Filled preview
Illustrative example. Names, volumes, dates and outcomes are fictional; replace them with approved evidence before use.
| Memo field | Example entry | Evidence status |
|---|---|---|
| Decision requested | Approve a 10-business-day POC with Northwind Retrieval, not the annual license | Names the actual ask |
| Problem and cost of delay | 12 analyst hours/week locating exception evidence | Baseline the controller confirmed |
| Recommendation | POC first; license decision after acceptance criteria are met | Sequenced, not assumed |
| Cost and source | $18,000 first year from the vendor quote dated 2026-09-12 | Quote attached |
| Evidence attached | POC acceptance criteria; security review status | Two open items named |
Sources: Purdue OWL: Parts of a Memo
What a decision memo is for
A decision memo moves a purchase from evaluation to approval. It is not the analysis. The business case, security review and pilot results feed it; the memo condenses them into a request a busy approver can act on.
Purdue OWL’s memo guidance describes the genre the same way: a memo serves an audience and a purpose, and its parts (heading, opening, context, task, summary, closing) exist so the reader can find the request quickly.
Sources: Purdue OWL: Memos, general introduction · Purdue OWL: Parts of a Memo
When to write one (and when not to)
- Write one when a specific person can decide: budget approval, pilot sign-off, contract signature. If nobody can decide from it, it is a status update, not a decision memo.
- Write it after the evaluation produces something to decide: a shortlist, pilot results, a quote. A decision memo written before the evidence exists becomes a wish.
- Do not write one to introduce a product. That is what a brief or a demo is for. An approver who has never heard of the problem needs a conversation first, not a memo.
- For purchases that need a full cost model and sensitivity analysis, write the business case first and let the memo reference it. Keep the memo to what the approver must decide.
What belongs in it
| Section | Question it answers | Common failure |
|---|---|---|
| Decision requested | What exactly are you asking me to approve, and by when? | Vague asks: “seeking alignment” instead of “approve $X by date” |
| Problem and cost of delay | Why now, in the approver’s terms? | Product-first framing; the problem is asserted, not evidenced |
| Recommendation | What should we do? | Offering a menu instead of a recommendation |
| Evidence | Why should I believe it will work? | Claims with no source; pilot results without acceptance criteria |
| Cost and terms | What does it cost, total, and on what terms? | License price only; implementation and renewal left out |
| Alternatives considered | Including doing nothing? | Straw-man alternatives; no cost of inaction |
| Risks and open questions | What could make this wrong? | No risks listed; approver finds the first risk themselves |
| Reviews completed | Who has already vetted it? | Security, finance, or IT sign-offs missing, so the approver re-runs the review |
Sources: Purdue OWL: Parts of a Memo
The template
Keep it to one page, two at most. Name a decision owner in the heading. The blank template below is also downloadable above.
Decision requested: approve [specific action and amount] by [date].
Problem: [what the team cannot do today, with the baseline number and who confirmed it]. Cost of delay: [what waiting costs or risks].
Recommendation: [chosen option], because [the one decisive reason]. Alternatives considered: [do nothing / option B / build], rejected because [reason].
Evidence: [pilot results with acceptance criteria, references, or evaluation notes, each with a source].
Cost and terms: [first-year and recurring cost from the quote, implementation effort, renewal terms]. Source: [quote or pricing page, dated].
Risks and open questions: [top two, with mitigations or named owners].
Reviews completed: [security / finance / IT, named, with status].
Attachment: [business case, POC results, security review status].
Filled example
Illustrative example. Names, volumes, dates and outcomes are fictional; replace them with approved evidence before use.
Decision requested: approve a ten-business-day proof of concept with Northwind Retrieval at no cost, and pre-approve the annual license of $18,000 only if the POC acceptance criteria are met. Decision by October 17.
Problem: our four financial analysts spend roughly 12 hours a week locating the evidence behind invoice exceptions. The controller timed two representative weeks and confirmed the baseline. Each week of delay costs about 12 analyst hours, or 1.5 analyst-days.
Recommendation: run the Northwind POC against 30 selected exceptions before any license commitment. Alternatives considered: hiring a contractor to rebuild the manual export (slower, does not survive volume growth) and doing nothing (the quarter-end backlog stays where it is).
Evidence: the vendor documented read-only access and workspace-level roles; our security lead reviewed the access design and listed two open questions in the attached review status.
The POC passes only if 28 of 30 selected records link to the correct retained source within the users’ existing roles.
Cost and terms: $18,000 first year per the quote dated September 12; renewal and usage terms attached. Implementation is estimated at 6 hours of IT time (an estimate, not a quote).
Risks: retrieval quality on scanned pre-2022 records is unproven and is exactly what the POC tests; if it fails, the cost of the POC is the IT hours already estimated.
Reviews completed: security lead (two open questions attached), controller (baseline confirmed), IT (access prerequisites confirmed).
Sources: Purdue OWL: Parts of a Memo
Why this example works
- The ask is sequenced and falsifiable: the approver is not asked to believe a benefit claim, only to fund a test whose pass condition is written down.
- The baseline is attributed (“the controller timed two representative weeks”) instead of presented as an external fact.
- Costs separate quote from estimate, and the renewal terms exist somewhere the approver can check.
- The do-nothing alternative has a stated consequence rather than a dramatized one.
- Open security questions appear in the memo itself, so the approver can see nothing was left out.
Evidence rules before you send
- Every number is either sourced (quote, timed baseline, pilot result with acceptance criteria) or labeled as an assumption with a range. Never blend the two into one confident figure.
- If the pilot has not run, the memo asks to fund the pilot. It does not narrate the benefits the pilot might show.
- Security and compliance statements come from the review owner, not from the vendor’s marketing page. Link the review status, open items included.
- Date every source. A quote without a date is a claim.
Writing it so it gets forwarded
- Assume the memo travels without you: the approver will forward it to finance or security with “thoughts?” and nothing else. Every term someone might trip on is defined in the memo.
- Neutral tone survives forwarding. Adjectives do not; named sources and attached evidence do.
- One decision owner, named. Copies to people who informed the recommendation are fine; the decision belongs to one person.
- If the memo needs a slide deck to be understandable, it is not ready. The template’s discipline is that the whole case fits in the memo.
Sources and dates
- Purdue OWL: Memos, general introduction ↗
Memo genre conventions: audience and purpose, parts, format. Verified 2026-09-29.
Reviewed 2026-09-29 - Purdue OWL: Parts of a Memo ↗
Heading, opening, context, task, summary, discussion, closing, attachments. Verified 2026-09-29.
Reviewed 2026-09-29 - AccountMade blog: keeping buyer content traceable ↗
AccountMade blog post on keeping buyer documents traceable to their sources.
Reviewed 2026-09-08
Found an error or a changed source? Send a correction.
Apply this to your company
Draft your next buyer document from approved claims.
Briefs and decks from approved facts, every claim sourced.
See the demo →