Research

Competitive Landscape Matrix

Builds a comparison matrix across competitors using criteria derived from buyer needs, with explicit unknowns rather than plausible-sounding filler in empty cells.

Prompt template

You are a product analyst building a competitive matrix that a team will use
to make a real decision. An unknown cell is useful; a confidently wrong cell
destroys trust in the whole table.

## Inputs
Category: {{CATEGORY}}
Competitors to compare: {{COMPETITORS}}
Buyer and their situation: {{BUYER_PROFILE}}
Source material: {{SOURCES}}

## Step 1 — Derive criteria
From {{BUYER_PROFILE}}, derive 6-8 comparison criteria. Each must be a
decision a buyer would actually change their mind over. For each criterion,
state the buyer need it serves in one line.
Reject criteria that are: universally true of the category, unmeasurable, or
vendor marketing language.

## Step 2 — Fill the matrix
Produce a Markdown table: rows are criteria, columns are {{COMPETITORS}}.
Each cell uses one of:
- A specific factual statement, followed by [S#]
- `UNKNOWN` where the sources do not say
- `N/A` where the criterion does not apply to that product

## Step 3 — Read the matrix
Write four short sections:
- Where the products genuinely differ (ignore criteria where all are similar)
- Which competitor wins for this specific buyer, and on which criteria
- The two UNKNOWN cells that most affect the decision, and how to resolve them
- One criterion the buyer has probably not considered but should

## Constraints
- Never estimate pricing, performance figures, or dates. If the sources do
  not state it, the cell is UNKNOWN.
- Do not write a cell in vendor marketing language. Describe capability
  behavior, not positioning.
- Do not declare an overall winner independent of {{BUYER_PROFILE}}.
- It is acceptable for a column to be mostly UNKNOWN. Say so plainly rather
  than padding it.

Why this works

The key move is deriving criteria from the buyer before filling any cells. Ask for a comparison directly and the model reuses generic category dimensions that flatter whichever product it knows best. The UNKNOWN token is a structured escape hatch: a table cell is a strong completion pressure, and without a legal way to leave one empty the model fills it with plausible invention. This is the same failure mode as forced-format hallucination in extraction tasks.

Example output

Example

## Criteria
1. Self-hosting support — buyer is in a regulated industry [need: data residency]
2. Time to first working workflow — buyer has no dedicated platform team
3. Audit trail granularity — buyer must evidence decisions to a regulator

## Matrix
| Criterion | Product A | Product B | Product C |
|---|---|---|---|
| Self-hosting | Docker deployment documented [S1] | Cloud only [S2] | UNKNOWN |
| Time to first workflow | UNKNOWN | Templates for common cases [S2] | UNKNOWN |
| Audit trail | Per-step logs with inputs [S1] | Run-level logs only [S3] | UNKNOWN |

## Where they differ
Only on deployment model and audit granularity. Everything else is
comparable and should not drive the decision.

## Decisive unknowns
Product C is unevaluable from these sources — three of three cells unknown.
Either get a trial or drop it from the shortlist.

Usage tips

Paste real source material — documentation pages, pricing pages, changelogs — rather than relying on model memory. Competitive facts age fast, and a matrix built from training data is a matrix of last year’s products.

If more than about a third of the cells come back UNKNOWN, that is the finding. It means the research is not done yet, and the honest output is more valuable than a complete-looking table.

Extending it

Add a WEIGHT column where the buyer rates each criterion 1-5, then ask for a weighted read of the matrix in step 3. Making the weights explicit usually surfaces internal disagreement about priorities before the tool decision is made — which is the argument worth having first.

Browse library

Source-Grounded Research Brief

Produces a research brief where every claim is tagged with its source and confidence, and unsupported assertions are quarantined instead of blended into the text.

The AI signal without the hype — new models, tools worth your time and what actually shipped. One email, no vendor pitches.

Choosing an AI stack for your team?

We help companies pick the right AI tools, wire them into existing systems and avoid the ones that quietly do not scale. Tell us what you are trying to build and we will tell you what we would use — no charge for the conversation.