
A business impact analysis: which systems must come back first
A business impact analysis answers one question before anything goes wrong: if this process stops, how soon does it hurt, and which systems have to be running again first? It is the step before the continuity plan. Without it, every recovery order is a guess, and usually the IT team’s guess rather than the board’s.
A business impact analysis starts with processes, not systems. Only once you know what the business cannot do without do you know which IT has to come back first.
Impact changes over time: an hour of downtime is often bearable, three days is not. The point at which it becomes unacceptable sets the recovery time.
The dependency list is the most valuable part. That is where you find the systems no process names and that still have to run first.
What a business impact analysis is, and what it is not
A business impact analysis (BIA) establishes, process by process, what happens if it cannot run for a while and how long that is allowed to last. The output is a ranking of processes, a permitted recovery time and permitted data loss for each, and a list of everything those processes depend on.
It is not a risk assessment. A risk assessment asks how likely a fire, an outage or ransomware is. A BIA does not ask why something fails, only what it costs when it does. That is what makes it useful: whether the server is down because of a breach or a botched update, the question of what comes back first stays the same.
Nor is it the plan yet. The continuity plan describes who does what when the time comes. The BIA supplies the numbers and the order. Write the plan without the analysis and the recovery times are filled in by feel; what that plan then looks like is covered in our business continuity plan template.
Start with processes, not the server list
It is tempting to start with the list of systems, because IT already has one. Don’t. A system list tells you what exists, not what matters. Start by asking what the business has to be able to do every day, every week and every month.
- Keep the list short and specific. Taking orders, shipping, invoicing, running payroll, keeping customer service reachable, keeping production running. Ten to twenty processes is enough for most SMEs.
- Give every process an owner. Someone from the business who knows how the work flows, not someone from IT. That owner answers the questions in the next step.
- Remember the peaks. Payroll runs once a month, year end once a year. A process that can wait on an ordinary Tuesday cannot wait on payroll day.
Hold the conversations process by process, with the owner and someone from IT in the room. A questionnaire sent by email produces answers that all say “critical”, and then you have no ranking.
Impact over time: when does it become unacceptable
The core question for each process is not whether downtime does damage but from when. So set fixed time windows side by side, for example the first hour, the first working day, three days and a week, and ask what happens in each.
- Money: lost revenue, penalties, the extra cost of working manually.
- Customers and commitments: deliveries that do not happen, contractual deadlines that pass.
- Law and regulators: reporting duties, filing deadlines, retention obligations.
- People and safety: in care or manufacturing this can outweigh everything else.
- Reputation: what customers see and what reaches the outside world.
Use a simple scale (low, noticeable, serious, unacceptable) rather than exact figures. The point at which a process reaches “unacceptable” is the maximum tolerable period of disruption. The recovery time you agree sits well before it, because recovery itself takes time. Also ask how much work may be lost: a morning of orders re-entered, or nothing at all.
From process to system: the dependencies
This is where the IT part starts, and where most of the value lies. For each process, record what it runs on, layer by layer:
- Applications: the ERP, the finance package, the web shop, the scheduling software.
- What those applications need: a database, a server or cloud environment, an integration with another package.
- The foundation: network and internet, sign-in and MFA, DNS, the backup itself.
- Third parties: the SaaS supplier, the hosting provider, the bank for payment files, the carrier with its own portal.
- People and kit: the one colleague who knows how the integration works, the label printer in the warehouse.
Draw it. A diagram on a single sheet shows what a table hides: the system five processes hang on at once, and the single point that can stop everything. Often that is not the server everyone knows but the account everything signs in with, or the one internet line at head office.
Which systems must come back first
The recovery order follows from the analysis, and it is almost never the same as the ranking of processes. Two rules make the difference.
The foundation comes first, even if no process names it. Network, identity and access, and a clean backup to restore from sit at the top, because without those three nothing else comes back. An analysis that only ranks applications leaves them out, and you discover during recovery that the ERP is ready but nobody can sign in.
