Use caseLive
Write the goal. Get three plans.
You state the result you want and the limits you will not cross. Jonbar searches the candidate plans (price level, discount depth, how long, which week it starts), drops every plan that breaks a limit and returns the three that fit best, each with its trade-off.
The search runs on Jonbar's market model for the category, calibrated with your own sales history.
Goal
Grow units 15% this month.
Limits
- Gross margin stays at or above 22%
- No discount deeper than 25%
- One campaign, at most 3 weeks long
Example goal, written the way a seller would write it.
The situation
One month, four levers, too many combinations to try by hand.
A goal is easy to write. The plan that reaches it without breaking anything is the hard part.
Example seller
Home textile: duvet covers and towels, sold on a marketplace and on her own store.
Three weeks before the November sales period, cotton has gone up and the list price has to move with it. Last November she cut 30% for ten days, cleared the warehouse and finished the month with less gross profit than the month before.
This year the question is not whether to run a campaign. It is which combination: list price a little higher or a little lower, discount 10% or 25%, one week or three, starting before the sales period or inside it. Every combination moves units and gross margin in opposite directions, and the discount is still being paid for in the weeks after it ends.
Four levers with a handful of settings each already make hundreds of plans. A seller can hand-build three of them, and the spread between a good and a bad one is wide: measured event by event, promotional ROI has run from −100% to more than 700%. Strategy& (PwC), CPG manufacturers survey
The search
Hundreds of plans in, three plans out.
Each dot is one candidate plan, placed by what it does to units and to gross margin over the month and the two weeks after it.
- 01Build the candidates: every price level crossed with every discount depth, length and start week.
- 02Play each one in the market model and read off units, gross margin and gross profit.
- 03Drop the plans that break a limit, rank what is left by gross profit, keep three with different price levels.
- Inside the limits
- Breaks the units or margin limit
- Discount deeper than 25%
Example grid: 7 price levels x 5 discount depths x 3 campaign lengths x 3 start weeks, plus the no-campaign plan at each price level.
Plan 1Leader
List price -4%
10% off, 2 wk · starts week 1
- Units
- +19.6%
- Gross margin
- 34.6%
- Gross profit
- -5.2%
Plan 2
List price -2%
10% off, 3 wk · starts week 1
- Units
- +17.7%
- Gross margin
- 34.6%
- Gross profit
- -6.7%
Plan 3
List price +8%
20% off, 3 wk · starts week 3
- Units
- +15.8%
- Gross margin
- 34.4%
- Gross profit
- -8.7%
Every plan that reaches +15% units gives up some gross profit. The shortlist shows which one gives up the least, and the leader's figure above is already corrected for the winner's curse.
The correction
The best-looking plan is always a little too good.
Out of hundreds of candidates, the one that comes first is partly there because the model's uncertainty happened to fall in its favour. That is the winner's curse, and it does not go away by running more plans. Jonbar shrinks the leader's estimate towards the rest of the shortlist before you see it, so the number you plan with is the corrected one.
- Units
- Raw search result+19.6%After shrinkage+18.6%
- Gross profit
- Raw search result-5.2%After shrinkage-6%
Example, from the search above.
What the engine looks at
The inputs behind one search.
Nothing here is a guess you have to supply. Most of it comes out of your sales history; the rest is the market model for the category.
- Your weekly sales history per product: units, price actually paid, campaign weeks, stock-outs.
- Price response for the product, from your own price moves where there are enough of them and from the category model where there are not.
- Campaign after-effects: how much of the extra volume is pulled forward from the weeks that follow.
- Unit economics: cost, commission, shipping, returns and ad cost, so the ranking is on gross profit and not on units.
- The calendar of the month you are planning: sales period, pay day, shipping cut-offs.
- Your limits: gross margin floor, deepest discount you allow, stock on hand, campaign length.
What you get
Three plans, with what each one costs you.
A ranking is not a decision. What you need is the three live options and the reason each one sits where it does.
- 01
Three plans with different price levels
Not three variations of one idea: the shortlist keeps one plan per price level, so the options are actually different decisions.
- 02
A probable range, not a single number
Each plan carries the range the market model considers probable for units, gross margin and gross profit, so you can see when two plans overlap and the choice between them is not resolved.
- 03
The trade-off in one line
What the plan gives up to reach the goal: margin points, discount depth, or the dip in the weeks after the campaign.
- 04
The plans that were dropped, and why
Which limit each rejected plan broke. If the goal turns out to be unreachable inside your limits, that is the answer you get, with the limit that is binding.
Honest limits
What this does not do.
- The search is as good as the market model behind it. For a product with no price history of its own, the response comes from the category model and the range is wider.
- It covers price level, discount depth, campaign length and start week. Bundles, new launches and assortment changes are not in the search yet.
- A competitor who reacts the day after your move is modelled as a response pattern, not as a known plan.
- The engine ranks on gross profit over the horizon you give it. A longer brand effect from heavy discounting is outside that window.
- Every forecast is written down before the month starts, and the error is used to correct the model. Until a pilot runs, the numbers you see on this page are an example, not a track record.
Bring one goal and the limits around it.
The pilot is two weeks and one decision. You write the goal, we run the search on your own history and you get the three plans before the month starts.