Turning on multi-factor authentication (MFA) can take an afternoon. Making it work across a business takes longer. The switch is one day; the real project is the six weeks after it, when staff need help, exceptions have to be resolved and overlooked systems begin to surface.
For UK businesses, MFA is now a baseline security expectation. Cyber Essentials addresses it where it is available, ISO 27001 assessors look for evidence that controls are implemented, and insurers and customers increasingly ask how access is protected. But a setting enabled in a cloud tenant is not proof that every important route into the business is covered. The gap between what a certificate asks for and what an actual estate can deliver is where strategy matters.
For most staff using a modern laptop and a company phone, the change is straightforward. The difficult cases are more varied: a shift worker on a shared PC, a warehouse handheld, a legacy line-of-business application with its own user database, an integration using a service account, or a scheduled job whose owner has left. The NCSC guidance on MFA for corporate online services is a useful reference for thinking about coverage across staff, contractors and administrators, and choosing stronger methods rather than relying on weak second factors.
MFA protects accounts when those accounts can present a second factor. It does not automatically modernise every system those accounts reach. Some applications cannot use the organisation’s central identity service; some devices have no practical way to prompt a person; and machine-to-machine processes should not be treated like interactive staff sign-ins. A sound rollout distinguishes those cases, then decides how each risk will be managed. Simply excluding difficult systems leaves a control on paper and a support queue growing in the background.
The first fortnight is where the day-to-day impact is most visible. People need help enrolling, replacing a lost phone or signing in after a device upgrade. New starters may not yet have a registered factor. Contractors and staff who work across two sites may need clear instructions for different devices or networks. Monday morning after a weekend phone upgrade is not an unusual edge case; it is a predictable support event.
Plan extra cover and a clear process before enforcement. A Helpdesk Support process should record enrolment and recovery requests, verify identity before resetting a factor, and explain what happens when a device is lost. The ticket log is useful evidence of the process in practice, not just a record of inconvenience. Managed IT Services can provide bolt-on execution for that work, but the strategy should define ownership, verification and escalation first. A lost-device route that relies on one administrator remembering an undocumented workaround is not a process.

The exceptions list is the real deliverable. For every system that cannot use the intended MFA method, record what it is, why it is excluded, who approved the decision, what compensating protection applies and when the exception will be reviewed. An exception with a name and a date is a control decision. An exception nobody recorded is a hole.
This is also where the rollout can uncover accounts that should no longer exist. A service account may have been created for a one-off integration; an old supplier login may still be active; a dormant mailbox may be attracting a licence charge without serving a business need. Technology Expense Management belongs before the rollout, not as an afterthought. Stronger authentication may require a higher licence tier, hardware tokens or identity tooling, so review the subscription estate and account inventory before purchasing upgrades. There is little value in paying to protect accounts the business no longer needs.
The longest tail often sits in established operational platforms. In IBM i Management, user profiles may support batch jobs, integrations and third-party support as well as human access. A 5250 session does not naturally present a browser-based second-factor prompt. Applying the same control to every profile without understanding its purpose can interrupt business processes; excluding the platform from the discussion entirely is the more common mistake.
Practical protections depend on the platform and how it is used. A gateway or jump host can put a modern authentication step in front of access. Network-level restrictions can narrow where sessions may originate. Named profiles can be reserved for human access, with machine-to-machine activity separated and documented. Where MFA genuinely cannot be applied to a particular path, monitoring and logging can improve visibility, alongside a named owner and a dated plan to reduce the exposure. Profiles nobody can explain are precisely the ones that need an owner. That requires specialist technical consultancy, not a blanket exception.

Disaster Recovery planning has its own MFA test: who can get in at 2am if the second factor is a phone at the bottom of a river? And what happens if the identity provider itself is unavailable? These are different failure scenarios, and a single emergency account may not solve both.
Microsoft's guidance on planning for mandatory multifactor authentication is a sensible reference for staging the rollout and thinking about exceptions and emergency access in advance.
Document and test emergency access before enforcing the policy. Keep recovery credentials and any physical security keys in controlled, secure custody, with access limited to named people. Alert on use and review it afterwards. If a cloud identity provider is down, an account hosted by that same provider may not be an independent route in. The recovery plan should spell out the appropriate escalation and recovery path for the systems involved, then test it. A plan that depends on someone’s personal authenticator is not a plan. Resilience means being able to get in on the worst day, not just being locked out during it.
A modern cloud core can make the identity design simpler. With Dynamics 365 Business Central, tenant configuration and platform authentication can provide a more consistent foundation than an ageing on-premise ERP with separate logins and accumulated exceptions. That is one practical argument for modernising, although migration still needs a considered plan. A Dynamics 365 Jump Start can help shape that conversation around the business system and the infrastructure needed to support it. The consultancy-first question is not merely “How do we switch on MFA?” but “Which access paths do we have, which are legitimate, and how do we protect each one without breaking operations?”
The same question reaches beyond the laptop. Network Solutions need to account for remote VPN access, while Voice and Connectivity may involve administration portals or phone services with their own sign-in controls. These are supporting parts of the wider security design, not reasons to treat every service as a separate project.
The costs are therefore broader than the licence change. They include staff and helpdesk time, tokens or identity tooling where needed, investigation of legacy systems, remediation of accounts and the work of documenting and reviewing exceptions. A deliberate plan helps direct that spend towards real exposure rather than paying to preserve unnecessary accounts. ISO 27001 and Cyber Essentials are useful consultancy domains here: assessors look for coverage, credible exceptions and evidence that emergency access has been tested, not a policy document in isolation.
A practical rollout starts by inventorying the systems and deciding which can present a second factor and which cannot. Agree the exception list with named owners and review dates. Plan the support fortnight with extra cover and a clear lost-device process. Set up and test emergency access before enforcing anything. Tell staff what will happen and when. Then diarise the exception review so go-live is a milestone, not the finish line.
For the wider Evestaff group of services, visit evestaff.co.uk.
Join The Discussion