dag: Should I automate this business process?
version: 1.0.0
image: https://images.unsplash.com/photo-1518770660439-4636190af475?w=1200&q=80
description: Determine the right level of automation for a business process, from keeping it manual through to commissioning custom engineering. Use this tree before purchasing any automation tooling or engaging an implementation partner to ensure the solution is proportionate to the process complexity and volume. Working through each question will help you avoid over-engineering simple tasks and under-investing in high-volume critical workflows.
tags: operations, automation, process improvement, rpa, efficiency
entry: Q1

Q1: Does this process run more than 20 times per week on a recurring basis?
  hint: Count all instances of the process across the team, not just your own workload — a task done twice a day by ten people is 100 executions per week. Include seasonal peaks when estimating frequency, as automation must cope with maximum load. Processes that run fewer than 20 times per week rarely deliver enough return on investment to justify tooling beyond simple templates or checklists. Capture the current frequency in writing before proceeding, as it will be needed to build any business case for tooling expenditure.
  yes -> Q2
  no  -> [NO_AUTO]

Q2: Is the process highly standardised with clear, documented rules and no significant judgement calls?
  hint: A standardised process follows the same logical steps every time regardless of who performs it and can be fully documented as a decision flowchart or written procedure. If the process requires contextual judgement, negotiation, or the interpretation of unstructured information, automation will require far more complexity to handle exceptions. Review any existing SOPs or process maps — if none exist, the process is likely not yet ready for automation. Invest in documentation and standardisation first, then re-run this assessment in the next quarter.
  yes -> Q3
  no  -> [NO_AUTO]

Q3: Does the process involve more than 500 transactions or data records per month?
  hint: Volume is the primary driver of automation ROI — low-volume standardised processes are often better handled by a checklist or template than a full automation tool. Count all records touched by the process, including lookups, updates, and downstream notifications sent to other systems. If volume is growing month-on-month, use a 12-month forward projection rather than the current figure when making this assessment. A low-volume process that is growing rapidly may justify early automation investment to avoid a painful retrofit later.
  yes -> Q4
  no  -> [SIMPLE_AUTO]

Q4: Do all systems involved in the process expose an API or have a supported integration connector?
  hint: Most modern SaaS platforms offer REST APIs or are listed in the connector libraries of tools like Zapier, Make (formerly Integromat), or Power Automate. Legacy ERP systems, bespoke databases, or on-premise applications may lack API access, requiring screen-scraping via RPA instead. Check the vendor documentation or ask your IT team before answering — an assumption of API availability that later proves false will delay the project significantly. Request a list of supported integrations from each system owner before committing to a tooling approach.
  yes -> Q5
  no  -> [RPA]

Q5: Is the process error-tolerant, or would an automation failure cause significant business or customer harm?
  hint: Error-tolerant processes can accept occasional failures that require a manual retry — for example, internal data synchronisation or report generation with a same-day correction window. Processes that are customer-facing, financially material, or safety-relevant require robust exception handling, alerting, audit trails, and fallback procedures that go beyond basic automation tools. Consider both the frequency of likely failures and the consequences of an undetected error persisting overnight or over a weekend when assessing tolerance. If in doubt, classify the process as low error-tolerance and invest in the more robust engineering approach.
  yes -> [WORKFLOW_PLATFORM]
  no  -> [CUSTOM_ENG]

[NO_AUTO]: No Automation — Keep Manual
  color: #546E7A
  description: This process does not currently meet the frequency, standardisation, or volume thresholds that make automation cost-effective. Focus instead on documenting the process as a clear SOP, training staff consistently, and using checklists or templates to reduce error rates. Review the process again in 6–12 months — if volume or frequency grows, re-run this tree to reassess the investment case. Capturing process metrics now will make the future business case for automation much easier to build when conditions change.
  code: OPS_AUTO_MANUAL

[SIMPLE_AUTO]: Simple Rule-Based Automation (Zapier / Make)
  color: #2E7D32
  description: This process is a strong candidate for lightweight, no-code automation using a tool such as Zapier, Make, or Microsoft Power Automate. Begin by mapping the full trigger-action sequence and identifying every data field that must be passed between systems, then build and test a proof-of-concept workflow in a sandbox environment using real sample data before going live. Assign an operations owner responsible for monitoring the workflow and handling exceptions on a daily basis during the first month. Document the automation in the process register with a scheduled quarterly review to catch any drift caused by upstream system changes.
  code: OPS_AUTO_SIMPLE

[RPA]: Robotic Process Automation (RPA)
  color: #F9A825
  description: Because one or more systems lack API access, Robotic Process Automation using a tool such as UiPath, Automation Anywhere, or Power Automate Desktop is the appropriate approach. RPA bots interact with application UIs as a human would, so engage your IT team early to confirm environment stability and to ensure application updates will not silently break the bot. Commission a process discovery workshop to document every screen interaction and exception path before development begins, and allocate budget for ongoing maintenance as underlying applications evolve. Establish a bot monitoring dashboard and a clear escalation procedure so that failures are caught and handed to a human operator within one business hour.
  code: OPS_AUTO_RPA

[WORKFLOW_PLATFORM]: Full Workflow Automation Platform
  color: #1565C0
  description: This high-volume, API-connected process warrants investment in a dedicated workflow automation platform such as Camunda, Nintex, or a native workflow module within your ERP or CRM. Engage your IT and architecture teams to select a platform that integrates with your existing technology stack, meets your data residency requirements, and can scale to projected future volume. Develop a formal requirements document, run a structured vendor evaluation against at least three platforms, and allocate budget for implementation, user training, and ongoing licence costs. Define SLAs for the automated process, build in alerting for failures and SLA breaches, and schedule a post-implementation review 90 days after go-live to validate performance against the original business case.
  code: OPS_AUTO_WORKFLOW

[CUSTOM_ENG]: Requires Custom Engineering
  color: #B71C1C
  description: The combination of high volume, API connectivity, and low error-tolerance means this process requires a bespoke engineering solution rather than an off-the-shelf tool. Raise a formal technology project request with your Engineering or IT leadership team, including a detailed process specification, volume projections, SLA requirements, and a risk assessment covering the consequences of automation failure. Engage a business analyst to produce a functional specification and obtain sign-off from key stakeholders before any development begins. Plan for formal UAT, a phased rollout with parallel manual operation, and a documented rollback procedure in case the automated solution performs unexpectedly in production.
  code: OPS_AUTO_CUSTOM