The Automation Decision Matrix

A framework for deciding what to automate, what to redesign, and what to leave human by design.

The Automation Decision Matrix is the visual centerpiece of Stage 3 of the Process First Automation methodology. It maps two dimensions, Process Stability and Driver Impact, against each other to produce a clear path recommendation for every process under evaluation.

This page is the public standalone version. Share it. Print it. Use it in your next automation planning session. The matrix belongs to your team as a decision tool whether or not you work with Axiant.

Part of the PFA Stage 3: The Automation Decision
Ungated. Free to use and share.
Printable format included
Take the Free Assessment
LOW STABILITYHIGH STABILITYHIGH IMPACTLOW IMPACTREDESIGNHigh value.Fix it first.AUTOMATEStable process.Go.PRESERVEDo not touch.INSTRUMENTAdd visibility.DECISION

The Automation Decision Matrix — Process First Automation™ — Axiant

The Automation Decision Matrix

Two inputs. Four outcomes. One clear decision for every process.

Plot any process on this matrix using two evaluations: how stable is execution, and how directly does this process connect to a measurable business outcome? The intersection determines the path.

Redesign

High value. Broken process. Fix it first.

This process connects to a meaningful driver but is too unstable or undocumented to automate reliably. Automating now locks in the dysfunction at machine speed. Redesign first, then re-evaluate.

Automate

Stable process. Strong driver connection. Go.

This process is ready. Rules are documented, execution is consistent, and the driver connection is measurable. The Process Readiness Score supports deployment. Apply technology with confidence.

Preserve

Unstable and low impact. Do not touch.

This process does not justify the investment or the risk. Preserve it as human-operated. If driver connection strengthens or execution stabilizes in the future, re-evaluate at the next PFA cycle.

Instrument

Stable but low impact. Add visibility.

Automation is not justified here. But visibility is. Add a data layer so leadership can see performance without the overhead of a full automation build. Re-evaluate when driver impact increases.

Automation Decision

The matrix above is a directional decision tool. In PFA engagements, every path recommendation is backed by the full five-dimension Process Readiness Score before being confirmed. Learn more about the PRS

Process First Automation™
Axiant — axiant.co
How to Apply It

Three steps to place any process on the matrix.

The matrix works as a quick-scan decision tool. For each process you are evaluating, answer two questions and plot the result. The intersection tells you where to start.

1
Step 01
Evaluate Driver Impact
Ask: if this process runs better or faster, which business driver moves? Revenue, margin, cycle time, or utilization? If you cannot name a driver, the process is a low driver impact candidate. If the connection is direct and measurable, it is high.
2
Step 02
Evaluate Process Stability
Ask: how consistent is execution? Are the rules documented and followed by every operator? Are exceptions rare and handled predictably? High stability means consistent, rules-based execution. Low stability means variability, tribal knowledge, or frequent exceptions.
3
Step 03
Find the Intersection
Plot the process on the matrix using your two evaluations. The cell it lands in is the path recommendation. Use the full Process Readiness Score to confirm the classification before committing to a build plan.
Worked Example

Texas Rural Hospital's claims processing workflow.

Driver Impact evaluation: the claims workflow was directly connected to cycle time and reimbursement velocity. Delays and handoff lag were measurable and operationally significant. Driver Impact: HIGH.

Stability evaluation: execution spanned multiple parties with manual transitions, inconsistent case handling, and limited end-to-end visibility. Rules were partially documented and exception handling varied by operator. Stability: LOW.

Matrix result: HIGH IMPACT + LOW STABILITY = REDESIGN path. After standardization, the workflow advanced to AUTOMATE through a coordinated five-agent architecture. The result was a 40%+ reduction in average processing time and 70%+ of cases completed without manual intervention.

Redesign(workflow standardized)Automate(after redesign was complete)
Read the full case study
What Each Path Means

Every path has a defined next action. None of them is a dead end.

The matrix gives you the path. This section explains what happens next for each one.

1
AutomateHigh Impact + High Stability

The process has earned automation. Deploy with confidence.

A process on the Automate path has cleared every qualification gate. The rules are documented. The driver connection is explicit. Execution is consistent. The data feeding the process is reliable. The human dependency is low enough that rule-based systems can handle the standard case and the defined exceptions. This process is ready.

