A multi-site maintenance RFP often begins with a spreadsheet of addresses and a request for comprehensive support. Providers then make different assumptions about assets, trades, frequencies, operating windows, parts, travel and specialist work. The resulting prices appear comparable but describe different services. Ambiguity may feel convenient during sourcing, yet it reappears as exclusions, approvals and unresolved work after mobilisation.

The purpose of an RFP is not to predict every fault. It is to make the operating boundaries, estate data and decision rules clear enough for bidders to design a credible response. Buyers can allow discovery where information is incomplete, as long as they label the gaps and explain how the final baseline will be agreed. The following structure helps procurement, facilities, finance and operations evaluate the same service rather than separate interpretations of it.

Start with an estate baseline and a data-quality statement

Provide a location register with address, format, operating hours, authorised contacts and known access conditions. If sites differ significantly, identify meaningful cohorts such as office, store, branch, restaurant or warehouse rather than asking bidders to infer them. Add the responsible landlord, internal team or incumbent vendor where that boundary affects the proposed work.

Asset data may be complete, partial or unavailable. Say which is true. Where an asset register exists, include category, identifier, location, basic specification, maintenance requirement and known status. Where only trade-based coverage is possible, describe the spaces and likely tasks and ask bidders to propose a validation method. Require a controlled process for correcting the baseline so new discoveries do not disappear into email.

  • Locations, formats, operating windows and access restrictions
  • Assets or trade categories, quantities and known condition
  • Current planned schedules, open backlog and material dependencies
  • Landlord, OEM, statutory and incumbent-vendor boundaries
  • Data gaps to be validated during mobilisation

Describe preventive and reactive work separately

Preventive maintenance begins with an approved schedule. For each asset or trade, state the task, frequency, checklist, skill boundary, evidence and action when a finding needs follow-up. Avoid using preventive maintenance as a label for a visit with no defined work. Bidders should explain how schedules are released, missed access is handled and findings become controlled corrective work.

Reactive maintenance begins with an issue. Define intake channels, required information, priority rules, triage responsibility and the difference between diagnosis, temporary restoration and resolution. A priority should reflect business impact and escalation, not just a colour or word. Ask how the provider will distinguish a job waiting on field action from one waiting on access, approval, parts, an OEM or the buyer.

  • Planned task and evidence by asset or trade
  • Reactive job categories and minimum intake fields
  • Priority criteria and buyer escalation for material impact
  • Explicit specialist and statutory exclusions
  • Definitions for complete, open, paused, repeat and cancelled work

Make approvals, parts and commercials testable

Many maintenance delays are commercial decisions disguised as technical delays. The RFP should identify who may diagnose, estimate, approve and procure. Define whether common consumables are included, held by the buyer, carried by the provider or approved per job. Do the same for replaced material, defective returns and evidence of consumption. If thresholds vary by site or category, give bidders the rule set.

Ask for a pricing schedule that maps to the work model. Retainers, visit charges, planned-task rates, labour units, material mark-ups, travel conditions and management fees can all be legitimate, but every bidder should disclose the same inclusions and triggers. Include scenarios such as no access, an unapproved estimate, a follow-up with parts, work beyond scope and a remote location. Scenario pricing reveals interpretation gaps that a rate card can hide.

  • Base-scope inclusions and explicit exclusions
  • Estimate format, approval threshold and validity
  • Consumable, spare, transport, storage and return responsibilities
  • Commercial treatment of failed access, repeat and additional work
  • Change control for new sites, assets, trades and frequencies

Specify evidence and reporting before choosing a platform

Start with the decisions the record must support. A useful closure entry may include reported issue, diagnosis, action, asset identifier, materials, photographs, residual risk and site acknowledgement. Preventive work may require checklist readings or observations; reactive work may need a cause and restoration status. More fields are not automatically better. Each should have an owner, a use and an appropriate access rule.

Reporting should preserve operational states rather than compress everything into open and closed. Ask for views of scheduled work, accepted requests, jobs awaiting appointment, field allocation, access, approval, material, technical input and final review. Then define a compact governance pack for trends: demand, ageing, repeat work, evidence exceptions, recurring assets or sites, and named improvement actions. A live dashboard cannot compensate for undefined data.

Evaluate mobilisation as seriously as steady state

Ask each bidder to submit a mobilisation plan with buyer dependencies, validation steps, roles, decision dates and cohort gates. The plan should explain how site and asset data is checked, how contacts and access are confirmed, how field instructions are prepared, how open backlog is handled and how systems exchange status. Treat unknowns as work to close, not assumptions to bury in a timeline.

Use a representative pilot across different site types and work categories. Agree what evidence will support expansion and what conditions will pause it. Evaluate bidders on the clarity of their questions and exception thinking as well as their answers. A provider that identifies a missing responsibility boundary during the RFP is offering more operational value than one that accepts every requirement without qualification.

  • Named mobilisation team and buyer counterparts
  • Baseline-validation method and sign-off
  • Pilot sites, tasks, scenarios and exit criteria
  • Open-risk and dependency register
  • Cohort rollout and change-control gates

Common buyer questions

What if the buyer has no asset register?

State that limitation and ask for a priced discovery or validation approach. A trade-based interim scope may be possible, but the contract should explain how discovered assets and condition are recorded, reviewed and incorporated.

Should the RFP prescribe service levels?

It should describe business impact and required operating windows, then ask bidders for feasible commitments with triggers, pauses, location assumptions and exclusions. Final measures must be defined consistently before price comparison.

How many sites belong in a pilot?

Choose enough sites to represent meaningful differences in format, access, asset mix and demand without making investigation unmanageable. Scenario coverage matters more than a universal site count.