | What it does here | Why to check it first |
|---|---|---|
| Price elasticity of demand | Scales the churn deviation and, at 1.25 times, the new business response | Widest band in the ledger at 40 percent either side, and the single biggest mover of this answer |
| Revenue under contract | Damps how much of the deviation a segment can act on | Usually read from a contract file. If it was a default, the whole shape of the first two quarters is a guess |
| Average months left on contract | Sets how fast each segment picks up the new price | A term length is not the same as months remaining. The model wants the second one |
| Cost to a customer of switching | Divides the price pressure before it reaches churn | Easy to overstate. If you have lost anyone in the last year, that is evidence and it belongs here |
| How hard competitors match a move | Sets the share of your rise they follow, before aggression | A default here means the competitor answer in month three is an industry prior, not your market |
| Months before competitors react | Scaled by 1.3 minus aggression to give the landing month | Moves the whole second half of the year backwards or forwards |
| Monthly customer churn | The base the deviation multiplies. One at today's conditions by construction | If this is wrong, every scenario is wrong by the same factor |
| Market share and regulatory exposure | Three thresholds that decide whether a regulator ever wakes | On the invented Harborline twin, 6 percent share never clears the 25 percent trigger, so the regulator is silent by arithmetic |
The brief lists these with defaults at the top, because a default sitting under a decision this size is the most useful thing the software can point at.
Honestly
A 20 percent rise is the question this engine handles best, and it still has holes worth knowing before you put the output in front of anyone.
On the invented Harborline record the arithmetic is clean: at 31 percent gross margin, a 20 percent rise takes gross profit per unit from 31 cents to 51 cents in the dollar, so volume could fall 39 percent before the company is back where it started. That number is a ceiling on the bad case, not a forecast, and it assumes the rise reaches everybody, which it does not.
The shape underneath
Because most of your revenue is already sold at a price that was agreed before you made this decision. The model reprices each segment at one over the average contract months times its contract multiplier, and reprices a named account in one step when its clock runs out.
The first month shows the rise on new business and on whatever was already repricing. The twelfth month shows most of it. If your book is heavily contracted, a twelve month horizon will understate the steady state, and the brief says so.
Run a sweep on the same lever. It runs the full Monte Carlo at every step, so each point on the curve has a band behind it rather than a single path. The best point it returns is the best point under your current assumptions, which is a different claim from the best point.
Yes, as a probability rather than a decision. Each named account has its own renewal month and its own leave probability. It is a separate scenario to force that loss and see the shape of it on its own.
Yes. That is a different lever, and it moves only the new business index. Existing customers keep what they have until they churn, which protects revenue now and stores up a fairness conversation for later. The model books the protection and does not model the conversation.
Next
A standard run is 200 replications of the scenario and 200 of doing nothing on the same seeds, which is 400 simulations and takes about a tenth of a second for a twelve month horizon.