In this guide
How a Verification Is Planned and Run
A verification runs in five stages: plan, sample, verify, reconcile, report. Planning fixes the cut-off date, the locations in scope and the register extract everything will be measured against, and that extract is frozen before anyone travels. Sampling decides which lines are visited, and because asset registers are heavily skewed by value, selection is weighted so the sample covers a stated proportion of carrying value rather than a stated proportion of lines. Verification is the floor work, running in both directions so that missing assets and unrecorded assets both surface. Reconciliation matches what was seen against the frozen extract and resolves what it can before anything is reported as a difference. Reporting states the outcome and its accounting consequence. Effort is driven by three things and only three: the number of locations, the number of register lines, and whether identification already exists on the assets.
Planning: Scope, Cycle and Coverage
Planning settles three things, and settling them in writing is what distinguishes a programme from an annual scramble. Full count against rotation is the first decision. A modest estate at one or two locations can be covered completely each cycle and generally should be, since the saving from sampling is small and the assurance from completeness is large. A large multi-site estate is covered on rotation, with the cycle defined so that every class and every location is reached within a stated period rather than whenever convenient. Sites and classes in scope follow, and the exclusions are recorded with their reasons. Assets below a value threshold, assets at locations being closed, and asset classes verified through another process are all legitimate exclusions, but they have to be visible in the plan rather than discovered as gaps afterwards. Cut-off and freeze arrangements close the planning. The register extract is frozen at the cut-off before anybody travels, because a verification against a moving register cannot be reconciled. Where assets will move during the fieldwork, the treatment is agreed in advance: either movement is suspended, or movements are logged and reconciled at the boundary.
Sampling: How Much Is Enough
Selection is weighted by value, because asset registers are almost always skewed and a sample drawn evenly across lines will spend most of its effort on items that carry very little of the money. In practice that means lines above a threshold are examined completely rather than sampled at all, and the remainder is stratified so that each band is represented. A sample built this way can reach a substantial majority of carrying value while touching a modest proportion of lines, which is the whole reason the approach is used. Population size matters far less than people expect, and this is the point most misunderstood. Beyond a certain size, the number of items needed to support a conclusion at a given level of confidence changes very little as the population grows, so a register of twenty thousand lines does not need ten times the sample of one with two thousand. Effort scales with value concentration and with the number of locations, not with line count. Full coverage is the only defensible answer in three situations: a small population where sampling saves little, a balance already contested, and any case where a single material line would otherwise go unverified.
Verification on the Floor
The floor work runs in both directions and a verification that runs only one is incomplete by construction. Tag-to-register takes an asset physically present and traces it back to its line, which is the direction that finds equipment nobody capitalised. Register-to-tag takes a line and looks for the asset, which is the direction that finds ghosts. Doing only the second, which is the instinctive approach because the register is the document in hand, guarantees that unrecorded assets are never discovered. Condition and usage are captured at the same time, against a defined scale rather than in free text, because these are the observations that cannot be made later from a desk and they are what triggers the impairment question. Whether an asset is actually running matters as much as whether it is intact. Assets in transit or off site need a rule agreed before the walk rather than a decision made in front of each one. Equipment at a repairer, out on hire, at an exhibition or with an employee is verified by documentation rather than by sight, and it is recorded as a separate category so the reader can see how much of the conclusion rests on paper rather than observation.
Reconciliation and Evidence
Reconciliation is where a verification either becomes useful or stops short. It starts by building the exception list: every register line not matched to an asset, and every asset not matched to a register line, held as two separate populations because they have opposite accounting consequences. Merging them into a single net figure conceals both. Each exception is then worked individually, and most resolve. An asset recorded at one site is found at another because it was moved and the register was never updated. A line appearing unmatched is matched once a duplicate description is unpicked. An asset with no line turns out to have been capitalised under a different description. Each resolution is supported by whatever document explains it, so the reconciliation stands up without the person who performed it having to narrate it. What remains after that work is the residue: items nobody can account for in either direction. The residue is quantified at carrying value, reported as its own figure rather than absorbed into an adjustment, and a management representation is obtained on it. A residue reported honestly is a finding; a residue quietly netted off is a misstatement.
Doing It Yourself, and When Not To
An internal team can credibly do a great deal of this. It can walk the estate, confirm existence, correct locations and custodians, identify unreadable tags, and build the exception list, and doing so before an external verification is almost always worth the effort. What an internal team cannot supply is independence, and independence is required wherever the result will be relied on by somebody outside the business. A lender assessing security, a statutory auditor gathering evidence on a material balance, and an insurer settling a claim each need the count performed by somebody with no interest in its outcome, and a self-count does not become independent because it was thorough. The register is too far gone for a self-count when descriptions are duplicated to the point that lines cannot be matched, when the register does not tie to the ledger, when several years have passed since the last verification, or when a previous internal count produced a result management did not accept. At that point asset verification by an external team resolves in one exercise what internal attempts have not.
