Enterprise Intelligence Group
Retail and CPG Planning Software RFP: Requirements, Demo Scenarios and a Scoring Sheet
A free RFP template for retail and CPG planning software: requirements, demo scenarios and a scoring sheet. Download the PDF and the CSV. No form.
One decision before one platform
Most RFP templates give you a list of features. This one starts with a decision. Name the one planning decision the software must improve, then write the RFP around it.
Use it to write requirements, to run demos on your own history, and to score what you see. It is free. There is no form and no email to give.
Download the scoring sheet (CSV)
Free. No form, no email. The CSV opens in any spreadsheet. The weights column is empty on purpose.
Using this on a live selection? Book a free 30 minute call.
Part of Retail and CPG technology selection.
THE TEST MOST TEMPLATES LEAVE OUT
Make each finalist replay one of your own past planning cycles
Pick one planning cycle that is finished. It could be last quarter’s promotion plan, or the forecast for one product family through a busy season. Give every finalist the same data from that cycle. Ask them to run it on their software, with your data, while your planners watch.
The cycle will hold overrides. An override is a time when a planner changed the number the system gave. Overrides are where your team’s knowledge meets the plan.
For each override, the finalist must show four things
- Where it is kept. The override sits in a record inside the product. A note in a side file does not count.
- Why it was made. The reason is saved, in the planner’s own words.
- Who approved it. A named person, and when.
- What happened next. The outcome, set beside the plan it changed.
Then ask one question about the contract
If our contract ends, does that record leave with us? Show us the export. It must hold every field above, in a file format we can open without your software, and say how long it takes you to deliver it.
Why this test matters
If the software cannot keep the reason, the approver and the outcome, that knowledge is lost the next time a planner moves on. If the record cannot leave with you, you are renting your own history.
A feature list will not show this. A replay will.
- Choose a closed cycle that your planners remember well.
- Ask the planners who made the changes to list the overrides they remember, and check the list against your own files.
- Share the data with each finalist in the same form, using only data you are allowed to share.
- In the demo, ask the finalist to open each override and show the four things. Do not accept slides.
- Score what was shown, using the 0 to 3 evidence scale on the main page. A promise on a roadmap scores 0.
The scoring levels are on the technology selection page.
What a replay looks like
ILLUSTRATIVE. Invented data. Not a client, not a company, not a real result.
The buying team replays a closed spring promotion for one shelf-stable snack, sold to one customer, across twelve weeks. Three finalists, called A, B and C here, are invented too. They are not real products.
| OVERRIDE IN THE CYCLE | REASON | APPROVER | OUTCOME |
|---|---|---|---|
| Week 6 forecast raised by a few hundred cases | The customer moved a display into a second region. The planner heard it by phone before it reached any system. | Demand planning lead, entered in week 5 | Orders landed close to the new number. The old number would have fallen well short. |
| Week 9 forecast lowered | A competing offer ran in the same weeks. The planner remembered the same clash from the year before. | Category manager, entered in week 8 | Orders landed below the old number and close to the new one. |
| Week 11 order moved forward by one week | The distributor asked for early stock ahead of a store reset. | Supply planner, entered in week 10 | The stock arrived in time. The following week then ran short. |
What each finalist showed
| FINALIST | WHERE KEPT, REASON, APPROVER, OUTCOME | LEAVES WITH THE BUYER? | SCORE ON THIS TEST |
|---|---|---|---|
| A | All four shown on one screen for each override. | Yes. Export holds every field in an open file format. | 3: Demonstrated for the agreed scenario |
| B | Reason and approver kept. The outcome sits in a separate report that is not linked to the override. | Partly. Export holds the values but not the reasons. | 2: Demonstrated with gaps |
| C | The reason is a free text note. No approver is saved. | No. Notes are not in the export. | 1: Partially shown |
No weights appear here, and no finalist is recommended. The example only shows what to look for. The buyer agrees its own weights before any demo.
Write requirements around the decision, not a feature list
A feature list tells a vendor what to say yes to. A decision tells them what to prove. Use one sentence for each requirement:
For [the decision], [the person who acts] needs [the evidence] by [the time] so that [the action can happen]. We will know it works when [what we can see in a demo].
| A FEATURE LIST LINE | THE SAME NEED, WRITTEN AROUND THE DECISION |
|---|---|
| Supports promotion planning. | For the weekly promotion plan for one customer, the account planner needs the last three comparable promotions and their results before the plan is sent, so that the plan is changed on evidence. We will see this when the finalist replays a past promotion. |
| Has a forecast override function. | When a planner changes a forecast, the change, the reason, the approver and the later result are kept together, so that the next planner can learn from it. We will see this in the replay test. |
| Integrates with our ERP. | The approved plan reaches order entry without re-keying, and a change made there comes back to the plan. We will see the round trip on our sample data. |
- Mark each requirement as mandatory or a preference. Do not let a preference sit in the mandatory list.
