18% discount for 4 weeks
- Incremental contribution
- −9.8 €M
- Trade spend
- 39 €M
- Store coverage
- +0.02 pp
- Below the minimum contribution guardrail: not ranked, not sent.
Signature product · Promotion Intelligence
Plan every promotion from what your plant can profitably deliver.
14 specialist agents design, compare and check each promotion against the factory, then recommend one for your team's approval. The trade plan becomes a consequence of what the plant can make, store and sell fresh.
Agents prepare, people decide
Want to see it on your own categories? A call is the fastest way to find out what the pilot would cover.
The problem
Promotions are planned at the wrong end of the value chain.
Most promotions start with the retailer and a depth of discount, and the factory hears about them last. Four things then take the profit you planned.
- 01
A subsidised baseline
Volume that would have sold anyway is paid for again, and counted as uplift.
- 02
Forward buying
The trade loads stock at the deal price, and the weeks after the promotion pay for it.
- 03
Cannibalisation
Extra units on the promoted product come out of your own range next to it.
- 04
Reflex and surprise
Last year's calendar is repeated by habit, and the factory hears about it too late to build ahead.
Start with the factory
Plan from the plant, and the trade plan follows.
Promotion Intelligence starts where the cost and the constraints are: raw material, lines, cold store and shelf life. Every design is checked against what the plant can profitably deliver before a retailer commits to it.
- Raw materialWhat arrives, when, and what it costs.
- LinesCapacity, changeovers and the weeks already full.
- Cold storeSpace to build stock ahead of the event.
- Shelf lifeHow early you can make it and still sell it fresh.
- Trade planThe promotions the plant can supply, at the margin you planned.
Every kind of promotion
Three kinds of promotion, judged differently.
A growth push, a seasonal event and a reply to a competitor are not the same decision, so each is measured against its own baseline.
- 01
Strategic growth
Judged against an undisturbed baseline.
For example: grow distribution in key accounts.
- 02
Seasonal event
Judged against the event baseline. The date never moves; the factory builds ahead within shelf life.
For example: New Year, Back to School, summer.
- 03
Competitive response
Judged against holding: what is lost if you do not react.
For example: a competitor offer detected at 20% off.
How it works
Four stages, 14 agents, one recommendation.
Context, response, economics and control, then a decision. When the plan needs the factory, the feasibility agent asks Manufacturing Intelligence through a typed contract and gets a typed answer back.
- Stage 14 agents
Context
Market, competitor activity, the event calendar, baseline demand and account behaviour are established.
- Competitor watch
- Event calendar
- Baseline
- Account behaviour
- Stage 24 agents
Response
Uplift, cannibalisation, forward buying and distribution effects are estimated for each design.
- Uplift
- Cannibalisation
- Forward buying
- Distribution effect
- Stage 33 agents
Economics and control
Trade economics are computed, the factory is asked for feasibility, and governance rules are checked.
- Trade economics
- Factory feasibility
- Governance
- Stage 43 agents
Decision
Designs are ranked against the objectives. A person approves, submits or revises, and every result feeds the next plan.
- Strategy designer
- Ranking
- Learning loop
- Your team approvesSubmit, revise or reject.
14 specialist agents in 4 stages. Every result feeds the next plan.
The request
Sent by the factory-feasibility agent.
- category
- tonnes
- weeks
- region
- uncertainty
Manufacturing Intelligence12 agents above the ERP
A typed contract, not an email to the plant. The factory answers in the same shape every time, from the same data your planners use.
- Raw material intake
- Demand
- Capacity
- Changeover
- Materials
- Cold store
- Shelf life
- Maintenance
- Quality
- Schedule
- Stress test
- Cost
The answer
Returned to the plan, for your team's approval.
- feasible volume
- confidence
- extra cost
- better start week
Inside the product
You set the direction. The agents do the analysis.
Eight moments from the product, on illustrative data: from where growth is to the factory check and the learning loop.
We can walk through the product live on a call, with your questions and your categories in mind.
