Retail Shrink and Loss Prevention Software: Build vs Buy

What Appriss Retail and Agilence actually do well

Take the buy answer seriously before reading the rest. Under about 40 stores, a productised exception reporting tool plus a consistent store audit calendar finds more money than a project will, and the honest constraint at that size is people to work the flags rather than software to produce them.

Appriss Retail is strong on returns specifically because it sees a consortium view of return behaviour across retailers that you cannot reproduce from your own data at any price. That is a genuine advantage and it does not go away when you build. Agilence covers till and service desk exception reporting competently, ships with a rule library assembled from a broad sample of retailers, and gets a mid size chain to useful findings in weeks. Sensormatic at the door and computer vision at self checkout each do their own job well inside their own boundary.

They also carry things a build never gives you free. Somebody else updates the rule library when a new fraud pattern spreads. Somebody else keeps the point of sale (POS) journal parser working through a till software upgrade. Somebody else is on the phone when the overnight load fails.

Buy while your loss is opportunistic and your estate is homogeneous. Read on if your loss has become organised and your estate has become three acquisitions.

Where they stop: the case file lives on somebody's desktop

A shrink number is an accounting result. A case is what recovers money and stops the crew, and no exception reporting tool owns the case.

It is nine at night and a regional asset protection manager is working the daily report for 340 stores. One store shows a cashier with 41 no sale drawer opens on a single shift. Another shows a service desk associate who has processed eleven no receipt refunds to gift cards in eight days, every one landing just under the supervisor approval threshold. A third flag is an order picked, marked collected and refunded 40 minutes later at a different store. Three rows in a report. Not one of them is a case yet.

Turning them into something a prosecutor accepts means pulling video from two recorder brands whose clocks disagree by ninety seconds, exporting journal detail at line level, matching a loyalty identifier to a phone number that also appears on a marketplace listing, writing a narrative, and assembling a package that survives a defence attorney reading it. In the asset protection teams we have built for, an investigator spends six to twelve hours on a single organised retail crime package, and most of that is copying, renaming and cross referencing files. That is why most flagged incidents quietly expire.

The second thing that stops is the rule library itself. Productised rules are built from somebody else's shrink. Your fraud is shaped by your policies. If you allow no receipt returns to a gift card up to $100, your abuse forms at $99. If your price match permits an override without a manager code, that is where the leakage goes. Vendor rules can be tuned, but only inside their model of a transaction, and when your pattern depends on an override reason code their schema does not carry, the rule cannot be written at all. So the team exports to a spreadsheet every Monday, and a shadow reporting function is born.

The arithmetic: cost per store per month versus a build

Exception reporting is usually priced per store per month, sometimes with a data volume component. Take your rate, multiply by store count, multiply by 36 months, and add the second and third tools, because most chains run at least three products across returns, till exceptions and video.

Then price the part the subscription does not touch. Take last quarter's flagged incidents, count how many became a working case, and count how many closed with a recovery, a civil demand or a referral. The distance between those numbers is the leak, and it is measured in investigator hours rather than in licence fees.

Now count the cases lost to recorder overwrite. Every chain has them and almost nobody counts them. A clip that existed on the day of the incident and did not exist by the time an investigator got to it is a case that was funded and then thrown away.

The crossover in our delivery experience arrives near 150 stores, and it moves earlier when you carry more than one point of sale journal format or more than two video platforms. Below 40 stores, buy and hire. Between those numbers it depends on how heterogeneous your estate is after acquisitions, which is a question about integration rather than about size.

What a custom loss prevention build actually costs

Bands, from delivery experience rather than a price list. A first release covering exception rules run against your own point of sale, returns and fulfilment data plus a real case file with an evidence chain and immutable audit logging runs $90,000 to $180,000 and ships in 14 to 20 weeks. A full platform adding video correlation with clock drift handling, electronic article surveillance and self checkout vision events, entity resolution across stores, civil recovery and restitution tracking, store audit workflows and shrink attribution reporting runs $250,000 to $600,000 phased across 9 to 15 months.