Technology selection follows the Automate classification. The platform is chosen based on the process requirements that emerged from the evaluation, not the other way around. The build specification is written from the Stage 2 process baseline. The human architecture is defined in Stage 4 before any code is written.

Next Action:
Select technology based on process requirements. Write build specifications from the Stage 2 baseline. Proceed to Stage 4: Human Amplification.
Not This Path If:
The rules are partially documented, the exception rate is high, or the driver connection was assumed rather than formally mapped.
2
RedesignHigh Impact + Low Stability

The process matters. But it is not ready. Fix it first.

A process on the Redesign path is too important to ignore and too broken to automate. The driver connection is strong, which means the investment is justified. But the instability, undocumented rules, or fragmentation means automation applied now would lock in the dysfunction at machine speed. The result would be Automated Chaos: the faster version of the original problem.

The Redesign path begins with a return to Stage 2. The Shadow Process is surfaced in full. Rules are documented. Fragmentation is addressed. The exception logic that currently lives in institutional memory is captured in a written decision guide. After redesign, the process is re-scored. In most cases it advances to the Automate path at that point.

Next Action:
Return to Stage 2: Operational Truth. Complete process redesign. Re-score using the Process Readiness Score. Re-evaluate for the Automate path.
Not This Path If:
The process is not connected to a meaningful driver. If driver impact is low, Preserve is the more appropriate classification regardless of stability.
3
InstrumentLow Impact + High Stability

Automation is not justified. But visibility is worth the investment.

A process on the Instrument path is stable and well-run but not directly connected to a meaningful business driver at this stage. Full automation investment is not warranted. But the absence of visibility into how this process is performing, at what volume, with what error rate, is a gap that a lightweight data layer can close at a fraction of the cost of a full automation build.

The Instrument classification is frequently misunderstood as the consolation path. It is not. For a stable, low-impact process, a well-designed data layer is the correct investment. It costs less, delivers value immediately, and creates the monitoring infrastructure that would support automation quickly if the driver case develops in a future cycle.

Next Action:
Design a lightweight data layer: monitoring, reporting, and performance visibility. Assign ownership. Establish a reporting cadence. Schedule the process for re-evaluation in the next PFA cycle.
Not This Path If:
The process execution is variable or exception-heavy. Low stability combined with low driver impact points to Preserve, not Instrument.
4
PreserveLow Impact + Low Stability

This process should remain human by design. That is the right answer.

A process on the Preserve path should remain human-operated. This is not a failure condition. It is the methodology working correctly. Some processes exist precisely because they require judgment, empathy, relationship management, or contextual interpretation that a rule-based system cannot replicate reliably. Others simply do not justify the investment or the risk given their current driver connection.

The Preserve path is the one that automation vendors will never offer. It signals that the engagement is governed by discipline rather than the assumption that more automation is always better. When Axiant classifies a process as Preserve, it is not dismissing the process. It is protecting it from an investment that would not return value and protecting the client from technical debt that would compound.

Next Action:
Document the Preserve classification with rationale. Schedule the process for re-evaluation in the next PFA cycle. If driver connection strengthens or execution stabilizes, re-enter the evaluation.
Not This Path If:
The process has a strong, measurable driver connection. High driver impact always warrants either Redesign or Automate depending on stability, never Preserve.
The Quantitative Foundation

The matrix is the summary. The Process Readiness Score is the evidence behind it.

Every matrix placement in a PFA engagement is backed by a quantitative assessment. The Process Readiness Score scores every process across five dimensions: Rule Clarity, Driver Connection, Process Stability, Data Integrity, and Human Dependency. Each dimension is rated 1 to 5. The composite score determines the path, with defined thresholds.

