The context
Every trade promotion starts as a brief from the brand team: which products, which shops, what discount or free goods, what budget and for how long. Someone then has to set it up in the DMS, field by field.
On my last DMS rollout I configured and tested six promotion types, on-invoice and off-invoice, for two markets. This concept uses that experience to design an AI-assisted set-up flow.
The problem
Set-up is manual, and mistakes surface only on invoices:
- Briefs arrive as PDFs and emails, worded differently each time.
- Each promotion is keyed into the DMS by hand.
- Overlapping schemes and wrong dates are found only when invoices go wrong.
- Corrections reach distributors and salesmen late.
What I did
- Listed the fields each of the six promotion types needs, from my configuration and testing work.
- Designed the extraction, validation and approval steps.
- Defined conflict rules for overlaps, budgets, dates and eligibility.
- Wrote a field-mapping sample and acceptance criteria for the draft step.
The process, before and after
Each column is a team or system. Numbered badges point to where time is lost today, and how the design saves it.
Scroll sideways to see the whole diagram →
- 1Promotion rules arrive as a PDF or email, worded differently each time.
- 2Each scheme is typed into the DMS by hand, field by field.
- 3Mistakes surface only when distributors see wrong discounts on invoices.
Scroll sideways to see the whole diagram →
- 1The assistant fills type, products, shops, value, budget and dates from the brief.
- 2Rules flag overlapping schemes, missing budgets and invalid dates before anything goes live.
- 3A person approves every promotion. The assistant only drafts.
What changed, step by step
| Step | Before | After |
|---|---|---|
| Promotion brief | PDF or email, any format | Standard brief per promotion type |
| Set-up | Keyed in by hand | Drafted automatically, then reviewed |
| Conflicts | Found on wrong invoices | Flagged before approval |
| Going live | DMS and app updated separately | Published to both at once |
The approach
- 1Standardise the briefOne brief template per promotion type, so the assistant always knows what to look for.
- 2Extract into a draftThe AI assistant reads the brief and fills a draft set-up: type, products, shops, value, budget and dates.
- 3Check before anyone approvesRules flag overlapping schemes, missing budgets, invalid dates and products not on the price list.
- 4Keep a person in chargeThe brand or trade marketing owner reviews and approves every draft. Nothing goes live on the AI's word alone.
- 5Publish onceThe approved promotion goes to the DMS and the salesman app together, so both show the same scheme.
Sample work: field mapping extract
| Field | From the brief | Draft value | Check |
|---|---|---|---|
| Promotion type | Buy 10, get 1 free | Free goods, on-invoice | Type is one of the six set up in the DMS |
| Products | All 1 litre packs | 3 product codes | Every code is on the active price list |
| Shops | Grocery outlets, North region | Channel: grocery · Region: North | No overlap with another live scheme |
| Dates | Whole of next month | 1 to 30 November | Starts after approval and ends inside the budget period |
| Budget | Up to 5 lakh | ₹5,00,000 | Budget is set and not already used |
Illustrative extract written for this portfolio.
Expected outcomes
- Faster set-up, because nobody types promotion fields in by hand.
- Fewer wrong discounts reaching invoices.
- Distributors and salesmen see the same scheme on the same day.
- A clear approval record for every promotion.
Design principles
- Standardise the input before automating it.
- AI drafts, people approve.
- Catch conflicts before launch. Fixing invoices costs far more.
Concept case study written during my 2026 sabbatical. It is not a deployed client solution. Examples are generic.
Next case study · AI-03Automating distributor claim settlement