In this article
The internet disappears. A customer is holding a purchase. Your cashier is holding their breath. A business continuity plan should explain what happens next before this becomes a live demonstration.
On-premise ERP runs on infrastructure controlled at a business location. With a properly designed local network, selected workflows can remain available when the outside internet connection fails. That is different from a browser application having an offline mode, and different again from a payment terminal being able to accept a particular transaction offline.


Can ERP work without the internet?
It can, for the functions deliberately designed and deployed to work locally. A workstation may still reach a local server over the office network even when the ISP is unavailable. Remote staff, cloud-only integrations and services that need external verification may remain unavailable. “Offline support” needs a written list of supported actions, not a hopeful tick on a proposal.
Offline invoices are not offline card authorisations
A local POS may record a sale, print a receipt and update its local stock without the internet. Whether it can accept a card or wallet payment depends on the payment provider, terminal, configuration and risk rules. Do not treat a queued sale as a bank-approved payment. Staff need a clear pending-payment status and a reconciliation procedure.
- Confirm which billing, stock and customer actions remain available locally.
- Define how transactions are identified and how duplicate submissions are prevented.
- Separate completed payments from payments awaiting provider confirmation.
- Plan what users see when prices, stock or customer records are stale.
- Document who resolves conflicts when branches reconnect.

The local server needs power, backups and an owner
A functioning local network does not help when the server has no power. Discuss the UPS, graceful shutdown, hardware maintenance, network security, update responsibility and off-site backup arrangements. Keeping everything in the office is not automatically safer; one damaged machine should not be the only copy of the business.
Agree a recovery point objective—the amount of data loss the business can tolerate—and a recovery time objective—the acceptable restoration time. Then test restoring a backup. “We have backups” and “we can restore operations” are related statements, but they are not the same statement.
How we scope an on-site ERP Enigma implementation
We assess the operating flow, locations, users, hardware and integration requirements before deciding which ERP Enigma functions belong locally, in the cloud or in a hybrid design. Offline operation, synchronisation and payment integration must be included in the agreed scope and tested against the actual deployment. This is implementation work, not a claim that every module works offline by default.
A Pakistani retail counter may prioritise billing during an ISP outage. A US warehouse may prioritise picking and dispatch on its internal network. An EU business may need additional access, retention and transfer controls. GCC hospitality teams may need branch-level continuity, while Australian and New Zealand operators should consider distance, connectivity and recovery support. The design starts with the interruption you need to survive.
Walk through an outage before buying offline support
Use this illustrative shop scenario during a demonstration. The local network is working, the internet connection has failed and two tills are still taking orders. Ask which actions remain available: looking up a product, checking locally held stock, recording a cash sale, printing a receipt and processing a return. Then ask which actions must stop. A local order record does not mean a remote card payment, tax service or delivery platform has approved anything.
The test needs more than switching off Wi-Fi on one laptop. Disconnect the external connection while keeping the intended local network running. Create transactions from both tills, restart one device and reconnect the internet. Check that every transaction reaches the central system once, with the right identifier, amount and status. This is a proposed acceptance test, not a claim that every deployment supports these functions.
Now try the awkward case: both tills sell the last locally visible unit. Decide whether a disconnected till can reserve stock, whether it uses an allocated quantity, or whether staff must confirm availability. There is no universal answer. The correct rule depends on how costly an oversell is and what the actual implementation can enforce. Write the rule down before the outage, when everyone is considerably less interested in arguing with a router.
Four failures that need different recovery plans
Internet loss, local-network failure, power loss and server failure are separate events. An on-site server may help with the first while remaining vulnerable to the others. A backup power supply also has a finite runtime; it needs a controlled shutdown plan rather than a hopeful label saying backup.
- Internet loss: identify local functions, remote dependencies and the queue that will synchronise after reconnection.
- Local-network failure: establish whether individual devices can work independently and how staff record activity when they cannot.
- Power loss: identify which server, router, switch, till and printer need power, and test the actual runtime under load.
- Server failure: restore to a designated replacement environment and verify that the team can access the recovered data.
Agree the synchronisation rules in plain language
Ask the supplier to show how duplicate uploads are prevented, how conflicting changes are flagged and who resolves an exception. A queued sale must not quietly disappear because a product record changed during the outage. Staff need a visible status such as awaiting sync, completed or needs attention, together with a safe route to resolution.
Define recovery point and recovery time separately. The recovery point describes how much recent information the business can tolerate losing. Recovery time describes how long the service can remain unavailable. Choose targets around operational consequences and verify the design against them; do not assume a backup file delivers either target automatically.
Before signing off, repeat the outage test with the employees who run the counter, review the resulting stock and finance records, and document any manual steps. If a proposed offline feature cannot be demonstrated within your agreed scope, treat it as unresolved rather than included. This is the right starting point for an on-site ERP discussion with Enigma.
A rollout checklist that beats surprises
- Disconnect the WAN deliberately and test the agreed local workflows.
- Test a power failure and the planned recovery sequence.
- Create two conflicting updates and confirm the reconciliation rule.
- Restore an agreed backup into a test environment.
- Train the team on pending payments, manual fallbacks and escalation.
Use the test results to choose your deployment model. Cloud, on-premise and hybrid each have sensible uses. The best option is the one your business can operate and recover—not the one with the most dramatic server photo.
Discuss on-site and hybrid ERP Enigma scope ↗
Build a practical POS and inventory workflow ↗

