Starting point
In contract logistics a provider runs warehouses for its customers, and the contract sets what each warehouse has to deliver: how many deliveries are received per day, how long unloading and putaway may take, how many shipments leave the building by when. The figures for this are produced in the warehouse management system. Site managers often see them only in an end-of-day report, when nothing can be changed in the running shift any more.
The client was the same consultancy for which we had already built a just-in-sequence system. This time the task was not steering a production line in seconds of takt time, but monitoring a warehouse's contracted performance over hours and days. The consultancy wanted to offer warehouse operators a product that keeps every site's KPIs in view, sets them against the agreed goals and reports in time when a goal is missed, on the move as well and not only at the desk.
Solution
Each site appeared as a flow in three areas: inbound, storage, outbound. Inbound showed deliveries and pallets, unloading, putaway and waiting times, the time spent on clarification cases and the staff deployed. Storage showed occupancy, overall, by zone, by ABC class and by category. Outbound showed shipments and parcels, pick, pack, consolidation and loading times and the staff in each area. The overview showed every site with its progress for the day against its inbound and outbound goals.
The values reached the platform without manual entry. Each site had a key of its own for the interface through which the warehouse management system or an operator's data pipeline sent KPIs in hourly and daily slices, along with alerts from its own system. Missing hours were filled in so the charts had no gaps. Forecast values came from the client's system: they appeared as dashed points for the rest of the shift, were replaced by the actual values as they arrived, and were kept for later comparison.
The agreed performance was data, not program logic. A contract value linked a KPI to a period (hour, day, week or month), a severity from low to highest and the sites it applied to. It was marked as an alarm, a notification or a goal, so the same value could feed the goal progress on the overview or trigger a warning.
A check ran every hour. It took each site's working days and shift hours into account, so an empty night shift woke nobody, compared the hourly values with the goals marked as notifications, and sent a push notification to every user who had switched warnings on. In the application the warnings were grouped by inbound, storage and outbound and were confirmed there.
For context, the platform set the day's average inbound and outbound times against the same weekday over the past eight weeks; the charts could overlay the weekday average or the previous period. The same codebase ran in the browser and as an app for iOS and Android, with native push notifications on both platforms and web push in the browser, in German and English. Companies, sites and users were separated by role, and a user with sites assigned saw only those sites.
- Inbound, storage and outbound per site, with amounts, lead times, occupancy and staff
- Connection through an interface with a key of its own per site, in hourly and daily slices
- Contract values as data: KPI, period, severity and sites, marked as alarm, notification or goal
- Hourly check against the goals, only on working days and within shift hours
- Push notifications to every user with warnings switched on, warnings confirmed by area
- Comparison with the same weekday over the past eight weeks; forecast and actual on one line
- One codebase for browser, iOS and Android
A look inside the product
Illustrative sample data; an anonymised product example that deliberately does not show the real product.
Result
The platform was built in 2021 in about five months, from the first commit to the iOS and Android apps with native push notifications. In 2022 we developed it further, adding automated builds that took the iOS app into test distribution.
It was designed so that a shift lead learns of a missed goal within the hour rather than in the next end-of-day report, while something can still be changed in the running shift. And because every site keeps the same KPIs against the same kind of contract value, sites can be set side by side.
Three decisions carried this: keeping the agreed performance as data instead of writing it into the check, taking the values over through an interface instead of having them typed in, and building browser, iOS and Android from one codebase.
- About five months of build time up to the apps with push notifications
- A missed goal on the phone within the hour instead of in the next end-of-day report
- Every site with the same KPIs against the same kind of contract value
- Browser, iOS and Android from one codebase