Grocery Shrink Tracking Software: Five Approaches, Compared

The margin math that makes grocery shrink urgent

Grocery shrink runs around 1.5% of sales. The National Retail Federation's National Retail Security Survey has put retail shrink at roughly 1.4% to 1.6% of sales for years, and grocery sits at the higher end because of perishables. That number sounds small until you set it against grocery's net margin, which usually lands between 1% and 3%.

Run the math on a $400M, 90-store chain. Shrink at 1.5% of sales is $6M a year. If your net margin is 2%, your net profit is $8M. Shrink is nearly as large as everything you take home. Cut it by a third and you add $2M to the bottom line, which is several new stores' worth of profit without the capex.

That is why this category gets bought. The question is which of five quite different things you are actually buying, because they go by the same name and catch different losses.

First, size your four buckets

No tool covers all of these well. Which approach is right depends entirely on where your loss actually sits, and most grocers have never measured that split.

• Theft, external and internal. Shoplifting and organized retail crime hit specific high-value categories: meat, baby formula, health and beauty, liquor. Internal theft shows up in voids, refunds, discounts, and cash handling at the register.
• Operational and process loss. Receiving errors, deliveries signed for but never fully checked in, markdowns that should have been taken and were not, mispriced items, scan errors at the lane. Quieter than theft and often larger.
• Spoilage and perishable waste. Produce, meat, dairy, deli and bakery that expire on the shelf. Grocery's perishable exposure makes this a big share of total shrink, and it is the bucket most other retail formats barely have.
• Unknown shrink. The gap between book and physical inventory you cannot assign to any of the above. Unknown shrink is not a cause. It is a measure of how much you failed to detect in time.

A store losing $80K a year to spoilage needs a different intervention than one bleeding $80K to register fraud. Do this sizing before you take a demo, or you will buy whichever tool demos best rather than the one aimed at your money.

The five approaches, compared

Every product sold as grocery shrink software is one of these five, or a bundle of two. The useful question is not which is best. It is which one sees the bucket you are losing to.

1. Shrink logging at the point of loss

Staff record the loss when it happens, from the floor. Scan the item, enter a quantity, pick a reason code. Inventory updates immediately and the adjustment flows into reporting, so margin reflects sellable product rather than book product.

Representative: Vori's shrink tracking is built this way. Its own product page describes mobile shrink entry where teams "scan an item, enter the quantity, and select a reason code," with standard reasons for damage, spoilage, theft and transfers plus custom ones, aimed at independent and multi-location grocers.

This is the only approach that produces clean, reason-coded data about why product was lost. Its limit is structural and worth being honest about: it captures exactly what someone scans. Product that walks out the door, gets binned without a scan, or was never received correctly leaves no record.

2. Physical count and self-scan audit

Periodic counting to establish the book-versus-actual variance. This is the only method that measures total shrink, because it is the only one that compares what you think you have to what is physically there.

Representative: Datascan provides self-scan physical inventory technology for grocery, with counts run by store teams, by Datascan's own staff, or as a hybrid. Its material focuses on perishable departments and cites department shrink rates of 4 to 8% in produce, 3 to 5% in meat and 2 to 4% in dairy against 1 to 2% in non-perishables.

The limit is cadence, not accuracy. Most grocers run a full physical once or twice a year. When a March count shows a $40K hole in the meat department, the cashier who was voiding sales left in January, the receiving pattern started in November, and the cooler that ran warm got fixed in February by a tech who never logged why. You are doing forensics on months-old evidence, and you cannot interview a number.

3. POS exception reporting

Rules over the transaction log: voids, no-sales, refunds, manual discounts and price overrides, grouped by cashier, register and shift. Internal theft has a signature and it lives here.

Most modern grocery POS platforms ship some version of this, which makes it the cheapest thing on this list to switch on. It is also narrow. It sees the register and nothing else, so it is blind to the operational and spoilage buckets that are usually larger than till fraud in grocery.

If you have never run exception reporting, do that before buying anything. It is often already paid for.

4. Video and sensor detection

Camera analytics, computer vision at self-checkout, weight sensors. Strong on external theft and on scan avoidance at self-checkout, which is a real and growing problem in grocery.

Two caveats. The cost is capital and recurring: hardware per store, network, and someone to tune the analytics and review the alerts. And coverage is literal. It sees what the camera sees, which means it contributes nothing to spoilage, receiving errors or missed markdowns.

RFID belongs in this family and rarely pencils out in grocery. You cannot economically tag a $1.99 can or a head of lettuce, and your highest-volume categories are exactly the ones it cannot cover. In apparel, where a tag sits on a $60 garment, the math is different.

5. Inference from POS and inventory data

No new hardware and no new data entry. The system reads the POS, ERP and inventory feeds you already generate and flags patterns that correlate with shrink: depletion that outpaces recorded sales, margin slipping in a category without a cost or promo reason, voids clustering on one register, perishables aging with depletion slowing against received volume.

The appeal for a multi-store grocer is that it covers all four buckets at once, runs every store every day, and deploys without capex. The honest limit: it tells you where and when, not why. It will tell you store 47's deli is losing product and it will not tell you whether that is a cooler, a cashier or a receiving door. Somebody still has to walk the floor.

Which one to start with

Match the approach to your largest bucket, not to the best demo.

• If your loss is mostly spoilage, start with point-of-loss logging. You need reason codes before you can fix anything, and no inference engine will tell you whether it was over-ordering or rotation.
• If you suspect till fraud, switch on POS exception reporting first. You may already own it.
• If your unknown shrink is large, which is the common case, inference is the only approach that shortens the loop between counts. Unknown shrink is by definition the loss nothing else caught.
• If you have a specific high-theft format or category, video earns its place in those stores, not chain-wide.
• If you do not know your split, run a count and a quarter of exception reporting before buying anything else. That is cheap and it tells you which of the other four to spend on.

Most chains that get this right end up running two: a counting cadence for truth and one continuous method for speed.

What to hold every vendor to

• Store-level and category-level granularity. A chain-wide shrink number is useless for action. You need to know it is store 47's deli, not "the company."
• Perishable awareness. Tools built for general retail treat everything like a durable SKU. The software has to understand expiry and depletion timing, not just theft signatures.
• Read-only integration. Pulling from POS, ERP and inventory without writing back means a faster security review and no risk to your transaction systems.
• Alert quality over volume. A tool that fires 200 flags a day trains your team to ignore it.
• Time to value in days. If a vendor needs six months before the first finding, the loss runs the whole time.
• No data team required. Mid-market grocers do not staff data scientists.
• A named limit. Ask every vendor what their method cannot see. Anyone who says "nothing" is selling the demo.

Where Ward fits

Ward is approach five. It connects read-only to your POS, ERP and inventory, watches those signals across every store, and ships an insight card naming the store, the category and what changed. First cards arrive within 48 hours of connection, and there is no hardware and no write-back.

It is not a counting system and not a camera. If your loss is mostly recorded spoilage you already reason-code, a logging tool will serve you better. Ward is for the unknown bucket and the speed problem: finding the gap in week one instead of month six, while the trail is still warm.

Methodology

Product descriptions above are taken from each vendor's own public documentation and checked on 2 October 2026. Vori's positioning comes from its shrink tracking product page, Datascan's from its grocery inventory accuracy material. Shrink rate figures are from the NRF National Retail Security Survey and, for the perishable department breakdown, from Datascan's published material.

Only products whose positioning we verified on the vendor's own site are named. Approaches three and four are described without named examples for that reason, and this is not a ranked shortlist. If a row here is wrong or out of date, that is a mistake rather than a position: write to brian@askotter.ai and it will be corrected or removed.