Obligation architecture and pre-award governance · Lesson 5

Build the baseline obligation-control-indicator register

7 min read

Lesson objectives

  • Make a documented decision on build the baseline obligation-control-indicator register with evidence and limits.
Reading tools

Once outcomes, obligations, dependencies and risks have been identified, the challenge is to unify them in a register people can operate. A list of clause names sends the team back through dozens of pages during each incident. A copy of every contract detail becomes hard to maintain and may expose sensitive material to people who do not need it. The baseline is an approved version of the elements that drive daily management and decisions, linked to the signed text and its version record, and revisable after authorised change. It is an operating map of the agreement, not a replacement for it.

Build each row around a specific decision. The minimum fields for this program are obligation identifier, lifecycle stage, intended outcome, location in a contract document or label as an internal control, actor, due point, performance evidence, verification method, monitoring indicator, non-performance risk, escalation trigger, decision holder, and confidence in the data. If the signed words lack an acceptance criterion, write “unresolved” and link it to an interpretation or change task. A guessed criterion only makes the spreadsheet look complete. Source location matters because a phrase may differ across a draft, an annex and the executed version. A user must know which version governed a past decision.

Do not turn the register into a set of traffic lights. Consider a visit report delivered on time but never checked against a sample of site access records. The supplier’s submission may be complete while the buyer’s verification control has not operated. A single green status erases that difference. Track performance of the act separately from verification of its effect. Where verification uses sampling, state the sample size and selection rule. If a control is meant to detect exceptions, allowing the supplier alone to choose the records for inspection would weaken it. Equally, low-risk routine work need not always receive a full audit.

A useful indicator states numerator, denominator, period, data source and exception rule. “95% service timeliness” says too little: 95% of tickets, operating minutes, or sites? If 19 of 20 tickets met a deadline, the count-based rate is 95%. If the one late ticket belonged to a critical site, however, its consequence may outweigh the others. Add a risk-class distribution or a critical-event view alongside the average. Do not add dozens of metrics with no associated decision. Ask what action would change if an indicator moved. If there is no answer, question its place in the register.

The baseline needs change governance. When an authorised variation changes scope, a deadline or price, do not silently overwrite the field. Retain the earlier version, change identifier, effective date, approving person, and dependent rows affected. Performance before the effective date may still use the earlier measure; merging old and new definitions into one trend can manufacture improvement. If a subcontractor or evidence method changes without changing the underlying duty, update the control or evidence source and explain comparability. Every operational edit needs a reason and version, but not every dashboard edit is a contract variation.

Include “open questions” in procurement-to-management handover rather than claiming false completeness. Who certifies the operating report? Where are site records held? Who may instruct the supplier? Does exit require data transfer? When must an extension decision be initiated? If an answer is missing at award, give it an owner and resolution date before it becomes an incident. In its UK government context, GovS 008 calls for documenting party obligations and a management plan covering roles, risks, contingency and exit, with proportionate performance monitoring. We use these topics to test the completeness of our register, not to impose a universal official template.

Lesson summary

The expanded exercise yields three separate actions: correct the asset identifier after proof, verify the visit against evidence for the right pump, and send the effect of late reporting on payment to the holder of contract interpretation. Test the new control at the first post-correction visit: all three identifiers match and no ticket remains assigned to an unknown asset. If the invoice remains unresolved, keep an explained pending state; a corrected asset number must not turn the whole file green automatically.