Alevli framework
The bad-case forecast
A forecasting method that chains a marketing budget to revenue in four multiplications, runs the chain as bad, base and good scenarios across a three-month ramp, and sets the owner's expectation at the bad case, because a business that can live with the worst realistic outcome can hold its nerve long enough to reach the good one.
Most marketing budgets are signed with a single number in the owner's head, and it is always the good number. Month one arrives at half of it, which is exactly what month one was always going to produce, and a working campaign gets cancelled for being on schedule. The bad-case forecast exists to move that number before the money moves. It is the Alevli gap pointed inward: the same law, satisfaction equals delivery minus expectation, applied to the owner's own expectation of marketing. Set it at the bad case, and every month afterwards is a pleasant surprise or a well-flagged problem.
It is the third leg of a budget system. The margin-back budget says what the business can afford. The spend floor says what the bidding algorithm needs. The bad-case forecast says what will come back, when, and inside what range.
The bad-case forecast
Five inputs
The chain
Three scenarios
The ramp
The signing test
Use it when
- A prospect asks "what will I get for €2,000 a month?" and deserves a range instead of a hope
- Onboarding any new account, so month one is judged against month one
- A client wants to cancel a campaign that is on schedule but under their imagined number
Do not use it when
- Brand or reach campaigns, whose returns arrive as memory rather than as this quarter's leads
- No close rate or contribution figure exists yet; a forecast built on guesses in every cell is a guess wearing a table
The steps
- Gather five inputs. Monthly budget; a cost-per-click range from Keyword Planner or auction data; a conversion-rate range from the account's history or a conservative benchmark; the business's own lead-to-customer close rate; and contribution per new customer, carried over from the margin-back budget.
- Chain the budget to revenue. Budget ÷ CPC = clicks. × conversion rate = leads. × close rate = customers. × contribution = money back. Four multiplications, no black box, every cell checkable by the owner.
- Run it three times, never once. The bad case takes the high CPC, the low conversion rate, and the low close rate; the good case takes the opposites; the base case sits between. A single forecast is a promise. A range is a forecast.
- Apply the ramp. Plan month one at roughly 50% of steady state, month two at 80%, month three onward at 100%. This is a practitioner heuristic from running small local accounts, not a research finding. It is the one input on this page with no source behind it, and the first thing to replace with the account's own history once it exists. Judging month one against steady state is a common reason working campaigns get cancelled.
- Set the expectation at the bad case, month three, then check two things. If the business can live with that number, sign; if it cannot, the good case was never a plan. Then compute the payback month from cumulative contribution against cumulative spend, and check whether the bad case's monthly leads clear the spend floor. A bad case under the floor is the forecast telling you to tighten scope before launch.
Worked example
A dental clinic at €2,000 a month, contribution €250 per new patient in year one, close rate from the clinic's own records. Illustrative numbers.
| Steady state (month 3+) | Bad | Base | Good |
|---|---|---|---|
| CPC | €4.00 | €3.00 | €2.50 |
| Clicks | 500 | 667 | 800 |
| Conversion rate → leads | 4% → 20 | 6% → 40 | 8% → 64 |
| Close rate → new patients | 55% → 11 | 60% → 24 | 65% → 41.6 |
| Contribution per month | €2,750 | €6,000 | €10,400 |
| Spend floor (30 leads) | Below | Clears | Clears |
Monthly contribution at steady state, against a €2,000 budget
The numbers
| Group | € |
|---|---|
| Budget | €2,000 |
| Bad case11 new patients a month | €2,750 |
| Base case24 new patients a month | €6,000 |
| Good case41.6 new patients a month | €10,400 |
The ramp, bad case only, the row the owner signs against:
| Month | Ramp | New patients | Contribution | Cumulative spend | Cumulative contribution | Cumulative net |
|---|---|---|---|---|---|---|
| 1 | 50% | 5.5 | €1,375 | €2,000 | €1,375 | −€625 |
| 2 | 80% | 8.8 | €2,200 | €4,000 | €3,575 | −€425 |
| 3 | 100% | 11 | €2,750 | €6,000 | €6,325 | +€325 |
Bad case, cumulative net by month
The numbers
| Group | € |
|---|---|
| Month 1 | €-625 |
| Month 2 | €-425 |
| Month 3payback month | €325 |
The conversation this table produces: "In the worst realistic case you get five or six new patients in month one, eleven a month from month three, and the campaign pays for itself in month three. Can you live with that?" If yes, the account is signed on numbers it will almost certainly beat. The bad case also fails the spend floor at 20 leads a month, so the launch plan tightens to one service on exact match, which lifts the conversion rate and lowers the CPC, pulling the bad case toward the floor before a euro is spent.
Why it works
Owners do not churn because results are bad; they churn because results are below the number they imagined, and nobody ever wrote that number down. The number they imagined is reliably the good case. Kahneman and Tversky named the pattern the planning fallacy in 1979 and proposed its correction in the same paper: forecast from what comparable cases actually produced, not from this case's story. Lovallo and Kahneman showed the pattern running through executives' decisions in 2003, and Flyvbjerg turned the correction into reference-class forecasting for project planning. The bad-case forecast is that correction cut down to five inputs. It writes the number down, sets it low on purpose, and attaches a date to it, the only combination that survives a slow month one.
It is also a premortem in numbers. Klein's premortem asks a team to assume the project has already failed and explain why; the bad-case forecast shows the owner the disappointing month before it happens. The psychology of why that works is Oliver's expectation-disconfirmation research: satisfaction tracks the gap between expectation and result, not the result alone. Here it is spent on the person who signs the invoice, not the person who buys the service. An agency that forecasts its own bad case is also making a claim about its honesty everywhere else, which is worth more than the good case ever was.
Download the spreadsheet behind this page: enter your own five inputs, and the three scenarios, ramp, payback month and spend-floor check recalculate for your business.
The bias this model corrects and the direction of the correction come from the research below. The four-multiplication chain is unit-economics arithmetic, and running bad, base and good cases is standard forecasting practice, applied here to a budget. The bad-case signing rule, the spend-floor check and the 50/80/100 ramp are original to this model, and the ramp is the one element that is a heuristic rather than a finding, pending validation against real account history.
Sources
- Kahneman, D., & Tversky, A. (1979). Intuitive prediction: biases and corrective procedures. TIMS Studies in Management Science, 12, 313-327.
- Lovallo, D., & Kahneman, D. (2003). Delusions of success: how optimism undermines executives' decisions. Harvard Business Review, 81(7).
- Flyvbjerg, B. (2006). From Nobel Prize to project management: getting risks right. Project Management Journal, 37(3), 5-15.
- Oliver, R. L. (1980). A cognitive model of the antecedents and consequences of satisfaction decisions. Journal of Marketing Research, 17(4), 460-469.
- Klein, G. (2007). Performing a project premortem. Harvard Business Review, 85(9).
Who this is for
The same method, read three ways.
01
Running it
Five inputs, the chain run three times, the ramp applied. Put the bad case at month three in front of the owner and ask whether the business can live with it; if the bad case sits under the spend floor, tighten scope before launch.
02
Teaching it
The planning fallacy and reference-class forecasting, cut down to a five-input spreadsheet. The teachable part is the signing rule: the expectation is set at the bad case on purpose, and month one is judged against month one.
03
Signing a budget
Before you sign, ask for the worst realistic month three and the payback month in writing. If you could not live with that number, the good case was never a plan.