Skip to content
Tallyfy
Developers

Introduction to DMAIC

How does DMAIC support process improvement?

Section titled “How does DMAIC support process improvement?”

When you hit a process problem, it’s tempting to jump straight to a fix. But without a structured approach, you’ll likely treat symptoms instead of root causes. DMAIC gives you a five-phase method that keeps changes data-driven and lasting.

DMAIC stands for:

  • Define
  • Measure
  • Analyze
  • Improve
  • Control

This approach evolved from manufacturing at companies like Motorola and General Electric, where reducing defects to near-zero was critical. Today it’s also used for service processes, such as client onboarding and invoice processing.

Define: what is the problem and what are the goals?

Section titled “Define: what is the problem and what are the goals?”

The Define phase pins down what you’re solving, who it affects, and what success looks like. Key activities:

  • Problem statement: State the issue concisely (e.g., “Client onboarding averages 15 days, causing dissatisfaction.”).
  • Goal statement: Set specific, measurable targets (e.g., “Reduce average onboarding to 7 days within 3 months.”).
  • Scope: Draw boundaries around the process under review.
  • People involved: Identify who’s affected by the process and the improvement effort.

This phase often produces a Project Charter summarizing these points. The team checks later decisions against it to stay within scope.

Precise problem statements matter

Section titled “Precise problem statements matter”

Avoid these pitfalls when you write the problem statement:

  • Jumping to causes: “Poor training causes delays” (that’s analysis, not definition)
  • Embedding solutions: “We need automation to fix slow processing” (that’s improvement, not definition)
  • Being too vague: “Customer service needs improvement” (which aspect? how much?)

Use the 5W1H approach instead - What’s happening? Where? When? Who’s affected? Why does it matter? How much impact? For instance: “Invoice processing errors have increased 40% over the past quarter, affecting 150+ customers monthly and causing $75K in rework costs.”

Tallyfy in the Define Phase:

  • Process Documentation: Use Tallyfy to document the current state of the process you aim to improve. Your Tallyfy template becomes the written definition of how the process runs today.
  • Identify key people: Note who’ll be affected or involved.

Measure: how does the process currently perform?

Section titled “Measure: how does the process currently perform?”

The Measure phase collects data to establish a baseline and pinpoint problem areas:

  • Identify key metrics (e.g., cycle time, error rates, customer satisfaction scores).
  • Develop a data collection plan.
  • Gather and validate the data.

Understanding variation

Section titled “Understanding variation”

Every process has variation. Ask whether yours is predictable (common cause) or unpredictable (special cause). A coffee shop has slight brewing time differences - that’s normal. A broken espresso machine creates abnormal delays. Managing by averages alone misses this distinction.

Look past the average to these measures:

  • Cycle time distribution: An average of 5 days can hide requests that take 2 days and others that take 10
  • First-pass yield: What percentage of work completes correctly without rework?
  • Process capability: Can your process consistently meet customer requirements?

If pizza delivery averages 30 minutes but ranges from 15 to 90 minutes, customers remember the 90-minute waits, and the average hides them.

Tallyfy in the Measure Phase:

  • Tallyfy Analytics: Tallyfy Analytics is an add-on that automatically copies every process and task, with its timestamps, into a database your BI tools can query. That gives you task completion times and process durations as a baseline without manual tracking, and you can query it to find bottlenecks (steps where tasks queue up). Tallyfy can also load your past processes when it sets up Analytics, so you can measure a baseline from work you’ve already done.
  • Custom Data Collection: If specific data points aren’t automatically tracked, you can add form fields to your Tallyfy tasks to collect this information as the process runs.
  • Seeing variation: Because Analytics holds a record for every process and task, your BI tool can chart the full spread of cycle times as well as the average.

Analyze: what are the root causes?

Section titled “Analyze: what are the root causes?”

With the Measure data in hand, ask “why” repeatedly until you get past the symptoms to the root cause. Common techniques include the 5 Whys and Fishbone Diagrams (covered in Simple root cause analysis techniques).

Moving past surface-level analysis

Section titled “Moving past surface-level analysis”

Teams often stop at the first plausible explanation, such as blaming slow service on understaffing. The real cause might be work arriving in unpredictable bursts, which creates artificial peaks, or a single step nobody noticed that causes most of the delay.

Elapsed time doesn’t equal effort time. A task might take 3 days but only need 30 minutes of actual work - the rest is waiting. See how to track time spent on tasks to capture real effort data.

