A market intelligence RFP should be built around a single principle: ask what a vendor can demonstrate, not what they can assert.
Most RFPs in this category fail that test. They are dominated by questions every vendor answers yes to, which is why they so reliably produce a shortlist of providers who look equivalent on paper and turn out not to be.
The test for a good question#
A useful RFP question meets two conditions:
- It can be answered with evidence — a document, a dataset, a demonstration.
- A poorly-equipped vendor cannot answer it the same way a strong one can.
| Vague question | Strong replacement |
|---|---|
| Do you have comprehensive coverage of our category? | List each platform you collect in our market, and name what is excluded. |
| Is your data accurate? | For each metric we will use, is it observed or estimated? If estimated, by what method? |
| How often is your data updated? | For each data type: collection cadence, processing lag, publication cadence — as three numbers. |
| Do you cover our competitors? | Show these ten named products in the sample, with history. |
| Can we integrate your data? | Provide a trial extraction in our required format at our required granularity. |
| Is your methodology sound? | What would have to change in it for this figure to move materially? |
Every replacement produces a checkable artefact.
The three questions that discriminate most#
1. What is excluded? From coverage, from a category definition, from a figure. Honest vendors answer specifically; evasive ones deflect to what they do cover. No vendor's marketing volunteers this, so the question has to be asked directly — and the quality of the answer is among the strongest signals in the whole process. See Data Coverage and Completeness.
2. Which figures are observed and which are estimated, and how? This separates vendors who have decomposed their own numbers from those who have not. A vendor that cannot draw the line internally cannot draw it for you. See Data Provenance and Due Diligence.
3. What would have to change for this figure to move materially? Rarely expected, and unanswerable without genuinely understanding the pipeline. A vendor who understands theirs names the classification rule or estimation assumption the figure is most sensitive to.
Require deliverables, not answers#
The most significant change a buyer can make to an RFP is to require artefacts:
- A category-scoped sample dataset — your platforms, your categories, your time window
- The methodology document that accompanies it
- A written coverage statement, per platform and per market, not in aggregate
- A demonstration answering a question you supplied, not one the vendor chose
The last point matters more than it looks. Letting each vendor pick the demonstration turns the comparison into a set of highlight reels; every vendor is strongest somewhere and will choose that ground.
Willingness to provide a scoped sample is itself diagnostic — it is precisely what a vendor with thin coverage in your category most wants to avoid. Sample Data Verification covers what to do with the sample once it arrives.
Commercial and legal questions that get left out#
These are cheap before signature and expensive afterwards.
Use rights, which are three separate permissions. May figures be redistributed to clients? Published in marketing material? Included in a regulatory or investor filing? A general subscription may grant none of them, and buyers routinely discover this after designing the data into a client deliverable.
Exit. What happens to already-delivered history if the relationship ends? Can it continue to be used, or does the licence lapse with the subscription? A multi-year analysis built on data you lose access to is a liability.
Change communication. How are methodology changes versioned, dated and notified? A change communicated only in release notes nobody reads is, in practice, undisclosed.
Export. Can the data leave the dashboard, at the granularity your systems need? Decisive for the hybrid approach in Build vs Buy Market Intelligence.
Continuity. What is the contingency if a source relationship ends or a platform changes its access terms?
Scoring#
Score against a fixed set of dimensions with one scale applied to every vendor, and record disqualifiers separately — a missing redistribution right, a category not covered, no export in the required format. These end the conversation regardless of a weighted total, and burying them inside a score is how a vendor with a fatal gap wins on aggregate. The dimension set and weighting method are in Market Intelligence Platform Comparison Criteria.
A note on scope#
Keep the shortlist to three. Each vendor added multiplies the work, because a real comparison means running the same tests against each on the same questions — and the time usually comes out of the sample testing, which is the part that actually discriminates.
Where to look next#
For the evaluation criteria the RFP is built on, see How to Evaluate Market Intelligence Providers. For scoring and weighting, see Market Intelligence Platform Comparison Criteria. For commercial structure, see Market Intelligence Pricing Models. For testing the sample, see Sample Data Verification.
Common questions#
What makes an RFP question useful?#
It can be answered with evidence rather than assertion, and a poorly-equipped vendor cannot answer it the same way as a strong one. "Do you have comprehensive coverage of our category?" fails both tests — every vendor says yes and nothing distinguishes the answers. "List the platforms you collect in our market, and name what is excluded" passes both: it produces a checkable artefact, and a vendor without the coverage cannot produce it. Most RFPs are dominated by the first kind, which is why they so often produce a shortlist of vendors who all look equivalent on paper.
Which questions discriminate most between vendors?#
Three do most of the work. First, what is excluded — from coverage, from a category definition, from a figure. Honest vendors answer specifically and evasive ones deflect, and no vendor's marketing volunteers this. Second, which figures are observed and which are estimated, and by what method — this separates vendors who have decomposed their own numbers from those who have not. Third, what would have to change for this figure to move materially, which almost nobody expects and which cannot be answered without genuinely understanding the pipeline.
What should the RFP require as deliverables, not answers?#
A category-scoped sample dataset covering your platforms, categories and time window; the methodology document that goes with it; a written statement of coverage boundaries per platform and per market; and a demonstration answering a question you supplied rather than one the vendor chose. Requiring artefacts rather than assurances changes what the process measures. A vendor's willingness to provide a scoped sample is itself among the more informative signals available, because it is the thing a vendor with thin coverage in your category most wants to avoid.
What is most often left out of a market intelligence RFP?#
Use rights, and they are usually discovered too late to renegotiate cheaply. Whether figures may be redistributed to clients, published in marketing material, or included in a regulatory or investor filing are three separate permissions, and a general subscription may grant none of them. Alongside those: what happens to already-delivered history if the relationship ends, how methodology changes are communicated, and whether the data can be exported at the granularity your systems need. Each is cheap to settle before signature and expensive afterwards.