Data migration adds 10 to 25 percent, and here it is mostly history and open cases. Closed cases can be loaded for retention. Open ones move with their evidence, their hashes and their custody log intact or they lose the property that makes them usable.

Year two runs 15 to 20 percent of build cost annually. Till software upgrades change journal formats, video platforms change export interfaces, and new fraud patterns need new rules written by people who understand your policies.

What pushes it up: the count of distinct video platforms, because each integration is real weeks and some older recorders have no usable interface at all. Point of sale heterogeneity after acquisitions, since two journal formats means two parsers and two rule sets. Transaction volume, which at several hundred stores forces a columnar store rather than a general purpose database. And legal review of retention and access, which is not optional when the records are discoverable.

What keeps it down: returns and refunds only, at your worst 30 stores, with video correlation deferred to phase two.

The four situations where building wins

• Regulatory fit. This system holds evidence, which changes the engineering. Under Federal Rule of Evidence 902(14), electronic records can be self authenticating where a qualified person certifies them by comparing a hash value produced by a reliable process, so hashing on ingest is not a nice detail, it is what keeps you out of an authentication argument. Retention has to be defensible in both directions, since preservation duties attach once litigation is anticipated while privacy rules push the other way. If your estate uses facial recognition anywhere, the Illinois Biometric Information Privacy Act sets written consent and retention schedule duties with a private right of action attached, and Texas and Washington have their own regimes. And recovery work now runs alongside the INFORM Consumers Act, which since 27 June 2023 has required online marketplaces to verify high volume third party sellers, giving your team a route to a name that did not exist before.
• Scale economics. Per store per month across three overlapping products, where every acquisition adds stores and another journal format at the same time.
• A workflow that is your competitive advantage. Entity resolution across suspects, vehicles, phone numbers, addresses and loyalty accounts is what turns scattered incidents into one organised retail crime case, and it depends entirely on the identifiers your own systems capture.
• Integration sprawl across three or more systems. Returns scoring, till exceptions, video, door alarms, self checkout vision, order management and the case folder on a shared drive. Every one is competent alone and none of them owns the object your team actually works.

Two of those true is a build. One of them is a better rule tuning session with your current vendor.

How to decide in a week, with ten flags

Take ten flags from last month's exception report, spread across returns, till and fulfilment.

Monday: assemble one of them into a complete case package exactly as you would hand it to a prosecutor. Time it honestly, including the video pull. That single number is the business case and most executives have never seen it.

Tuesday: for the other nine, check whether the relevant footage still exists. Count how many are already gone. Then look up when each incident was flagged and when an investigator first opened it.

Wednesday: write down your three most abused policies and try to express each as a rule in your current tool. The ones you cannot write are the ones costing you money, and the reason is almost always a field the vendor schema does not carry.

Thursday: take last month's inventory adjustments in your five worst stores and try to split them into theft, receiving variance, damage, markdown error and unrecorded write off. If you cannot, every intervention you fund is a guess applied evenly.

Friday: price it. Cases lost to overwrite, plus incidents flagged and never worked, plus three years of subscription across every tool. Under roughly $250,000 a year, tune the rules and hire an investigator. Above it, build the case file and evidence chain first and leave video correlation to phase two.

If it points to build, the next step is a paid discovery rather than a proposal. Two to three weeks, fixed fee, ending in a signed product requirements document covering the case model, the evidence chain and retention rules agreed with your counsel, the rule set, the integration list by vendor and version, and acceptance criteria. You keep that document whoever builds from it.

We are wrong for you if you run under 40 stores, if your loss is mostly receiving variance and damages rather than theft, or if you are choosing on hourly rate, and we would rather say so than bid it. Where Digital Heroes fits: more than fifty specialists, over 2,000 projects, our own products including ShopScore, HeroCheckout and Section Vault, and India LLP, US LLC and UK LTD entities so intellectual property assigns under your own law. This system holds evidence you may have to produce in court, so the repository and the cloud accounts are yours from the first commit, and our record is checkable on Clutch, Trustpilot, Fiverr Vetted Pro and our D-U-N-S listing.

Book a 30-minute call with Digital Heroes and get a written plan and a fixed quote within 48 hours.