We built a procurement platform for a hospitality franchise network that forecasts demand from live usage data and sends purchase orders to suppliers automatically. Operating costs fell 25%, and waste fell 18%.
Client: Sentinel NetworkIndustry: HospitalityService: AI & Automation
Overview
Client
Sentinel Network
Hospitality franchise network
Engagement snapshot
The challenge
Franchise managers at 100+ locations ordered supplies through separate spreadsheets and emails. Reported stock was 15% out from what was actually on the shelves, which led to frequent emergency shipments and a lot of waste.
Orders were based on what managers could see and their own judgement. Turning a low-stock alert into a purchase order could take 48 hours, and head office had to hire more admin staff for every 10 new sites.
Unexpected stock-outs meant last-minute deliveries at 3x the standard shipping rate.
Over-ordering at 40% of locations tied up capital and caused high spoilage of perishables.
Franchise owners named manual ordering as their #1 pain point, and it was slowing the opening of new sites.
The solution
We built one IoT connection that reads live usage data from point-of-sale (POS) systems and smart scales, keeping an up-to-date digital copy of each site's inventory.
AI models study past consumption, local events and seasonal occupancy to forecast what each site will need.
The system creates purchase orders and sends them straight to vendor APIs, with an automatic audit trail for every line item.
Key features
AI forecasting models study past consumption, local events and seasonal occupancy to predict what each site needs, reaching 96% accuracy.
Business impact
96% forecast accuracy
AI picks the best vendor per site, and purchase orders are raised and matched to invoices automatically.
Business impact
POs raised within 2 minutes
For each site, the AI picks the best vendor based on live prices, lead times and regional stock.
Business impact
Zero manual sourcing
POs, invoices and delivery slips are matched automatically, so invoice fraud and short deliveries are caught straight away.
Business impact
100% financial accuracy
Order levels change as hotel bookings rise and fall, which stops overstocking when guests cancel unexpectedly.
Business impact
18% less waste
Before and after
Before
Best guess
Franchise managers ordered from visual checks and gut feeling.
After
Based on real usage
Forecasting models trigger orders from measured consumption.
Before
Scattered emails
Order records sat in different inboxes, with no central audit trail.
After
One shared ledger
Every PO trigger, vendor response and delivery is logged in one traceable ledger.
Before
More sites, more staff
Head office had to hire more admin staff for every 10 new franchise sites.
After
Same team at any size
The same system runs 100 or 1,000 sites without adding to head office operating costs.
Results
25% Operating Cost Reduction
Automation removed manual admin tasks and cut corporate procurement overhead by 25%.
Before: ManualAfter: 25% Lower
18% Less Waste
Matching orders to forecast demand reduced spoilage and over-stocking across the franchise network.
Before: High SpoilageAfter: Optimised
99.9% Fulfilment Rate
Orders are now fulfilled 99.9% of the time, and automated triggers raise a vendor PO within 2 minutes of stock running low.
Before: < 10 MinsAfter: 2 Mins
Testimonial
“Purchase orders now go out within 2 minutes, matched against invoices and delivery slips. Forecasts of what each site needs are 96% accurate, waste is down 18%, and operating costs across the network are 25% lower.”
Kenneth EdwardsDirector of Procurement, Sentinel NetworkKey results
Trust and control
Every vendor API connection uses encrypted, scoped tokens, so no vendor can reach the client's database.
Secure accessEach PO carries a data-origin tag that shows exactly which stock trigger caused it.
TraceableThe cloud setup can expand to 500+ sites without redesigning the system.
Scales with new sitesThe client owns 100% of the automation engine, demand models and custom vendor connectors.
100% ownedFaster delivery
Reusing tested components saved about 8 weeks, which the team spent on demand forecasting and rules for each franchise.
Provided tested connectors and scoped access tokens for sending orders to supplier systems.
Turned POS and smart-scale readings into live stock levels for each site.
Tracked each order from PO to delivery, so late or short shipments showed up quickly.
Checked each automatic order against budget before it went to a vendor.
More than 100 sites ordered supplies through spreadsheets and emails from franchise managers. Reported stock was 15% out from real stock, so emergency shipments at 3x the normal rate were common. At the same time, 40% of locations over-ordered, tying up cash and spoiling perishables. Franchise owners called ordering their biggest pain point.
Live data from POS systems and smart scales keeps a digital copy of each site's stock up to date. AI forecasts demand from past consumption, local events and seasonal occupancy with 96% accuracy. The system then picks a vendor on price, lead time and stock, and sends the purchase order automatically.
Operating costs fell 25% as manual admin disappeared, and waste dropped 18% because orders now follow bookings and real usage. Stock records are 100% accurate, order fulfilment reached 99.9%, and a low-stock alert becomes a vendor order in about 2 minutes. Head office no longer needs extra staff as sites open.
If your sites still order by email and gut feeling, we can help you automate purchasing from real usage data and keep a full audit trail.