The latest room plan is not automatically the agreed plan

Two room plans can both look finished. Each can carry a recent date, a familiar project number and details entered by someone who understands part of the work. The difficulty appears when the plans contain different changes and the system gives one of them authority simply because it arrived last. That rule is convenient for finding a current file. It is less useful for establishing which set of decisions the people involved have actually agreed to use.

At Nerqali, a showroom colleague could revise a doorway detail while a survey specialist corrected a dimension elsewhere. The later upload might not include the earlier change. A reader opening only the current file would have little reason to know that another relevant decision existed. I had seen revision problems in survey applications earlier in my career, yet our retail workflow still allowed a date to stand in for a considered decision. The lesson returned in a setting where the apparent simplicity of the interface made the omission easy to overlook.

## Chronology is not consent

Our 2023 review found competing plans treated as approved in twenty-two of 119 sampled projects. That was the finding, not a count of twenty-two failed installations. Some differences were caught through conversation before work began. Others required clarification that the system had not prompted. I wanted the report to preserve that distinction because overstating the consequence would make the case more dramatic but less useful. We had evidence of an ambiguous instruction path. That was sufficient reason to examine the process without inventing a larger loss.

The supplier proposed selecting the most recently uploaded file while retaining the older version in history. It was a reasonable answer to a narrower question: how should a user find the newest file? It did not answer whether the author of that file had seen and intended to supersede the other change. A timestamp cannot establish that knowledge. Keeping an earlier version somewhere in the system also does not ensure that the person about to use the current plan knows there is a relevant disagreement to examine.

I did not want to solve the problem by declaring that every revision required a large approval meeting. Most changes could follow an ordinary, proportionate path. The consequential distinction was whether two versions carried changes that had not been reconciled. The system needed to expose that condition and connect it with the person authorised to resolve it. If the matter required a new measurement or a customer conversation, the workflow should make that work visible rather than imply that a technical comparison of files had supplied the answer.

We approved ₹76 lakh capital and ₹19 lakh implementation expense for the revised application behaviour and associated work. Finance owned the accounting treatment. The distinction mattered in the investment account because a capital amount and a one-off expense are not automatically the same as a recurring operating commitment. I also wanted the proposal to describe the review effort it would create. A system that exposes disagreement may increase the visible workload before it improves the quality of the instruction. Concealing that consequence would make the recommendation incomplete.

## Expose the branch before work follows it

The revised workflow retained the relevant versions and showed where a branch required reconciliation. An identified project specialist then established the issued plan. The software did not automatically merge room geometry, certify the measurements or infer customer consent. Those are different responsibilities from storing and comparing documents. IT made the state intelligible and prevented an unresolved branch from quietly becoming the apparent final instruction. The qualified professional remained responsible for deciding what the plan should say and what further evidence was needed.

Some colleagues worried that visible exceptions would slow projects and make the service appear less capable. Their concern deserved attention. A branch that remains unresolved for days can be an operating problem, and the company needs a practical response standard. But returning to an automatic last-upload rule would not eliminate the disagreement. It would move the consequence to whoever encountered the missing detail later. I preferred to account for the review while there was still an opportunity to clarify the instruction before a team organised work around it.

The design also needed to help people understand what had been resolved. It was not enough to add another approval label if nobody could see which changes the reviewer had considered. We wanted the issued plan to have a clear relationship with the relevant revisions and the decision made between them. That did not require displaying every historical file to every user. It required keeping sufficient context for the people who needed to review the choice, while giving the installation team an unambiguous account of the version they should follow.

I found it useful to separate three questions in the review. Which file is newest? Which changes are still unresolved? Which plan is authorised for the next activity? A single current-status field had been carrying all three meanings. Once we separated them, the application could support each question more honestly. This was not complexity added for its own sake. It was complexity already present in the work, represented clearly enough that people did not have to rediscover it through calls, memory or a disagreement at the site.

## Count reconciliation without calling it failure

In the six-month review ending September 2024, nine of 150 projects required explicit reconciliation before issue. I did not present that as a directly comparable reduction from the earlier twenty-two of 119. The populations and review conditions differed. Nor did I describe the nine as nine bad installations. They were nine occasions when the process identified a decision that needed attention before an instruction was issued. The useful result was visibility and resolution, supported by evidence about how the workflow behaved in those cases.

That distinction affects the incentives around a system. If every exposed exception is treated as a performance failure, teams learn to prefer processes that report fewer exceptions, even when the underlying work is no clearer. If every exception is celebrated as a success, the organisation may tolerate a needlessly difficult process. Leadership needs a more discriminating account: which disagreements are inherent in the work, which are avoidable, how quickly are they resolved and what does the resolution establish for the next person? The count alone cannot answer those questions.

For a board, I would translate this into questions about the meaning of reported status rather than a request to examine room-plan software. When management says an instruction is current, what does current establish? Where different professional contributions meet, who resolves a conflict between them? Does a low exception count reflect agreement or merely a process that does not expose disagreement? These are questions about the reliability of the operating account. They leave management and its specialists responsible for designing and running the practical workflow.

My early survey work taught me to respect the difference between a measured space and its representation. The revision project added a related lesson: a representation can preserve the latest information while losing an earlier agreement. A useful system does more than organise files. It helps people recognise when a decision is still required and prevents that absence from being mistaken for certainty. The aim is not to make every project slower or every record more elaborate. It is to ensure that the instruction carrying work forward has the authority that its presentation appears to promise.