Lidl 2018: The SAP Project Cancelled After Seven Years and Hundreds of Millions of Euros
Lidl spent seven years trying to replace its in-house merchandise system with SAP-based eLWIS before cancelling the project in 2018 after costs reportedly reached hundreds of millions of euros.
eLWIS was meant to modernize the heart of Lidl’s retail operations
Lidl began the eLWIS merchandise-management transformation around 2011. The company wanted integrated process chains, simpler master-data management, real-time analytics, and a platform that could scale across its international store and logistics network. SAP Retail and HANA were central to the design. [1] Retail merchandise systems sit at the center of purchasing, inventory, price, promotions, store replenishment, logistics, and finance. Replacing them is therefore closer to changing the company’s operating language than installing a new back-office application.
The hidden assumption
Lidl wanted to replace a long-lived in-house merchandise-management platform with a standardized global SAP Retail environment running on HANA. The engineering implication is that fit-to-standard discipline, early requirements-gap decisions, controlled customization, benefits tracking, and explicit stop-loss checkpoints cannot remain an aspiration or a postmortem recommendation; it has to be turned into measurable release criteria, ownership, and tested operating procedures.
The project ran for seven years and reached selected countries
Industry reporting says the new environment went live in Austria and some other smaller markets while the wider rollout remained incomplete. The project became one of the largest transformation efforts in Lidl’s history, involving large internal teams, consultants, and extensive system integration. [2] Partial rollout can be useful evidence if management treats each country as a hypothesis test. It becomes dangerous if successful operation in a smaller market is assumed to prove readiness for much larger countries with different volume and complexity.
Where the rollout became fragile
The project aimed to support thousands of stores and logistics locations, so even small gaps between Lidl’s operating model and standard software could create huge customization costs. The engineering implication is that fit-to-standard discipline, early requirements-gap decisions, controlled customization, benefits tracking, and explicit stop-loss checkpoints cannot remain an aspiration or a postmortem recommendation; it has to be turned into measurable release criteria, ownership, and tested operating procedures.
Process fit and customization became a central source of complexity
Reports about the failure repeatedly point to differences between Lidl’s established merchandise and inventory practices and the assumptions of standard SAP Retail. The deeper issue was not one configuration setting; it was the strategic choice between changing business processes to fit standard software and customizing the software to preserve existing operating logic. [3] Standard ERP creates leverage through common behavior. Heavy customization spends that leverage. The organization inherits custom code, custom testing, custom upgrades, and a wider gap between vendor documentation and actual operation.
Why the failure spread
The longer eLWIS continued, the more other systems, teams, and transformation work became attached to the target architecture, increasing the cost of stopping. The engineering implication is that fit-to-standard discipline, early requirements-gap decisions, controlled customization, benefits tracking, and explicit stop-loss checkpoints cannot remain an aspiration or a postmortem recommendation; it has to be turned into measurable release criteria, ownership, and tested operating procedures.
By 2018 management decided the strategic goals were not achievable at reasonable effort
Lidl stopped eLWIS in 2018 and decided to continue developing its in-house merchandise-management system. Company statements reported by German industry press emphasized that the decision was not a wholesale rejection of SAP but a judgment about this specific transformation and its cost-benefit path. [4] Stopping also required disentangling integrations already built around eLWIS. The more surrounding systems point to the target ERP, the more cancellation becomes a second migration project.
What recovery really required
By 2018 management concluded that the original strategic goals could not be achieved at acceptable effort and chose to reinvest in the older in-house system. The engineering implication is that fit-to-standard discipline, early requirements-gap decisions, controlled customization, benefits tracking, and explicit stop-loss checkpoints cannot remain an aspiration or a postmortem recommendation; it has to be turned into measurable release criteria, ownership, and tested operating procedures.
The widely reported cost was around €500 million, not an audited project figure
Handelsblatt, Computer Weekly, Computerwoche, and other industry sources cited roughly €500 million as the amount invested or lost. Lidl did not publicly publish a detailed audited project ledger establishing one exact number, so the responsible historical phrasing is ‘reported’ or ‘estimated’ hundreds of millions of euros. [5] The uncertainty around the exact cost is itself a useful reporting lesson: dramatic round numbers become sticky in ERP folklore. A careful article should preserve the difference between trade-press estimates and audited corporate disclosure.
ERP programs need an early decision on process conformity
Every large package implementation discovers gaps. The organization must decide which legacy practices are genuine competitive advantages, which are historical habits, and which should be retired. If that decision is postponed, customization grows until the package becomes a new custom system with vendor-upgrade constraints. Fit-to-standard workshops need empowered business leaders who can change process, not only IT teams who can configure software. Otherwise every gap becomes a technical customization request by default.
Benefits tracking should make cancellation possible before year seven
A transformation should have measurable outcomes for data quality, process cycle time, store support, logistics, forecasting, and cost. If those benefits repeatedly move into the future while customization and integration cost rise, governance needs explicit stop-loss checkpoints independent of sunk cost. Benefits should be measured in production, not only in business-case spreadsheets. A platform that promises real-time analytics or lower master-data effort should show those outcomes in early deployments before global rollout continues.
Why Lidl 2018 belongs in software-failure history
Lidl belongs in software history because the company eventually did the hardest thing in a giant ERP program: it stopped. Seven years of work and reported spending in the hundreds of millions created enormous sunk-cost pressure, yet leadership concluded that continuing was worse. The enduring lesson is to decide process fit early, constrain customization, measure business benefit continuously, and preserve the option to abandon an architecture before it becomes irreversible. The decision to return to the legacy system also shows why legacy software is not automatically bad software. A mature in-house platform can remain economically preferable when a replacement fails to fit the operating model.
Works Cited
- 01Computer Weekly — Lidl Dumps €500m SAP Project computerweekly.com
- 02Handelsblatt — How SAP and Lidl Sank Hundreds of Millions handelsblatt.com
- 03Computerwoche — Lidl Stops eLWIS SAP Project computerwoche.de
- 04Heise c't — Elwis lebt – nicht mehr heise.de
- 05Panorama Consulting — Lidl ERP Failure Case Study panorama-consulting.com
CodeHistory is a living archive. Citations document the evidence used for this edition; later evidence may refine the account.