Choosing the right IT vendors is not primarily about picking the best vendor — it is about avoiding the wrong one. Across our engagements, the strongest single predictor of a successful vendor relationship is not vendor capability (most credible vendors can deliver) but vendor fit — the match between vendor model and buyer architecture, governance maturity, and strategic intent. The framework that works in practice scores vendors across five dimensions: capability, commercial, delivery model, relationship fit, and risk profile.
The wrong question.
The wrong question is "which vendor is the best?" Most credible vendors can deliver most credible scopes. Variability in vendor capability is real but usually less than variability in vendor fit. A globally-recognized top-tier vendor delivering against a scope that doesn't match their operating model will under-perform a less-prestigious vendor delivering against a scope that does. The selection question, framed productively, is "which vendor fits this engagement?"
The five evaluation dimensions.
1. Capability.
Can they actually do the work? This is table stakes; most credible vendors clear the bar. The question that matters here is not "do they have the skills" but "do they have the skills delivered in the form we need." A vendor with deep expertise in cloud migration may have that expertise sitting in a different geography, a different business unit, or a different practice from the one that would deliver to your engagement. The capability matters less than its accessibility.
Diagnostic questions:
- Reference engagements at comparable scale and complexity in the last 24 months.
- Named delivery resources, not just leadership names from the pitch deck.
- Concrete evidence of methodology, not just the existence of one.
2. Commercial.
Day-one price matters; structural pricing logic matters more. A vendor with a marginally higher headline rate but a more disciplined indexation model, clearer change-request mechanics, and better benchmark-refresh provisions usually produces lower total cost of ownership over the contract life.
Diagnostic questions:
- Rate card with volume bands, indexation rules, geographic breakdown.
- CR pricing mechanics — how does the vendor price scope changes against the master agreement?
- Benchmark provisions — is the vendor willing to commit to periodic benchmark-driven price resets?
- Exit and migration support pricing — what does it cost to leave?
3. Delivery model.
How will the work actually be delivered? Onshore vs nearshore vs offshore. Dedicated team vs shared pool. Fixed price vs T&M vs outcome-based. Each model has trade-offs; the question is whether the model matches the engagement.
For example: a fixed-price model works well when scope is genuinely fixed and well-defined. A T&M model works well when scope is fluid and the buyer has the governance maturity to manage utilization actively. An outcome-based model works well when outcomes are clearly measurable and the vendor has enough operational control to commit to them. Matching the model to the engagement is more important than the abstract merits of any particular model.
4. Relationship fit.
The dimension most often skipped, and frequently the dimension that decides whether a strategic relationship succeeds or fails. Two areas matter:
- Cultural compatibility. How does the vendor work? Aggressive escalation versus quiet collaboration. Engineering-led versus account-led. Specific governance ceremonies the vendor expects. Does that fit how your organization operates?
- Account model. What is your relative size in the vendor's portfolio? A tier-3 account in a large vendor's portfolio receives different attention than a tier-1 account in a mid-sized vendor's portfolio — and the latter often produces better outcomes despite the smaller vendor's smaller resources.
5. Risk profile.
The seven-category vendor risk view, applied at selection time. Financial health of the vendor. Concentration risk if they win this engagement on top of their existing scope. Geopolitical exposure. Security posture. Regulatory alignment. Exit posture. Reputational risk.
This is the dimension most likely to surface deal-breakers that price would have masked.
The decision matrix.
The standard tool is a weighted evaluation matrix with vendors scored across all five dimensions and sub-criteria within each. Weights reflect what matters for the specific engagement: a transformation program weights capability and relationship fit heavily; a commodity-replacement weights commercial and delivery model.
Critically, weights should be locked before responses are received. Adjusting weights after seeing responses is one of the most common ways evaluation matrices codify post-hoc bias.
The failure modes.
Brand selection.
Selecting the highest-prestige vendor because their brand provides cover for the decision. Common in regulated industries and at board-visible engagement levels. Often produces under-delivery because the buyer is not the vendor's most strategic client and does not receive their A team.
Relationship selection.
Selecting the vendor with the strongest existing relationship to the buyer's leadership. Often the right answer; sometimes a structural failure that bypasses the actual selection criteria. The discipline is to make the relationship-led decision visible and defended on its merits, not concealed inside a sham evaluation matrix.
Procurement-led race-to-bottom.
Selecting on day-one price without weighting structural commercial logic, delivery model fit, or relationship trajectory. Produces short-term savings and long-term cost as the contract decays through changes the procurement-led selection didn't anticipate.
Vendor capture.
Allowing the incumbent or preferred vendor to shape the requirements during the evaluation period. The deepest form of bias and the hardest to detect because the buyer believes they ran a fair evaluation. The fix is procedural — limit incumbent access to the requirements-definition phase, run blind reference checks, sequence the evaluation so vendor advocacy cannot influence weighting.
The Aventario perspective.
"Most failed engagements we audit had perfectly defensible vendor selection in retrospect — on paper. But the post-mortem usually surfaces one of two things: the wrong vendor was structurally favoured during requirements definition, or the right vendor was selected but the engagement was scoped for a delivery model the vendor's account team wasn't actually structured to deliver."
— Markus Kern, CEO, Aventario
FAQ.
What is the most under-weighted vendor selection criterion?
Relationship fit — specifically the buyer's relative size in the vendor's portfolio. A tier-1 account in a mid-sized vendor's portfolio usually receives stronger attention than a tier-3 account in a tier-1 vendor's portfolio, and the resulting service levels reflect that asymmetry.
How many vendors should the shortlist contain?
For a complex IT engagement, 3–5 shortlisted vendors after the RFI. Fewer creates the appearance of inadequate market exploration; more produces evaluator fatigue and dilutes attention. The strongest selection processes spend more time on fewer vendors rather than running a broad pool.
Should the existing vendor be included in the RFP?
Almost always yes — incumbents should compete on the same terms as challengers. The exception is structural: where the existing relationship is being deliberately exited or where their inclusion would compromise the evaluation discipline.
