A workflow audit is the diagnostic layer of the Workflow Friction framework. It helps turn “this process feels clunky” into evidence about where work is being slowed, repeated, fragmented or made harder than necessary.
Workflow audit
Define the audit boundary
The define the audit boundary step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to scope, trigger, start point, finish point, dependencies. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Write an outcome statement
The write an outcome statement step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to value, constraints, quality, safety, service expectations. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Build the intended workflow map
The build the intended workflow map step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to inputs, actions, decisions, outputs, owners, waiting. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Observe real work
The observe real work step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to recent cases, neutral prompts, workarounds, evidence. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Build the observed workflow map
The build the observed workflow map step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to unofficial actions, deviations, loops, informal channels. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Audit information flow
The audit information flow step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to source of truth, identifiers, context, validation, trust. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Find search friction
The find search friction step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to locations, failed searches, verification, retrieval cost. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Audit handoffs
The audit handoffs step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to sender, receiver, payload, acceptance, ownership, queues. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Measure waiting and queue friction
The measure waiting and queue friction step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to active work, waiting causes, dependency, sequence. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Detect rework and loops
The detect rework and loops step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to rejections, missing inputs, repeated checks, upstream causes. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Measure context switching
The measure context switching step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to interruptions, recovery, restart points, necessary versus accidental. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Audit decision friction
The audit decision friction step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to rules, defaults, ambiguity, escalation, judgement. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Separate tool friction from workflow friction
The separate tool friction from workflow friction step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to interface problems, system boundaries, process causes. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Separate necessary complexity from accidental friction
The separate necessary complexity from accidental friction step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to controls, risk, judgement, counterfactual tests. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Interview people without leading them
The interview people without leading them step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to specific events, neutral prompts, multiple roles. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Create an evidence table
The create an evidence table step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to stage, behaviour, friction type, evidence, cause, consequence. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Build a practical friction taxonomy
The build a practical friction taxonomy step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to search, switching, waiting, handoff, duplication, rework, ambiguity. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Prioritise findings
The prioritise findings step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to frequency, effort, impact, confidence. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Add reversibility to prioritisation
The add reversibility to prioritisation step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to experiments, rollback, risk, learning. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Design the smallest useful intervention
The design the smallest useful intervention step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to remove, simplify, standardise, automate. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Decide when automation is appropriate
The decide when automation is appropriate step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to repeatability, stable rules, reliable inputs, exceptions. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Design workflow experiments
The design workflow experiments step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to hypothesis, sample, baseline, outcome, side effects. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Measure more than speed
The measure more than speed step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to outcome, quality, effort, waiting, errors, rework. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Audit the unhappy path
The audit the unhappy path step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to exceptions, urgent cases, escalations, rare versus frequent. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Avoid interviewing managers only
The avoid interviewing managers only step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to triangulation between policy, practice and system data. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Avoid activity-only productivity measures
The avoid activity-only productivity measures step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to proxies, gaming, downstream quality, value. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Avoid buying another tool too early
The avoid buying another tool too early step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to problem statement before solution, alternatives. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Run a 15-minute rapid audit
The run a 15-minute rapid audit step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to one workflow, one case, three priority findings. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Run a 60-minute facilitated audit
The run a 60-minute facilitated audit step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to scope, reconstruction, friction mapping, decision. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Audit a multi-team workflow
The audit a multi-team workflow step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to interfaces, assumptions, identifiers, ownership. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Document decisions during the audit
The document decisions during the audit step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to rationale, alternatives, owner, review condition. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Turn findings into a backlog
The turn findings into a backlog step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to problem statements, evidence, owner, next step. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Validate after the change
The validate after the change step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to before/after, displacement, quality, maintenance. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Recognise systemic problems
The recognise systemic problems step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to patterns across workflows, shared causes, validation. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Know when policy is the problem
The know when policy is the problem step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to purpose of rules, risk controls, simplification. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Improve information design
The improve information design step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to labels, defaults, grouping, links, status, ownership. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Make the audit a continuous loop
The make the audit a continuous loop step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to observe, understand, prioritise, intervene, validate. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Workflow audit
Audit before you optimise
The audit before you optimise step gives a workflow audit a practical boundary. The focus is not to produce a perfect diagram; it is to make the part of the system under investigation observable. For this stage, pay particular attention to evidence, proportionality, system-level outcomes. A useful audit names what is known, what is assumed, and what still needs evidence.
