Separate Service Impact from Activity Urgency

Business impact analysis asks a different question from daily task prioritization: what happens to a defined service and its beneficiaries as disruption lasts longer? Every manager may call their activity urgent because their staff are busy, but workload alone is not an impact. Start with a service whose output is specified, a stated period and a unit for each measure. Appointment booking can be described by valid requests left unconfirmed, waiting time for priority groups, repeat contact effort, and a documented obligation if one exists. Payment settlement may be measured by pending transactions and the risk of duplicate entries. Do not invent a number for reputational damage. If prior complaints are documented, use them within their limits; otherwise record reputation as a qualitative risk requiring observation. The aim is a defensible recovery order, not a numerical score that creates false certainty.

Divide disruption time into decision points meaningful to the service: perhaps two hours, eight hours, one day and three days in a learning case. Those periods are not universal sector standards; the team chooses them from service cycles and when consequences appear. At each point ask how much valid output is missing, whether that output can be caught up without harm, and which consequence exceeds the decision maker's tolerance. Write an explicit unacceptable-impact threshold, such as more than 200 priority appointments unconfirmed before the evening shift ends. This is a local managerial decision, not a threshold ISO imposes on every organization. Describe the maximum plausible time before that unacceptable effect is reached as a conclusion from data and the threshold, not a figure copied from an old plan. Record who approved it and which assumptions were used so a later exercise can challenge them.

Separate product or service impact from the time needed to restart one activity. Telephone staff may still receive booking requests, but valid confirmation ceases without availability data. The service impact remains despite continued call intake. Conversely, an automatic confirmation message may fail while a worker can reliably deliver the decision by another route; activity failure then differs from complete service failure. Write the causal chain: lost resource, affected activity, delayed service output, then consequence for a beneficiary or obligation. That chain shows where an alternative can interrupt the impact. Do not count unconfirmed requests and follow-up calls as if they were different people. State the denominator for any percentage. Sixty follow-up contacts among 200 delayed requests equal 30% of delayed requests only if each contact represents a different request, which needs evidence. Otherwise report 60 contacts without claiming a percentage of people.

Impact may change nonlinearly with time. Settlement may be delayed for an hour with little visible effect and then require an exception when an internal accounting cut-off arrives. Suppose 50 appointment requests accumulate hourly. After recovery, staff can process 70 hourly while 40 new requests arrive, giving a net backlog clearance rate of 30, not 70. Four hours of disruption create a backlog of 200; clearing it takes about six hours and forty minutes after system restoration if those rates hold. The calculation demonstrates that a running system is not the same as restored service. Check whether rates stay constant, whether cases have equal effort, and whether staff can work that long. Priority cases may need a different sequence. Record the result as a bounded operational estimate rather than a promise to customers.

Conflicting evidence should not be smoothed into a convenient average. A sales manager may say every delay is critical while records from previous months show most customers accepted one-day rescheduling. That history does not disprove a claim about a small group under a special commitment. Split requests into impact groups rather than generalizing the average. For each group record the source, number of cases, period and exceptions. Distinguish a documented contract term from an internal target and an unverified customer expectation. If a supplier contract does not guarantee restoration within the period you need, do not record that period as proven capability. Ask for the agreement or a test result and send a conditional decision to the risk owner. Impact analysis produces priorities and requirements; strategies come later. A wish to purchase a preferred solution must not determine the impact that analysis is meant to measure.

Close the analysis with a reviewer who did not fill the table. Give that person the service definition and the source of every figure, then ask for an independent recalculation of one row and a priority explanation without showing the recommendation. If answers differ, check denominator, period and request group before changing weights. An explicit unknown is safer than a zero that implies no impact. Put the needed field measures into the first exercise: arrival rate, fallback throughput and output validity. State the interim decision that protects the highest-impact group while those measures are collected.