Lock the criteria before vendors enter the room
Most evaluation problems start when criteria are written after the first demo. By then, a charismatic sales team has already shaped what the buying group thinks matters.
Before issuing the RFP:
- Collect requirements from each stakeholder group. Include end users, IT, security, finance and legal, not just the project sponsor.
- Sort requirements into three buckets. Pass/fail gates (for example, SSO support or data residency in a specific region), weighted criteria, and nice-to-haves that only break ties.
- Agree on weights and get sign-off. A short email from the sponsor confirming the weights prevents relitigating them later.
- Write the scoring rubric. Define what a 1, 3 and 5 look like for each criterion so scorers are measuring the same thing.
Pass/fail gates matter. A vendor that fails one should drop out before scoring, no matter how strong the rest of the offering looks.
The seven criteria groups
Adapt the detail to the purchase, but most software evaluations cover these areas:
- Functional fit: Does the product handle your core workflows out of the box, through configuration, or only through custom work? Configuration beats customization almost every time.
- Technical fit: Integrations with your existing systems, API quality and documentation, performance at your data volumes, and admin effort.
- Security and compliance: Certifications (SOC 2 Type II, ISO 27001), data handling, access controls, breach history, and any regulatory needs specific to your industry.
- Implementation and support: Realistic timeline, who does the work (vendor, partner or your team), support hours, escalation paths and named contacts.
- Vendor viability: Financial health, customer base in your segment, product roadmap, and ownership changes.
- Total cost of ownership: Licenses, implementation, training, integrations, internal staff time and expected price increases over the full term.
- Contract terms: Renewal protections, termination rights, SLAs, liability caps and data return.
A common mistake is folding price into every category. Keep cost in its own group so a cheap vendor with poor fit cannot score well by accident.
Weighting and scoring
There is no universal weighting, but a typical starting point for a business application looks like this. Adjust for your situation.
| Criteria group | Example weight |
|---|---|
| Functional fit | 30% |
| Technical fit | 15% |
| Security and compliance | 15% |
| Implementation and support | 10% |
| Vendor viability | 10% |
| Total cost of ownership | 15% |
| Contract terms | 5% |
For infrastructure or data-heavy tools, shift weight toward technical fit and security. For commodity tools with many similar vendors, cost and contract terms can carry more.
Use a 0 to 5 scale with written definitions. For example, for a functional requirement:
- 0: Not supported.
- 1: Possible only through custom development.
- 3: Supported through configuration or a workaround users would accept.
- 5: Supported out of the box and demonstrated live with your data.
Have each evaluator score independently before any group discussion. Then compare, and talk through any criterion where scores differ by two points or more.
Verify claims instead of trusting answers
RFP responses tell you what a vendor says. Your job is to confirm what the product does.
Scripted demos. Send each shortlisted vendor the same set of 5 to 10 scenarios drawn from your real workflows, with sample data. Ban slide decks during the demo portion. Score live against the rubric.
Proof of concept. For larger or riskier purchases, run a time-boxed trial with defined success criteria written in advance. Two to four weeks is a common length. Keep it scoped to the workflows that decide the purchase.
Reference calls. Ask for customers similar in size and use case, and try to find one or two yourself through your network. Useful questions:
- How long did implementation actually take compared with the original plan?
- What did you have to build or buy that you did not expect?
- How did the vendor handle your worst support issue?
- What happened at your first renewal?
- Would you choose them again?
The renewal question often reveals more than anything else on the call.
Security and viability due diligence
Run this in parallel with demos so it does not delay the decision.
Security checklist:
- Current SOC 2 Type II report or ISO 27001 certificate, with the bridge letter if the report is older than a few months
- Completed security questionnaire (your own, or a standard one such as SIG or CAIQ)
- Data locations and list of subprocessors
- Encryption at rest and in transit
- SSO, role-based access and audit logs
- Incident notification timeline and past incidents
- Penetration test summary from the last 12 months
Viability checklist:
- Years in business and funding or profitability status
- Number of customers comparable to you
- Any recent acquisition, layoffs or leadership turnover
- Roadmap for the features you depend on
For private vendors, you may only get partial financial information. Ask directly, and weigh a refusal to share anything as part of the score.
Commercial and contract criteria
Score the contract, not just the price. Two vendors with similar year-one quotes can look very different over a three-year term.
Build a total cost model covering the full term: licenses, implementation, training, integrations, premium support, and expected increases at renewal.
Contract terms worth scoring:
- Renewal price caps. Limits on annual increases.
- Termination rights. Exit for material breach, repeated SLA failures or change of control.
- SLAs with teeth. Uptime commitments tied to service credits.
- Data return. Export in a usable format at no extra cost.
- Liability caps. Adequate for the data involved, with carve-outs for data breach and confidentiality.
- Auto-renewal notice. A notice window you can realistically track.
Example renewal clause:
Upon renewal, Vendor may increase fees by no more than the lesser of 5% or the annual change in CPI for the preceding twelve months. Vendor shall provide written notice of any increase at least 90 days before the renewal date.
Example data return clause:
Within 30 days of termination or expiration, Vendor shall make all Customer Data available for export in a standard, machine-readable format at no additional charge, and shall delete Customer Data within 30 days after export confirmation.
A vendor's willingness to negotiate these points is itself a useful signal.
Make the decision and document it
Hold a final scoring session with all evaluators. Present the weighted totals, then discuss the top two candidates on their biggest risks, not just their scores.
The written recommendation should include:
- Final weighted scores for each vendor
- Pass/fail gate results
- Key risks for the chosen vendor and how the contract or implementation plan addresses them
- Three-year total cost comparison
- Dissenting views, if any
Keep the scoring sheets and notes. They support audit requirements, answer questions from losing vendors, and give you a baseline when the contract comes up for renewal and you need to decide whether to re-run the evaluation.
Let pingpong run it for you
pingpong drafts the RFP, finds and invites vendors, collects proposals through a private portal, scores them with five AI models and flags the gotchas above. It drafts every negotiation message for your approval, then keeps watching the market so you renegotiate before renewal. $100 for the first month, then $799 a month.
Common questions
How many vendors should we include in a software RFP?
A common approach is to send the RFP to four to six vendors, then shortlist two or three for scripted demos. Fewer than three limits your negotiating leverage and comparison points. More than six usually creates evaluation work without improving the decision.
Should price be the deciding factor if scores are close?
Look at total cost over the full term, not the year-one quote, before treating price as a tiebreaker. If scores and total costs are both close, contract terms such as renewal caps and exit rights are often a better deciding factor. They determine how much risk you carry after signing.
How do we stop stakeholders from favoring a vendor they already like?
Fix the criteria, weights and rubric before any vendor contact, and require independent scoring before group discussion. Scripted demos using your own scenarios reduce the effect of polished presentations. Documenting the reasons for each score also makes unsupported preferences easier to spot and discuss.