Objectives and guardrails
Your leadership sets the weights. The guardrails decide what is allowed.
Designs are ranked by a weighted score across eight objectives. A design that breaks a guardrail is not ranked at all.
Eight objectives, weighted by leadership
- 1Incremental profit
- 2Absolute contribution
- 3Store coverage
- 4Market share
- 5Invoice lines
- 6Net volume
- 7Active products per store
- 8Lasting effect after the promotion
Every ranking shows the weighted score behind it, so changing a weight is a visible leadership decision, not a hidden setting. In the worked example below the ask is distribution, and the recommended design ranks first on store coverage.
Seven guardrails
- Budget
- Minimum contribution
- Maximum discount
- Cannibalisation ceiling
- Capacity and supply confidence
- Shelf-life protection
- Retailer contract rules
Worked example · illustrative
Same ask, two designs.
A dairy brand wants wider distribution of its cheese range. The deeper discount looks like the obvious answer. The numbers say otherwise.
The ask
Grow distribution of the cheese range. A dairy brand, illustrative data.
5% range incentive for 4 weeks
- Incremental contribution
- +3.6 €M
- Trade spend
- 5.0 €M
- Store coverage
- +0.85 pp
- Supply confidence
- 93 %
Ranked first against the weighted objectives.
The mechanism, not the depth, decided the outcome.
The deep discount pays for volume the brand would have sold anyway and barely moves coverage. The range incentive rewards stores for carrying more of the range, so a fraction of the spend wins far more coverage, and the factory can supply it.
- Uplift contribution+7.1
- Trade spend−5.0
- Cannibalisation−1.2
- Halo on other products+2.7
- Incremental contribution+3.6
Halo and cannibalisation are the economics agent's figures; the post-promotion dip is netted into the uplift. Illustrative data, not a client result.
Bring one of last year's promotions to a call. We will show how it would have been designed, checked and ranked.
The closed loop
Plan, execute, measure, learn.
Results are measured against matched control stores, not against hope, and every result changes the next plan.
- 01PlanDesigns ranked against your objectives, within the guardrails, checked with the factory.
- 02ExecuteThe approved plan goes to the promotion calendar and the ERP, after a human decision.
- 03MeasureResults read against matched control stores, so the uplift is the real one.
- 04LearnResponse estimates are updated, and the next plan starts from what actually happened.
What it is made of
6 agent teams from the catalogue.
Promotion Intelligence is assembled from 6 of the 48 catalogue teams, on the same platform as the rest. Open one to see its agents, its data and who approves.
Works with Manufacturing Intelligence, which answers the factory's side of every plan. See all 48 teams.
The path
From pilot to scale, one decision at a time.
Each phase ends with a decision point for you. Nothing moves to the next phase until you say so.
- Phase 04 to 6 weeks
Discovery
Your data, systems, objectives and guardrails mapped. One category chosen for the pilot, and the baseline agreed.
Your decision: Go to pilot, or stop. - Phase 1About one quarter
Pilot
One category and all three kinds of promotion, run alongside today's process and backtested on recent history.
Your decision: Scale, adjust, or stop. - Phase 2Two quarters
Scale
More categories, regions and accounts on the same platform, with the factory check connected.
Your decision: Integrate with S&OP, or hold. - Phase 3Ongoing
Integrated S&OP
Promotions planned inside the S&OP cycle, with one set of numbers from the factory to the shelf.
Your decision: Your roadmap, reviewed each cycle.
Questions
Data, systems, governance and security.
The questions a commercial, supply or IT leader asks first.
Weekly sales by product and customer, your promotion history (mechanism, depth, dates), prices and trade terms, and for the factory check: capacity by line, stock and shelf life. Discovery confirms what you have and what the pilot category needs; gaps are flagged as gaps, never filled with guesses.
Where to start
See where agentic AI could pay in your business.
Answer ten questions and get a free written report on your own workflows, in about ten minutes. Or bring one workflow to a free 20-minute call.