This manual is built in three layers. Read the training sections once to understand the reasoning. Print the worksheets and put them where the work happens. Then follow the 30-Day Implementation Plan on page 44 rather than trying to launch everything at once.

An audit is a scheduled moment where a specific person is required to look at a specific thing and write down what they found. That is the whole definition. It is not an inspection, not a punishment, and not an accusation.

The three functions of an audit

1. Detection. It surfaces variances while they are still traceable to a date, a shift, and a person.
2. Deterrence. Opportunity is the one leg of the fraud triangle you fully control. Unpredictable, visible auditing removes it.
3. Documentation. It creates the record that turns isolated events into a pattern — and protects you legally when you eventually act.

Why owners discover losses late

• Inventory variance is only measured once a year, so the window is 365 days wide.
• Cash is reconciled as a store total, hiding offsetting register variances.
• Receiving is signed rather than counted, so vendor shortages become your shrink.
• Damage and waste are discarded rather than logged, so they resurface as "theft."
• Exception data (refunds, voids, discounts) exists in the POS but is never read.

The opening routine determines whether any later investigation is possible. Every control that follows depends on a clean, attributable starting point.

The opening sequence

1. Exterior walk before entry — doors, glass, locks, signs of forced entry.
2. Alarm log reviewed for overnight activity; anomalies documented, not dismissed.
3. Safe count verified against the prior night's closing figure by two people.
4. Starting banks counted in front of the receiving cashier; both initial the sheet.
5. One cashier per drawer, one login per cashier, recorded on the daily sheet.
6. Back door, compactor area, and stockroom checked before the first delivery.

Non-negotiables

• Drops at a published threshold — most small formats set $200–$300 per drawer.
• No cash counted on the sales floor or in camera-blind areas.
• Every no-sale requires a reason; the POS should force one.
• Manager overrides require the manager's own credentials, never a shared code.
• Deposits prepared by one person and verified by a second before leaving the building.

Receiving is where inventory becomes real. A shortage accepted here is untraceable later — it will appear as shrink months after the vendor's claim window has closed.

Blind receiving in five steps

1. Receiver counts without the expected quantity visible.
2. Count is compared to invoice by a second person or by the system.
3. Discrepancies noted on the invoice with the driver present and signing.
4. Photos taken of damage, reseals, short pallets, or substituted items.
5. Credit request submitted the same day; tracked until resolved.

Ten to twenty SKUs typically account for the majority of theft exposure in a small format. You cannot count the store daily. You can count these.

Counting rules that keep the data honest

• Rotate which SKUs are counted; never publish the rotation.
• Expected quantity hidden from the counter until after the count is recorded.
• Count at varied times — a 4 p.m. count and an 8 a.m. count tell different stories.
• Record zeroes. A "no variance" entry is data, not wasted effort.

Log everything, especially the small things

Damage, spoilage, shoplifting events (including those where nothing was recovered), safety issues, equipment failures, customer complaints tied to transactions, and any refusal at the register. Each entry needs date, time, location, people involved, dollar value, and a factual narrative.

The ten-minute walkthrough route

1. Front end — drawer security, register area clutter, bag check discipline.
2. Sales floor blind corners and any product concentration off-planogram.
3. Fitting rooms / restrooms — packaging, tags, empty containers.
4. Back hallway, stockroom staging, and anything sitting near an exit.
5. Compactor, dumpster, and trash exit — the most-used internal theft route.
6. Back door status and receiving area order.

1. Each drawer reconciled independently against its own sales.
2. Variance recorded by register and by cashier — never blended into a store figure.
3. Exceptions from the day (voids, no-sales, refunds) printed and attached.
4. Deposit prepared, verified by a second person, and logged.
5. Safe counted and signed; figure carried to tomorrow's opening verification.
6. Closing checklist completed and initialed — alarm, doors, lights, cameras recording.

The random mid-shift till audit is the single highest-yield control available to a small retailer. It is fast, it is unpredictable, and it produces attributable data.

