On paper, splitting an infrastructure project across six specialist vendors looks cheaper than awarding it to a single integrator. Each line item is competitively quoted, no one is bundling margin on top of margin, and the buyer keeps direct control of every relationship. In practice, the same scope under multi-vendor delivery routinely ends up 15–30% more expensive once the project closes — and that figure excludes the operational cost of the disputes that follow.
This article is for buyers evaluating whether to award an integrated infrastructure scope to a single enterprise systems integrator or to break it apart. It describes where the hidden costs of multi-vendor delivery actually sit, how to evaluate an integrator on something other than headline price, and the project profiles where each model is appropriate.
Section 1
Where multi-vendor delivery actually leaks money
The headline savings on a multi-vendor BoQ are real on day one. The leaks appear between day 30 and project closure, and they show up in predictable places: scope gaps at vendor boundaries, duplicated mobilisation costs, untracked variation orders, and the buyer's own staff time absorbed into coordination work that no one priced.
- Scope gaps at vendor boundaries: the cabling contractor leaves a service loop the switching vendor cannot terminate; both invoice for the rework.
- Duplicated mobilisation: six vendors each price site access, lifting equipment, and supervision that a single integrator would have shared.
- Variation orders without a common change-control: each vendor raises VOs independently and the buyer has no consolidated view.
- Buyer-absorbed coordination: a project manager on the buyer's side becomes the de facto integrator without the contractual authority of one.
- Warranty fragmentation: a fault at the cable-to-switch boundary becomes a multi-month finger-pointing exercise no single vendor owns.
Section 2
What an enterprise systems integrator actually does
The integrator's contractual value is not buying-power on hardware — it is owning the seams. A competent integrator takes single-source responsibility for design, procurement, delivery, commissioning, and warranty across the disciplines in scope. Their margin pays for design coordination, integrated programme management, single warranty interface, and an in-country support organisation that survives the project closeout.
- Integrated design across disciplines (cabling, active network, security, surveillance, power) before any BoQ is finalised.
- Single programme with one critical path, not six independent ones competing for the same site access windows.
- One commissioning authority and one acceptance protocol covering the whole stack.
- Single warranty interface — the buyer raises one ticket regardless of which sub-system is implicated.
- Documented as-built package, asset register, and operational runbook handed over as a defined deliverable.
Section 3
How to evaluate an integrator on something other than price
If price is the only evaluation criterion, multi-vendor will always look better and the integrator model never gets a fair hearing. A defensible evaluation weights the integrator's track record on comparable scopes, the depth of their in-country engineering bench, their design and documentation discipline, and the quality of their warranty and support organisation — not just unit prices on the BoQ.
- Reference projects of comparable scale, in the last 24 months, with named client contacts the integrator will let you call.
- Bench depth: how many certified engineers per discipline, employed (not subcontracted) by the integrator in-country.
- Design output: ask to see a real as-built package from a previous closeout — not a sales brochure.
- Warranty SLA: response time, on-site time, escalation path, and what is included vs chargeable.
- Financial standing: a project that runs 12–18 months requires an integrator that will still exist at the warranty handover.
Section 4
When multi-vendor is the right answer
Multi-vendor delivery is not always wrong. For small scopes, mature buyers with their own in-house engineering organisation, or commodity refreshes within a single discipline, the coordination overhead of an integrator is not justified. The model becomes risky when the scope spans disciplines and the buyer does not have an internal team able to own the seams.
- Small single-discipline scopes (e.g. a switch refresh in an existing rack room) — multi-vendor is fine.
- Buyers with a mature in-house engineering function that has owned similar integrations before — multi-vendor is workable.
- Multi-discipline new-builds, campus deployments, or compliance-driven scopes — single integrator is almost always cheaper at closeout.
- Anything where downtime carries a real operational cost — single warranty interface pays for itself on the first incident.
Conclusion
The strategic question is not "integrator vs multi-vendor" in the abstract — it is who owns the seams between the disciplines. If the buyer has the internal capacity to own them, multi-vendor can work. If they do not, the integrator's margin is buying the most expensive line item on the project: the absence of disputes.
Evaluating an integrator for a current scope? We can share the evaluation rubric we use with institutional buyers, including the reference-call script and the as-built deliverable checklist. Get in touch via the contact page.
