Asset audit: how to run a physical inventory count

An asset audit compares what is physically there with what the register says should be there. Done properly it confirms identity, location, custodian, condition, and status — but only when the team records discrepancies instead of editing the source until it matches the room.

This guide addresses the operational asset audit: the physical count and the reconciliation that follows it. It does not define financial-audit sampling, accounting policy, security certification, regulatory evidence, or legal requirements. Coordinate those responsibilities with the appropriate finance, security, compliance, or external-audit owner.

Define the objective and scope

State why the count is being performed. Different objectives require different observations:

• verifying high-value or frequently moved equipment;
• preparing for an office move or program closeout;
• checking a category after repeated custody exceptions;
• confirming equipment assigned to departed employees;
• validating a new register or migration;
• supporting an authorized accounting or insurance review.

Write the scope in concrete terms:

• included categories;
• locations and sublocations;
• expected custodians;
• included lifecycle states;
• cutoff date and time;
• exclusions such as items in transit, under repair, or held by an approved third party;
• observation fields;
• required evidence and approval.

Avoid “count everything” unless the organization has defined what everything means and can control changes during the count.

Assign separate roles

The same person may perform several roles in a small team, but the responsibilities should remain explicit.

When possible, do not let the person responsible for a record silently certify and correct the same exception.

Prepare the count list

Create the expected population from the authoritative register at the cutoff time. Include enough information to identify the item without revealing the answer the observer is meant to verify.

Depending on the objective, the count list may include:

• asset ID or scannable code;
• asset name and category;
• manufacturer, model, or serial fragment;
• expected location or custodian;
• expected status;
• last verification date;
• notes about approved exclusions.

Decide whether the count is book-to-floor, floor-to-book, or both:

• Book-to-floor begins with register records and asks whether each expected asset can be observed.
• Floor-to-book begins with physical items and asks whether each observed asset has a valid record.

Using both directions helps reveal missing assets and unrecorded assets.

Freeze or control changes

Assignments, receipts, repairs, and disposals may continue while the count is happening. Define how those events are handled:

1. freeze ordinary changes for a short approved window; or
2. keep a transaction log that the reconciler applies after the cutoff; or
3. divide the count into locations or categories with separate cutoff times.

Without a cutoff rule, the team may classify a legitimate movement as a discrepancy or use a later update to rewrite what existed at the observation time.

Copyable physical count template

Use one row per observation:

Record what is observed. Do not copy the expected location or condition into the observation column without confirming it.

Count procedure

Before entering the location

• Confirm the scope and cutoff.
• Confirm access with the location owner.
• Test scanning devices and backup lookup method.
• Define how sealed, unsafe, inaccessible, or operating equipment is handled.
• Remind observers not to move or relabel items without approval.
• Confirm how photos or other evidence may be collected under policy.

For each item

• Scan or read the stable asset ID.
• Confirm the physical item matches the record identity.
• Record the observed location.
• Record the custodian only when the procedure permits and the fact is verified.
• Record observed condition using approved values.
• Record whether the label is scannable, damaged, or missing.
• Record the observation time and observer.
• Create a discrepancy rather than correcting the expected value during the count.

Before leaving the location

• Review uncounted expected assets.
• Review observed items with no matching record.
• Confirm all temporary notes have a count ID.
• Record inaccessible areas or items.
• Obtain location-owner acknowledgment when the procedure requires it.

Discrepancy log template

Every discrepancy needs a status, owner, and next action. “Not found” is an observation, not a final conclusion.

Investigate before correcting

Common explanations include:

• a legitimate transfer occurred after the cutoff;
• the assignment was never updated;
• the item is under repair, in transit, or in approved storage;
• the label is missing or attached to the wrong item;
• a duplicate record exists;
• the item was disposed of without the final event reaching the register;
• the observed item was never recorded;
• the expected custodian or location uses a different approved naming convention;
• the item is genuinely missing and requires the organization's escalation process.

Retain the observation and investigation even after correcting the current record. The correction should explain what changed, who approved it, and when.

Reconcile the count

At minimum, reconcile:

• expected assets;
• observed expected assets;
• expected assets not observed;
• observed assets without a valid expected record;
• location and custodian mismatches;
• status and condition mismatches;
• missing, damaged, duplicate, or incorrect labels;
• approved exclusions and changes after cutoff;
• open discrepancies and their owners.

Counts alone do not explain the quality of the register. Report the population, scope, dates, methods, unresolved exceptions, and limitations alongside any completion percentage.

Apply and review corrections

Corrections should follow the normal asset workflow:

• transfers close an earlier assignment and create a new one;
• condition changes record the observed state and date;
• missing labels create a replacement action;
• disposals require the approved lifecycle event and evidence;
• duplicate records are resolved according to the migration or data-governance rule;
• missing assets follow the organization's escalation and decision process.

Schedule a follow-up review for open exceptions. Do not call the count complete merely because the observation period ended.

Using scanning and AssetCenter

The barcode asset-tracking guide explains how scans identify records and how to plan exceptions, and AssetCenter's barcode asset tracking software shows what a count looks like when each scan opens the record it belongs to. The asset-tag guide covers durable identifiers and placement.

In AssetCenter, scanning can open the corresponding asset record, and the record can retain assignments, locations, events, condition updates, files, and notes. Use the barcode-scanning manual, managing-assets manual, and asset events manual for current product instructions.

A count of this kind is only as useful as the register it is checked against. AssetCenter's fixed asset management software holds that register and the verification history behind it, so each observation is recorded against the same asset rather than in a separate audit file.

Next step: pilot one location

Choose one bounded location with representative categories and at least one known exception. Freeze or log changes, run both book-to-floor and floor-to-book observations, and complete the discrepancy approval cycle.

Expand only after the team can show how each observation becomes an approved correction or an owned unresolved exception without erasing the evidence behind it. A count that keeps finding the same discrepancies is usually reporting a process gap rather than a counting problem; simple asset management covers the ownership and update rules that stop the register drifting between audits.

By Jeremy Francis

Founder & CEO, AssetCenter