- Attach one scenario to each requirement. If you cannot say how you would test it, rewrite it.
- Name who is responsible for data preparation, integration and measuring forecast quality. Keep these inside the evaluation, not for later.
See the short answer on running an RFP for CPG forecasting software. See a business requirements document example.
Five exception scenarios every finalist must run on your history
Take these from your own files and from your planners, not from a script the vendor prepared. The main page has five general scenarios. These five are built from your own past. See the five general scenarios and the scoring levels.
1. A promotion that did not go to plan
- Take from your history
- One promotion where sales ran well above or well below the plan.
- Ask the finalist to show
- How the plan changed while the event ran, who changed it, and how the review after the event reads.
- Write down
- The changes, their reasons and approvers, and whether the review can be found later.
2. A supply shortfall
- Take from your history
- A week when a supplier or plant could not ship the full quantity.
- Ask the finalist to show
- How the shortfall is spread across customers, who decides it, and how each customer is told.
- Write down
- The rule used, the person who decided, and what each customer was told.
3. An override that was right, and one that was wrong
- Take from your history
- Two overrides from your files. Do not tell the finalist which was which.
- Ask the finalist to show
- How the product keeps the reason and the outcome so that a later planner can tell them apart.
- Write down
- The reason, the approver, the result and where each is stored.
4. A new or changed item with thin history
- Take from your history
- A recent item with little sales history, or one whose pack size changed.
- Ask the finalist to show
- What the product assumes, where the assumption is shown, and how a planner corrects it.
- Write down
- The source of each assumption, the data flag, the correction and who approved it.
5. A customer changes the plan late
- Take from your history
- A case where a retailer or distributor moved dates, quantities or the offer after the plan was set.
- Ask the finalist to show
- How the change is entered, who is told inside your company, and how your answer goes back.
- Write down
- The original plan, the change, the reply and the time each step happened.
The plan you share with suppliers and retailers
Many planning products promise a shared plan. Ask five questions about it, and have the finalist show each answer in the product.
| QUESTION | WHAT TO ASK | A GOOD ANSWER SHOWS |
|---|---|---|
| Who owns it? | Who owns each line of the shared plan, on your side and on theirs? | A named owner for each decision, and a history of every change with the person who made it. |
| How does an exception travel? | A planner spots a problem. How does it reach the partner who has to act? | Who is told, in what form, how the partner replies, and where the reply is kept. |
| Where does a distributor’s view enter? | Distributor stock and orders often differ from retailer sell-through. Where do they go? | A place to enter them, labelled by source, kept apart from retailer data, with a named person who can change them. |
| Who may see what? | Which partner can see which lines, and which fields? | Permissions by partner and by field, and a log of who looked. |
| What happens when the shared plan and the actual order disagree? | Replay a case where they did. | Which one wins, who is told, and whether the gap and the reason are kept. |
Read how to separate sell-in from sell-through.
AGENTS IN PLANNING
What may an AI agent do in planning? Set the level before the demo
Some finalists will show an AI agent. Do not ask whether it is smart. Ask what it is allowed to do, and what proves it can be trusted with more. These five levels give you the words.
| LEVEL | WHAT THE AGENT DOES | AN EXAMPLE IN PLANNING |
|---|---|---|
| Retrieve | Finds and shows information. A person does the rest. | Pulls last year’s results for a promotion and the notes planners left on it. |
| Prepare | Puts together a draft. A person opens it and decides what to do. | Drafts the weekly list of plan changes for the planner to review. |
| Recommend | Proposes a change and gives the reasons. A person decides. | Suggests lowering a forecast and shows the evidence behind it. |
| Act with approval | Makes the change after a named person approves it. | Updates the plan after the planner approves, then tells the partner. |
| Bounded action | Makes one defined kind of change alone, inside limits, and reports it. | Adjusts a replenishment setting inside a range the buyer set. |
What evidence moves a change up or back
Move a kind of change up one level when
- The same kind of change has been made many times at the level below.
