1,610 receipt submissions in the campaign report

RaftLabs delivered a receipt-entry and prize-draw platform for Musgrave's SuperValu and Centra campaign in Northern Ireland. In the first week, per the client, 471 users joined and 573 receipts were validated.

users in the campaign report
1,062
receipt submissions at the reporting snapshot
1,610
project period recorded in the task
About 2-3 months

Short answer

This receipt scanning software case study shows how RaftLabs built an AI receipt validation platform for Musgrave's SuperValu and Centra stores in Ireland. Shoppers uploaded receipts to enter weekly prize draws. Per the client, the campaign recorded 1,062 users and 1,610 receipt submissions in four weeks.

Engagement

The engagement

Client
Musgrave Group (SuperValu and Centra)
Sector
Retail and grocery
Market
Northern Ireland
Engagement
Fixed-price project with a dedicated team, about 2-3 months; BrandFire led the campaign, RaftLabs delivered the platform
What shipped
Receipt-entry and prize-draw platform with AI validation, strict receipt uniqueness checks, and store matching for SuperValu and Centra
Campaign scale
18 stores, 10 winners per store per week, EUR 10 minimum spend
Campaign record
1,062 users and 1,610 receipt submissions in four weeks, per the client

The situation

This receipt scanning software case study covers a prize campaign for Musgrave's SuperValu and Centra stores in Ireland. Musgrave wanted more footfall in its stores. Shoppers who spent EUR 10 or more could upload a receipt and enter a weekly draw.

Visual note: the opening image is an illustrative reconstruction based on the documented workflow. Its sample receipt values and interface details are not campaign records.

The draw picked 10 winners per store per week across 18 stores. That volume made hand checking slow and hard to keep consistent. BrandFire led the campaign, and RaftLabs built the platform.

In the first week, per the client, 471 users joined, 573 receipts were validated, and 76 winners were picked across 18 stores. Over four weeks the campaign reached 1,062 users and 1,610 receipt submissions.

Illustrative reconstruction of the Musgrave receipt scanning campaign flow
A supermarket receipt turning into loyalty points, hand-drawn illustration

Checking receipts at campaign scale

  • Keep one receipt to one entry

    Each receipt had to count once, and the check had to work from the photo alone. Every printed receipt carries details that, together, belong to one sale: the store, the till, the brand, and the exact time.

    A till rings up one sale at a time. When those details match a receipt already entered, the platform treats the upload as the same sale, even when the shopper photographs the receipt again.

  • Match each receipt to the right store

    A receipt only counts in the draw for the store printed on it. The platform matched the details on each receipt against the list of store addresses, then checked the entry against the campaign rules for that store.

    With 18 stores running the same campaign under two brands, SuperValu and Centra, a correct store match decided which weekly draw the shopper entered. The EUR 10 minimum spend formed part of that check.

  • Train the AI until its checks held up

    Reading a receipt photo is the first step. Turning what it says into a store match and a yes or no on the campaign rules takes repeated tuning.

    Most of the tuning time on this project went on the instructions the AI follows. Each round tightened how it read receipts and how reliably it picked the right store.

The call

Make the AI's choice simple, then check it

Each store had a long internal reference code. Asking the AI to pick the right store by copying one of those codes exactly made matching unreliable.

We gave each store a short number instead: 1, 2, 3, and so on. The AI picked a number, and the platform turned that number back into the right store and checked it.

That one change lifted validation reliability by about 20% in a single step. Together with the AI tuning, overall validation accuracy moved from about 80% to near 99%.

the campaign workflow

How the platform worked

A shopper registered, uploaded a receipt image, and got a campaign decision for that submission. Approved receipts entered the weekly draw for the store on the receipt. Operators could follow every submission by its recorded status across both brands.

  1. A shopper uploads a receipt and enters the store draw

    Illustrative reconstruction: the controls, copy, and values shown in this image are examples, not campaign records.

    The shopper flow ran in one place for SuperValu and Centra. A shopper registered and submitted a grocery receipt image. The platform checked the EUR 10 minimum spend, matched the receipt to its store, and recorded a status for the submission.

    Approved receipts moved into the weekly draw for that store, with 10 winners picked per store per week. This is the campaign pattern behind our grocery loyalty program build work for multi-store retailers.

    Illustrative reconstruction of the shopper receipt submission flow
  2. Validation checks uniqueness and store match on every upload

    Illustrative reconstruction: the receipt, extracted values, and interface shown in this image are sample content.

    On each upload, the platform read the receipt, matched it to a store, and compared it with receipts already entered. A full match on store, till, brand, and time marked the upload as a repeat of a sale already in the draw.

    The campaign record kept each outcome visible to operators. At the reporting snapshot, 1,158 submissions were AI approved and 409 were AI rejected. A further 4 were manually approved, 10 were manually rejected, and 29 were still processing.

    Illustrative reconstruction of receipt field extraction and store matching

Proof

Receipt scanning software case study results

Figures from the campaign record and the post-campaign report, per the client.

ResultWhat changedPeriod or contextEvidence and limitation
Users in week one471 usersFirst seven days of the campaignPer the client
Receipts validated in week one573 receipts validatedFirst seven days, across 18 storesPer the client
Winners in week one76 winnersFirst week, across 18 storesPer the client
Users over four weeks1,062 usersFour-week campaign totalPost-campaign report, per the client
Receipt submissions over four weeks1,610 receipt submissionsFour-week campaign totalPost-campaign report, per the client
Validation accuracyFrom about 80% to near 99%After simplifying store matching and tuning the AIDelivery team measurement during validation tuning

Where this applies

Use AI validation when receipt volume passes what a team can check by hand

A small campaign can check receipts by hand. A campaign with 10 winners per store per week across 18 stores produces a volume where hand checking slows the draw and makes each decision harder to keep to the same standard.

The delivery team put the point plainly: checking 500 receipts in one go is hard for a person to do without a mistake. AI validation fits that point. Each receipt gets checked against the same entry rules, and every decision leaves a record the operator can review.

If your retail campaign runs on receipt uploads, with a minimum spend and one entry per transaction, this pattern fits your project. Our loyalty platform development page covers the wider member platform work that sits around a campaign like this.

Frequently asked questions

Straight answers to what buyers ask before they scope. Still stuck? .

Have the store list ready with the correct address for every store taking part. Add sample receipts from each store, the entry rules, and the draw dates. The platform matches receipts against store addresses, so clean store data and real receipt samples let validation tuning start on the material the campaign will actually receive.

Yes. Musgrave ran SuperValu and Centra in one campaign on this platform. Each receipt was matched to the store printed on it, so the entry sat in the draw for that store and that brand, while operators worked from one campaign record.

It stays visible in the campaign record for a manual decision. The post-campaign report recorded manual outcomes alongside the AI ones: 4 submissions manually approved and 10 manually rejected, with 29 still processing at the snapshot. Operators kept the final call on those entries.

Work with us

Recognise this problem in your business?

Tell us what's broken. We'll diagnose it and show you exactly what to fix first, before you commit to anything.

  • Scope and cost agreed before work starts. No surprises. No obligation.
  • Working prototype within 3 weeks of kickoff.
  • Pay by milestone. You see progress before each invoice.
  • 60-day post-launch warranty. Bug fixes, UI tweaks, and deployment support. No retainer.
  • All conversations are NDA-protected.