A system inherits the strictest requirement of every process that relies on it. If invoicing runs on the same package as the payroll payment files, that package gets payroll’s recovery time. Then group the systems into a few recovery tiers rather than one long list: what must be back within hours, within a day, within a week. How to get the recovery time for each tier right is covered in our article on the recovery time objective.
The gap between what the business asks and what IT can do
Almost every first BIA reveals a gap. The business asks for a recovery time of a few hours for the ERP, and the current backup restores it in a day or two, if all goes well. That gap is not a failure of the analysis; it is exactly why you do it.
For each gap, set two things side by side: what the current set-up actually achieves (measured in a restore test, not estimated) and what it would take to meet the requested time. That might be more frequent backups, a second site, spare hardware on the shelf or a manual fallback that keeps the process going for a few days without the system.
What happens to the gap is the board’s decision. A shorter recovery time costs money and management effort; a longer one costs risk. That trade-off is about the business. IT says what each option requires, the board says which it takes, and the decision is written down, including the gaps that deliberately remain.
How the analysis stays current
A BIA goes out of date faster than you expect: a new package, a process moving to the cloud, another office. So agree when it will be reviewed. Once a year is a reasonable minimum, plus every significant change to processes, applications or suppliers.
- Keep it short. A fifty-page analysis does not get updated. A table per process and one dependency diagram does.
- Tie it to testing. A restore test or tabletop exercise shows whether the order holds. Whatever goes wrong there feeds back into the analysis.
- Let it feed the plan. Ranking, recovery times and dependencies go straight into the continuity plan and the IT recovery plan. Two documents with different numbers are worse than one with old ones.
If your organisation falls under NIS2, business continuity is among the measures you must be able to demonstrate, and a maintained BIA is the supporting evidence rather than extra paperwork. The international standard for continuity management is ISO 22301, with ISO/TS 22317 as guidance for the impact analysis itself.
Where our work stops
What we do. We do the IT part of a business impact analysis: mapping dependencies from application to internet line, measuring the recovery time your current backup and set-up actually achieve, working out the recovery tiers and saying what it takes to close a gap. We are ISO 9001 and ISO/IEC 27001 certified, and that describes how we work.
What we do not do. We do not decide which impact is acceptable for your business or which recovery times you commit to; that is for your board, with the process owners. We do not calculate financial loss for your accounts or your insurer, we are not an insurer, and we certify nobody against ISO 22301 or NIS2: that is for an accredited body or your regulator.
Questions we get about this
The five that come up most often.
What is a business impact analysis?
A business impact analysis (BIA) establishes, for each business process, what happens if it cannot run for a while and how long that is allowed to last. The output is a ranking of processes, a permitted recovery time and data loss for each, and a list of the systems and third parties they rely on. It is the step before the continuity plan.
What is the difference between a BIA and a risk assessment?
A risk assessment looks at how likely a threat is, such as fire, an outage or ransomware. A BIA looks only at the consequences of disruption, whatever the cause. You need both: the risk assessment shapes what you try to prevent, the BIA what comes back first if it happens anyway.
Who should carry out a business impact analysis?
Process owners from the business answer the questions about impact and recovery time, because they know what downtime costs. IT maps the dependencies and says what is technically achievable. The board makes the decision on recovery times and on the gaps that remain.
How often should a BIA be reviewed?
At least once a year, and also after every significant change: a new package, a process moving to the cloud, a new office or supplier. Tie the review to a restore test or tabletop exercise, because that is where you find out whether the order still holds.
Can you carry out the business impact analysis for us?
The IT part, yes: mapping dependencies, measuring the recovery time your current set-up achieves and saying what it takes to close a gap. Which impact is acceptable and which recovery times you commit to is decided by your board with the process owners. We do not certify against ISO 22301 or NIS2 and we are not an insurer.
Further reading
What the analysis feeds into.
Shall we map your dependencies once?
Name the three processes you can least do without. We will show what they run on, what your current recovery actually achieves and where the gap is; which recovery time you choose is your call.
Practical IT knowledge in your inbox
New guides on management, security and the workplace, written by the people doing the work. No sales talk, and you can unsubscribe in one click.