Good analysis combines multiple angles:

  • Process analysis: Where does work get stuck? Which handoffs fail?
  • Data patterns: Do problems cluster around certain times, customers, or conditions?
  • Human factors: What makes the process hard to execute consistently?

In Out of the Crisis, W. Edwards Deming estimated that 94% of problems come from the system itself, not the people working in it. Look for the cause in the process before you blame individuals.

Tallyfy in the Analyze Phase:

  • Visualizing Bottlenecks: The Tracker view provides visibility into where work is piling up or taking longer than expected. Tallyfy Analytics lets you connect your own BI tools to analyze this data in custom dashboards.
  • Reviewing Comments: Task comments can reveal qualitative data about problems, frustrations, or recurring issues within the process.
  • Pattern recognition: Filter the Tracker by assignee, tag or folder, or query Analytics by template and date, to see whether delays cluster around certain conditions or happen at random.

Improve: how can we fix the root causes?

Section titled “Improve: how can we fix the root causes?”

Once you’ve identified root causes, the Improve phase is about brainstorming, evaluating, and testing solutions:

  • Generate a range of improvement ideas.
  • Select the most promising based on impact and feasibility.
  • Pilot solutions on a small scale to test effectiveness.

Solution design basics

Section titled “Solution design basics”

Common strategies for fixing a root cause:

Error-proofing (Poka-Yoke): Make mistakes impossible or immediately obvious. A diesel pump nozzle that’s too wide for a gasoline car’s filler neck prevents costly errors better than warning signs.

Flow optimization: Reduce handoffs, eliminate waiting, process work continuously. One way is to reorganize work from departmental batches into end-to-end case teams.

Standard work: Write down the current best way to do each task, so everyone can work the way your best operator does.

Always pilot before a full rollout, because a small test shows problems that weren’t obvious on paper while the stakes are still low.

Tallyfy in the Improve Phase:

  • Modifying Templates: Implement your proposed solutions by directly editing the Tallyfy template. The change applies to every process launched after you save, and processes already running aren’t affected.
  • Pilot Small Changes: Duplicate the template, make the pilot changes in the copy, and run a few processes from it before you apply the change to the main template.
  • Error-proofing: Mark form fields as required so a task can’t be completed until they’re filled in, and use automations to show hidden steps or route work based on earlier answers.

Control: how do we make improvements stick?

Section titled “Control: how do we make improvements stick?”

Control has three parts:

Process discipline: Standardize the new process and make it easier to follow than the old way. If people have to remember a long list of new instructions, they’ll revert under pressure, so build the improvements into the process itself.

Visual management: Show performance on a live dashboard the team can check at any time, so drift shows up while it’s still small. A monthly report shows it weeks later.

Response plans: Define what happens when performance drifts, and name who owns it. Who investigates? What triggers escalation?

Tallyfy in the Control Phase:

  • Standardized Execution: Every process launched from the updated template gives people the new steps, assignees, and required form fields as tasks, so the new method is what they see and complete.
  • Ongoing Monitoring: Continue to use Tallyfy Analytics to monitor the performance of the improved process against the baseline and desired targets.
  • Alerts & Notifications: Tallyfy emails people when they’re assigned a task and sends digest emails listing tasks that are due soon or overdue. The Tracker’s Overdue processes filter shows which processes are late.
  • Documentation is Live: The Tallyfy template itself is the documentation of the controlled process, and each edit applies to every process launched after it.
  • Accountability: Each task shows who it’s assigned to and when it’s due, so it’s clear who owns each step and which ones are late.

Quick wins vs. transformation

Section titled “Quick wins vs. transformation”

Not every problem requires a full DMAIC project. Use this guide to decide your approach:

Quick wins (days to weeks):

  • The problem sits in a single step.
  • The root cause is clear and the fix is known.
  • Few people are involved.

Rapid improvement (a few weeks):

  • The scope is focused and the problem is moderately complex.
  • A team works out the solution, using some data.

Full DMAIC (a few months):

  • The problem is complex and crosses functions, and its root causes are unknown.
  • The impact is large enough to justify a project of that length.
  • Decisions must rest on data.

For a targeted improvement, you can use single DMAIC tools, such as the 5 Whys or a small pilot, without running all five phases.

Section titled “Related articles”
Get in touch
About Tallyfy