Procedure

1. Select the register and time without announcing it. Vary cashier and daypart.
2. Note the exact time; pull the sales total for that drawer to that moment.
3. Remove the drawer to a secure counting area — never count on the floor.
4. Count twice. Record the count before comparing to expected.
5. Reconcile, record the variance, and return the drawer promptly.
6. Discuss the result with the cashier the same day, whatever the outcome.

Every POS produces exception data. Almost no independent retailer reads it. Ten minutes a week here finds more than a hundred hours of video review.

A cycle count is a full count of one department or category, performed on a rotation so the entire store is covered each quarter without closing.

1. Pick the department at the start of the week; count on a varied day.
2. Count blind — no expected quantities on the sheet.
3. Recount every variance greater than your threshold before recording it.
4. Check for receiving errors, misplaced product, and unit-of-measure issues first.
5. Record the result even when clean; the clean weeks define your baseline.

Weekly receiving reconciliation

• Match every invoice to a receiving document and to what entered inventory.
• List open credits by vendor and age; anything over 30 days gets escalated.
• Note which employee received which vendor — rotate assignments monthly.

Adjustment audit

Pull every manual inventory adjustment for the week. For each: who, why, how much, and who approved it. Adjustments made by the same person who counts, receives, and sells the product represent a complete absence of segregation of duties.

Fifteen-minute targeted review

Do not watch hours of footage. Pick three moments from the week's data: the largest refund, one late close, and one large delivery. Review those specific timestamps. Data picks the moment; video confirms it.

Track failures, not just fixes. The finding is never "someone propped the door." The finding is that your workflow requires the door to be propped.

The most revealing number is the known-to-unknown ratio. A store that logs damage, waste, and markdowns well should be able to explain a meaningful share of its shrink. When unknown shrink is 90% of the total, the problem is documentation before it is theft.

Rank departments two ways — by shrink dollars and by shrink percentage. The lists will differ, and both matter. Dollars tell you what to work on. Percentage tells you what is broken.

Three questions for your top two departments

1. Is it losing product? Look at receiving and theft controls.
2. Is it losing margin? Look at pricing, markdowns, and discounts.
3. Is it losing accuracy? Look at counting method and system entry.

Vendors correct for the customers who count. Three months of documented discrepancy data changes both the conversation and the behavior — and the discrepancies that remain get credited instead of absorbed.

Map every monthly finding to one of three outcomes:

Training record minimums

• Who was trained, on what, by whom, on what date.
• Acknowledgement signature for cash handling, refunds, and receiving.
• Refresher cadence — at minimum annually and after any related incident.

Every monthly audit ends with no more than three corrective actions. Three completed beat fifteen listed.

Keep five numbers on one running sheet, month after month:

1. Shrink % (and shrink dollars)
2. Total cash variance (absolute value, not net)
3. Receiving discrepancy rate
4. Exception outlier count
5. Open corrective actions

Review the line, not the dot. A store improving from 3.1% to 2.4% is being managed well even if the target is 1.5%. A store holding steady at 1.8% for six months has a stable system and can now afford to expand its program.

1. Verify the data. Recount, confirm balances, rule out timing and posting errors.
2. Identify access. Who could have — from schedules, logins, keys, and override rights.
3. Establish the timeline. Last verified present; first confirmed missing.
4. Preserve evidence. Export video immediately; save transaction detail and originals.
5. Review transactions. Full history for the window, not just the flagged items.
6. Review video against data. Match timestamps to specific transactions.
7. Interview last. Open questions, two people present, unlocked door, no accusations.
8. Document objectively. Facts, dates, amounts, sources. No adjectives.
9. Take corrective action. Always paired with the control change.
10. Follow up. Re-audit the same control at 14 and 30 days.

Before

• Facts assembled and written down; you should not be learning during the interview.
• Two management representatives present; door unlocked; reasonable length.
• Know your state's requirements and your own policy before you begin.

During

• Open questions first: "Walk me through how you handle a no-receipt return."
• Let silence work. Do not fill it.
• Never accuse, never threaten, never promise leniency or outcomes.
• Document exact language; avoid paraphrasing anything that matters.

After

• Written summary the same day, signed and dated by both managers.
• Decision made on facts and policy — consistently applied across every employee.
• Control change implemented regardless of the outcome.