Start with a real example rather than an abstract description. Ask what triggered the work, what happened next, what information was needed, and what the person did when the expected path did not work. Record the behaviour in plain language. Avoid replacing observations with explanations too early. If someone says a process is “slow”, ask where the time was actually spent and what caused it.
Then test the finding across more than one case. A single awkward event may be an exception; a repeated pattern is stronger evidence of workflow friction. Compare frequency with consequence. A tiny annoyance that occurs hundreds of times can matter more than a dramatic problem that happens once a year. Also look for displacement: a local improvement may simply move work to another person, team or stage.
The output from this step should be actionable. Capture the observation, likely cause, consequence, confidence and the next question to investigate. Do not require the solution to be known yet. The purpose of the audit is to create enough clarity that the eventual intervention can be chosen deliberately rather than because a familiar tool, policy or automation pattern happens to be available.
Template
Example evidence table
| Stage | Observed friction | Evidence | Consequence | Next question |
|---|---|---|---|---|
| Intake | Clarification | Required identifier missing in 3 of 5 cases | Work waits for follow-up | Can the identifier be captured earlier? |
| Research | Search | Three locations checked for current source | Extra effort and verification | Can one source become authoritative? |
| Handoff | Context loss | Receiver asks for decision history | Sender repeats explanation | Can the decision travel with the work? |
| Review | Rework | Submission returned for missing field | Cycle time increases | Can validation happen at intake? |
Practical tool
Reusable workflow audit checklist
- Define the trigger and intended outcome.
- Set the audit boundary and name dependencies.
- Gather the documented process without treating it as ground truth.
- Observe at least one recent real case.
- Map the observed workflow separately from the intended workflow.
- Record searching, switching, waiting, handoffs, duplication and rework.
- Capture missing context and information sources.
- Identify recurring decisions and unclear ownership.
- Distinguish necessary controls from accidental friction.
- Classify each finding using a consistent friction taxonomy.
- Record evidence before proposing solutions.
- Score frequency, effort, impact and confidence.
- Consider reversibility and risk.
- Prefer removal before simplification, standardisation and automation.
- Define an outcome measure and friction measure.
- Run a contained experiment when uncertainty is high.
- Validate for displacement and unintended consequences.
- Document decisions and keep the audit available for future review.
Questions
Frequently asked questions
How long should a workflow audit take?
A focused audit can take fifteen minutes to an hour. Complex cross-team workflows require more observation. Start narrow and expand only when evidence shows that a wider boundary is necessary.
Who should participate?
Include people who perform the work as well as process owners. For cross-team workflows, include representatives at important handoffs. Different roles expose different forms of friction.
Should every manual task be automated?
No. Manual judgement can be valuable. Remove unnecessary work first, simplify what remains, standardise recurring cases, and automate only where the workflow is stable enough to be reliably supported.
What if people disagree about the workflow?
Treat disagreement as evidence. Reconstruct a recent real example and compare it with the documented process. The goal is not to decide whose memory is correct but to understand what the system actually requires.
What should be measured after a change?
Measure the intended outcome plus a small set of relevant friction indicators such as active effort, waiting, rework, errors or handoffs. Check whether work was merely displaced elsewhere.
From audit to action
A good workflow audit does not end with a diagram. It ends with a clearer understanding of what should be removed, simplified, standardised or automated—and evidence for deciding what to do first.
Use the Workflow Friction flagship to place your findings in the wider framework, then return to this audit method whenever a workflow starts to feel harder than it should.
Explore Workflow Friction Get in touch