When to write your B2B demand generation software rubric

What to take away

The commonest mistake is booking demos first, then inventing criteria to justify the favourite. Write the rubric before a vendor sees your calendar.
CAP must-haves at three. Long lists push teams towards the biggest platform rather than the best fit.
Price three years of ownershipsubscription, data, implementation and internal hours.
Score dated evidence such as a reference call, a security questionnaire or a live export test.
Agree the exit before you sign, including how you get your data out.

Define the job before you look at vendors

Most selection projects fail at the start. The team writes a feature wish list instead of describing the job the software must do. That list grows with every demo, and nothing gets cut.

Define the job first

Write outcome in one sentence
Include a number and a date
Three must-haves, not twelve
Phrase must-haves as trial tests
Decide who signs before final week

Write the problem in one sentence

Describe the outcome in one sentence, with a number and a date. For example: double qualified pipeline from mid-market manufacturing accounts by March 2027. That sentence rules out tools which cannot reach those accounts.

Keep it short enough to repeat in a meeting without notes. If two stakeholders write different sentences, settle that gap before briefing any vendor, because the rubric inherits the disagreement.

If the wider plan is still unsettled, the tools and supplier guide for 2027 sets out the categories you may need before shortlisting anything.

Split must-haves from nice-to-haves

Three must-haves is a working number. Everything else belongs on a second list, scored but not decisive. Teams that keep twelve must-haves buy the platform that does all twelve badly.

Phrase each must-have as a test you can run in a trial account, such as importing 5,000 records with the source field intact. Show the list to the people who will use the tool, then cut anything that only sounds impressive.

Decide who signs. If marketing, sales and IT each hold a veto, settle that before the final week, not during it.

Build a scoring rubric you can defend

A rubric is a table of criteria, weights and evidence. It turns an argument about taste into a conversation about proof. Keep the scores visible to everyone on the panel.

Weighted scoring rubric

Criterion

Fit to target accounts: 25%
Data and consent handling: 20%
Total three-year cost: 20%
Handover to sales: 15%
Reporting you will act on: 10%
Support and exit terms: 10%

Weight

Fit to target accounts: Live data test
Data and consent handling: Contract and DPA
Total three-year cost: Written quote
Handover to sales: Sandbox test
Reporting you will act on: Your data imported
Support and exit terms: Contract clause

Evidence

Fit to target accounts
Data and consent handling
Total three-year cost
Handover to sales
Reporting you will act on
Support and exit terms

Weight the criteria

Weights should reflect your one-sentence problem, not the vendor's pricing tiers. This rubric is a workable starting point.

Criterion	Weight	What scores 5	Evidence
Fit to target accounts	25%	Native filters for your firmographics	Live data test
Data and consent handling	20%	Clear lawful basis, UK GDPR terms	Contract and DPA
Total three-year cost	20%	Within budget including services	Written quote
Handover to sales	15%	Leads reach the CRM with context	Sandbox test
Reporting you will act on	10%	Pipeline by source, not clicks	Your data imported
Support and exit terms	10%	Named contact, export in 30 days	Contract clause

Set a pass mark, for example 3.5 out of 5 on every must-have. A vendor can win on total score and still fail one.

Score evidence, not promises

Ask every vendor for the same artefact: a written answer, a screenshot from a live tenant, or a reference call. Platform pages are useful background. Salesforce's B2B marketing automation page shows how broadly vendors define this category. That breadth is why your rubric has to be narrower.

Score every criterion on the same day from the same evidence pack, so a later demo cannot move an earlier number. Where two vendors look identical, the supplier comparison method explains how to break a tie on cost and contract terms rather than features.

If paid media sits inside the plan, read IAB UK's Back to Basics guide to programmatic before you compare ad tech promises. It explains mechanics that buyers are often asked to accept on trust.

Price it and plan the exit

Cost the three-year total

List every line: subscription, onboarding, data enrichment, integrations and internal hours. A platform at £2,000 a month plus £30,000 of implementation is a different decision from a £2,000 a month subscription alone. Those figures are illustrative.

Agree the exit before signing

Will the vendor commit to exit terms in writing?

Yes

Export formats, notice periods and deletion terms fixed

Price that risk into the score

Ask for renewal uplifts in writing, because a percentage buried in the contract quietly changes the three-year total. Small teams often start with a free lead record such as HubSpot's CRM while they define requirements, then migrate. That keeps discovery cheap.

Agree the exit before you sign

Ask what happens in month thirteen. Get export formats, notice periods and deletion terms in writing. For example, a 60-day notice period decides when you must make the call. Note who owns the data during that notice period, and whether reporting survives to the end date. If the vendor will not commit, price that risk into the score.

Common questions

How long should selection take?

Six to ten weeks suits most mid-market teams. Longer projects lose momentum and let requirements drift.

How many vendors should we shortlist?

Three to five. Fewer gives no comparison, and more turns the rubric into a full-time job.

Who should sit on the panel?

Include the person who owns the pipeline number, plus whoever will run the tool daily. Add someone who can veto on data protection.

Do we need a consultant?

Only if you lack internal time to run the process. A consultant can build the rubric, but the panel must own the scores.