Trigger. Weekly exception review showed one cashier with 41% of no-receipt refunds while working 18% of transactions.

Step 1 — Verify. Report rerun for 90 days. Pattern held across all three months.

Step 2 — Access. Refunds processed under her login only; no shared credentials in use.

Step 3 — Timeline. Pattern began seven weeks after a manager departure removed the second-approval requirement.

Step 4 — Preserve. Nine days of video remained. Exported the same afternoon.

Step 5 — Transactions. 14 refunds averaging $22 with no matching original sale that day.

Step 6 — Video. Refunds processed with no customer present at the counter.

Step 7 — Interview. Facts presented neutrally; admission obtained without accusation.

Outcome. $312 documented over 90 days; restitution agreed; termination consistent with policy.

Most audit findings end in coaching, not investigation. Coaching done well prevents the escalation that leads to real cases.

1. Poor training — the employee never learned the correct method.
2. Poor procedures — the correct method does not exist or cannot be followed during a rush.
3. Vendor shortages — you paid for product that never arrived.
4. Administrative mistakes — price files, markdowns, unit-of-measure, duplicate receiving.
5. Employee theft — internal, usually escalating, usually attributable.
6. External theft — concentrated on concealable, resellable product.

Work them in this order. The first four are more common and cheaper to fix, and ruling them out honestly makes the remaining case dramatically stronger.

• Consistency beats intensity. Ten minutes, 300 times a year.
• Documentation is the product. An unrecorded audit did not happen.
• Accountability runs both ways. Managers audit staff; owners verify managers.
• Trends, not snapshots. Six months of one metric beats one perfect month.
• Data must change decisions. Schedules, layouts, vendors, policies, thresholds.
• Improve the program twice a year. Retire checks that never find anything.

Full separation is impossible with four employees. Partial separation is not, and partial is enough to remove most opportunity.

• Identical forms at every site — variation destroys comparability.
• Same submission deadline weekly; late is a finding of its own.
• Owner reviews a ranked comparison, not individual site reports.
• Rotate an unannounced site visit monthly; verify audits were actually performed.
• Share what the best site does differently — most gaps are practice, not people.

How long should the daily audit take?

15–25 minutes total, split across open, midday, and close. If it takes an hour, it will be abandoned.

Should I tell employees about audits?

Announce the program, never the schedule. Known-but-unpredictable is what removes opportunity.

What is an acceptable shrink percentage?

Benchmarks vary widely by format. Your direction over six months matters more than any published figure.

What if I find a small shortage repeatedly?

Coach first, verify with a re-audit inside seven days, and escalate only when the pattern survives retraining.

Can I review video without cause?

Reviewing your own business footage is generally permissible, but audio recording and certain locations are restricted. Know your state law and your posted policy.

Do I need a POS with exception reporting?

It helps enormously, but a manual tally of refunds and voids by cashier for two weeks will surface the same outliers.

What if my manager resists auditing?

Audit the manager's controls first and frame it as protection. Resistance to documentation is itself a finding worth understanding.

How do I audit when I am not on site?

Require same-day submission of completed forms, verify a random sample in person monthly, and never accept a week of sheets completed at once.

A large retailer facing a shrink problem opens a case, assigns an analyst, pulls exception data, dispatches an investigator, and implements a corrective action tracked at the district level. An independent retailer facing the same problem has a bad feeling, a spreadsheet, and no time.

I spent years on the enterprise side of that comparison, and then years talking to owners on the other side of it. The gap was never about intelligence or effort. Independent owners understand their business more deeply than any corporate analyst ever will. What they lacked was infrastructure — the boring, essential machinery of scheduled audits, connected records, exception reporting, and follow-up that makes loss visible.

Professional loss prevention tools existed. They cost more than most independent stores could justify, required specialists to operate, and were built around organizational structures that do not exist in a four-employee business.

My LP Portal exists to remove that barrier: professional-grade audits and checklists, incident management, high-risk merchandise tracking, receiving intelligence, loss heat maps, structured training through LP Academy, and AI guidance — in one affordable platform designed for the owner who is also the buyer, the scheduler, and the closer.

You worked too hard building this to lose it a hundred dollars at a time.