01
Rule Clarity
Are the rules governing this process documented and understood, or do they live in institutional memory?
Drives Stability axis
02
Driver Connection
Does this process connect directly to a measurable business outcome: revenue, margin, cycle time, or utilization?
Drives Impact axis
03
Process Stability
How consistent is execution day to day and operator to operator? High variation signals fragmentation or undocumented exception handling.
Drives Stability axis
04
Data Integrity
Is the data feeding this process reliable and consistent? Low data integrity disqualifies otherwise strong automation candidates.
Modifies all paths
05
Human Dependency
Is this process deterministic and rule-based, or does it require judgment, interpretation, or relationship context?
Drives Preserve path
20-25
High Readiness
Automate path
14-19
Moderate Readiness
Automate or Instrument
8-13
Low Readiness
Redesign path
5-7
Not Ready
Preserve path
Stage 3 in Context

The Automation Decision Matrix lives inside Stage 3 of a six-stage continuous methodology.

The matrix does not stand alone. It is the executive-friendly simplification of a quantitative evaluation that happens inside Stage 3 of the Process First Automation Loop. Understanding where Stage 3 sits helps frame what needs to happen before the matrix is applied and what follows once the path is confirmed.

Stage 01
Economic Gravity
Stage 02
Operational Truth
Stage 03
The Automation Decision
You Are Here
Stage 04
Human Amplification
Stage 05
Visible Systems
Stage 06
Proof and Iteration
What Comes Before

Stages 1 and 2 produce the inputs that make the matrix work. Stage 1 (Economic Gravity) establishes the Driver Map that determines Impact. Stage 2 (Operational Truth) surfaces the process baseline that determines Stability. Without these inputs, the matrix placement is a guess.

What Comes After

Processes classified as Automate proceed to Stage 4 (Human Amplification) for architecture design. Redesign processes return to Stage 2 after the redesign work. Instrument and Preserve processes are documented and scheduled for re-evaluation in the next cycle.

Common Questions

What practitioners ask about the matrix.

Can the same process appear in different quadrants at different times?

Yes. Path classifications are not permanent. A process assigned to Redesign re-enters the evaluation after the redesign work is complete and typically advances to Automate. Preserve and Instrument processes are re-evaluated at each PFA cycle. If driver connection strengthens or execution stabilizes, the classification changes. The matrix is a snapshot of current readiness, not a permanent categorization.

What if a process scores high on Driver Impact but I am not sure about Stability?

That uncertainty is itself a signal. If you cannot confidently evaluate Stability, the process likely has undocumented rules, variable execution, or exception handling that has not been formally mapped. That profile points toward the Redesign path. The Operational Truth stage of PFA (Stage 2) is specifically designed to resolve that uncertainty before the path is confirmed.

Is the matrix the same as the Process Readiness Score?

No. The matrix is a simplified directional tool that uses two inputs: Driver Impact and Process Stability. The Process Readiness Score is a five-dimension quantitative assessment that evaluates Rule Clarity, Driver Connection, Process Stability, Data Integrity, and Human Dependency on a 1 to 5 scale per dimension. The matrix tells you roughly where a process sits. The PRS confirms it with evidence.

How should I use the matrix if I have 20 or more processes to evaluate?

Use the matrix for a first-pass triage. Plot all 20 processes using your best current assessment of their Driver Impact and Stability. This will quickly surface your highest-priority Automate and Redesign candidates and identify which Preserve and Instrument cases do not warrant further investment at this stage. Then run the full Process Readiness Score on your top five to ten candidates to confirm the classifications before building your automation roadmap.

What does the matrix say about AI automation candidates specifically?

AI automation candidates are evaluated on the same two axes as any other process. Driver Impact: does this AI application connect to a measurable business driver? Process Stability: is the underlying process stable and rule-based enough to support reliable AI execution, or does it require the kind of contextual judgment that AI systems currently handle poorly? Many AI initiatives fail because they are applied to high-variability, judgment-heavy processes. The matrix identifies those candidates as Redesign or Preserve before the investment is made.

Apply It to Your Processes

The Diagnostic produces a formal Four Paths classification for each of your process candidates.

The matrix gives you directional clarity. The PFA Diagnostic backs every classification with a full Process Readiness Score, documented rationale, and a written output you can act on. Forty-five minutes. A named practitioner. A written assessment that belongs to you whether or not an engagement follows.

Full PRS scoring for each candidate
Written Four Paths classification
Yours to keep regardless of next steps
Take the Free AssessmentOr take the DRIFT Self-Assessment first