- A person reviewed each one, and the results were recorded.
- Every record holds the reason, the approver and the outcome.
- The limits are written down and tested.
Move it back one level when
- A change had to be reversed.
- A record is missing the reason, the approver or the outcome.
- The data behind the change was wrong.
- A new type of item, customer or season appears.
- Nobody can say who approved it.
Three questions for every finalist
- At which level does the product run each of the five scenarios today?
- Can we set a ceiling for each decision?
- Can we move a decision down one level in a single step, and is that move recorded?
A scoring sheet with weights agreed before any demo
A weight set after the demos is a weight picked to fit a favorite. Agree the weights first, in writing, with everyone who will score.
- List your requirements, written around the decision. Add the rows below that apply.
- Mark each row mandatory or not.
- Agree a weight for each row with the whole buying team. Write the date. Do this before any demo is booked.
- Score what you see from 0 to 3 using the evidence scale on the main page. Write where the evidence is kept.
- Do not change a weight after a demo. If you must, write the reason and the date.
- A failed mandatory row stays failed, whatever the total says.
This sheet sets no weights and recommends no finalist. Every weight cell is empty so that your team fills it in.
| ID | AREA | WHAT THE FINALIST MUST SHOW | WEIGHT |
|---|---|---|---|
| R1 | Override record | Each override is kept in a record inside the product, with its date and the person who made it. | Agree before any demo |
| R2 | Override record | Each override keeps its reason, its approver and its later outcome together. | Agree before any demo |
| R3 | Override record | If the contract ends, the full record leaves with the buyer in an open file format, with a stated delivery time. | Agree before any demo |
| Q1 | Requirements | Each requirement is met on a replay of the buyer’s own past planning cycle, not on sample data alone. | Agree before any demo |
| S1 | Scenario | A promotion that did not go to plan. | Agree before any demo |
| S2 | Scenario | A supply shortfall. | Agree before any demo |
| S3 | Scenario | An override that was right, and one that was wrong. | Agree before any demo |
| S4 | Scenario | A new or changed item with thin history. | Agree before any demo |
| S5 | Scenario | A customer changes the plan late. | Agree before any demo |
| P1 | Shared plan | Each line has a named owner and a history of changes. | Agree before any demo |
| P2 | Shared plan | An exception reaches the partner, and the reply is kept. | Agree before any demo |
| P3 | Shared plan | A distributor’s view has its own labelled place and an owner. | Agree before any demo |
| P4 | Shared plan | Visibility is set by partner and by field, and viewing is logged. | Agree before any demo |
| P5 | Shared plan | When the shared plan and the actual order disagree, the outcome, the gap and the reason are kept. | Agree before any demo |
| A1 | Agent permissions | The level is set for each decision, with a ceiling the buyer controls. | Agree before any demo |
| A2 | Agent permissions | The evidence that moved a change up or back is recorded. | Agree before any demo |
| A3 | Agent permissions | A decision can be moved down one level in one step, and the move is recorded. | Agree before any demo |
| D1 | Delivery | Data preparation, integration and the way forecast quality is measured have named owners. | Agree before any demo |
Download the scoring sheet (CSV)
FREQUENTLY ASKED QUESTIONS
How to run an RFP for CPG forecasting software?
Start with the forecast decision, not a feature list. Ask each finalist to replay one of your own past planning cycles on your data, then score the demos against weights you agreed first. The short answer is on the technology selection page. Read the short answer.
What is RFP in CPG?
An RFP is a formal request a consumer packaged goods company sends to chosen suppliers, asking how they would meet set requirements. The plain definition is on the technology selection page. Read the definition.
How do I build a TPO vendor shortlist for an enterprise CPG?
Shortlist against your own promotion decisions, not a market map. Then ask each finalist to replay a promotion your retailer actually ran. The five steps are on the technology selection page. Read the five steps.
Is the template really free?
Yes. The PDF and the CSV scoring sheet download without a form or an email address.
Can I use it for pricing, assortment or space planning software?
Yes. The replay test, the shared plan questions and the permission levels apply to any planning decision. Swap the five scenarios for ones from your own history.
This template is a starting point. Your procurement and legal teams own the contract terms.
IF YOU WANT HELP
Want a second pair of eyes on your RFP?
Tell us which planning decision you are buying for and where you are in the process. You can also book a free 30 minute session on the same page.
Discuss your technology selection
The EIG Library · Insights · AI Maturity Assessment · Contact