Automation Opportunity Matrix™: How to Prioritize Automation Projects

Oct 8, 2026 · Dan Saunders

Introduction: Several Ideas, One Available Slot

You probably have a list of things you would like to automate, and an automation prioritization matrix helps you decide which one comes first. New inquiries sometimes wait half a day for a reply. Appointment confirmations still go out by hand. Onboarding depends on whoever remembers the next step, and pulling the weekly numbers together eats an hour you would rather spend elsewhere.

You have time to fix one of these properly, not all four. So which one?

The short answer: choose a ready, manageable automation tied to the bottleneck that matters most, not simply the one that looks easiest. This guide walks through the Automation Opportunity Matrix™, AGL’s version of that method. You will compare your candidates on impact and effort, check whether each process is ready, and leave with one pilot, one owner, and one way to measure it.

What Is an Automation Prioritization Matrix?

An automation prioritization matrix is a simple grid for comparing possible automation projects. You place each candidate according to two judgments: how much it would improve an important problem, and how much work it would take to build and keep running.

Impact and effort grids are a common project-management tool, and AGL did not invent the general idea.

The Automation Opportunity Matrix™ Definition

The Automation Opportunity Matrix™ is AGL’s framework for comparing automation candidates by business impact and implementation effort, checking whether each process is ready, and choosing one practical next project.

What AGL adds is the order and the rules around the grid. You start with the bottleneck, not a brainstorm. You count the effort of maintaining an automation, not only setting it up. You check readiness before building, and you leave with one pilot instead of several projects running at once.

An Opportunity Is a Task, Not a Department

The matrix works best with bounded candidates. “Notify the assigned owner when a new inquiry arrives” is a candidate. “Automate sales” or “use AI” is not, because nobody can judge its impact or effort.

Each candidate needs four things: a trigger, an action, an intended outcome, and a named owner. If you cannot fill in all four, the idea needs more definition before it goes on the grid.

The matrix also comes after you have picked the problem. If your candidates span lead follow-up, invoicing, and onboarding at the same time, choose the most important bottleneck first. Otherwise, easy tasks from unrelated parts of the business can crowd out the work that matters.

Compare Business Impact With Implementation Effort

The matrix rests on two plain judgments: high or low impact, and high or low effort. Write a short reason beside each one. The reasons matter more than the labels, because they are what you will check against reality later.

Business Impact: The Vertical Axis

Impact runs from low at the bottom to high at the top. It asks how much a candidate would improve the bottleneck you chose. That improvement might be recurring time reclaimed, shorter delays, fewer missed handoffs, fewer errors, or a better customer experience.

Write down the evidence for the benefit you expect. A task that happens dozens of times a week has more room to matter than one that happens twice a month. Frequency alone does not prove value, though. A frequent task with no connection to the bottleneck is still low impact for this decision.

Keep reclaimed capacity separate from money. Freed hours are real, but they become cash savings only when they let you stop paying for something. They become revenue only when they go into work that earns more. The matrix compares operational impact; it does not calculate a financial return.

Implementation Effort: The Horizontal Axis

Effort runs from low on the left to high on the right. Count the whole job, not just the setup screen:

• Clarifying and preparing the process
• Setup and connections between tools
• Cleaning up the data the automation will rely on
• Testing, including the exceptions
• Training the people involved
• Human review of what the automation produces
• Monitoring and maintenance after launch

Start by checking what your existing tools can already do. A rule built inside software you already pay for often takes less effort than a new connection across several systems. Record any added cash cost on its own line. It is a real constraint, but it sits beside the effort judgment rather than inside it, and it is not an ROI calculation.

Effort is also relative. A setup that is easy for a team with a technical operations person may be hard for a solo owner. Judge every candidate against the same capacity and the same time horizon, and the comparison stays fair.

What to Collect for Each Candidate

Evidence to collect	Why it matters	Where it usually comes from
Current baseline for the bottleneck	Gives the impact judgment a starting point	A one- or two-week tally, or reports you already have
How often the task happens	Shows scale, though not value on its own	Your inbox, CRM, calendar, or a manual count
Expected improvement and its evidence	Supports the impact judgment	Recent examples of delays, misses, or errors
Tools involved and what they already do	Shows whether existing software might cover it	Current settings and the vendor’s documentation
Dependencies on data, other steps, or people	Reveals effort that is easy to miss	Walking through the process step by step
Monitoring and maintenance	The effort that continues after launch	Who checks it, how often, and what tends to break
Added cash cost	A separate constraint on the decision	New subscriptions, plan upgrades, or outside help

The Four Parts of the Automation Opportunity Matrix™

Once each candidate has an impact and an effort judgment, it lands in one of four quadrants. Each quadrant comes with a default action. Treat that action as a starting point for your judgment, not an automatic rule.

Quadrant	Business impact	Implementation effort	Default action
Quick Wins	High	Low	Consider first for a limited pilot when readiness checks pass
Plan Carefully	High	High	Break into smaller steps, resolve dependencies, and plan the work
Optional Improvements	Low	Low	Leave until the important bottleneck is being addressed
Defer or Skip	Low	High	Avoid for now; reconsider if circumstances change

Quick Wins: High Impact, Low Effort

Consider these first for a limited pilot, as long as they pass the readiness check. Suppose your bottleneck is slow lead response and your CRM can already notify the assigned owner when a form arrives. That rule may be a Quick Win. “Quick” is a relative judgment here; it does not promise a setup time or a particular return.

Plan Carefully: High Impact, High Effort

These projects matter, which is exactly why they deserve planning rather than avoidance. Break them into smaller steps, resolve dependencies first, and look for a piece that could become its own Quick Win. Automating lead qualification often sits here: the payoff could be real, but the rules, data, exceptions, and human review take serious work.

Optional Improvements: Low Impact, Low Effort

These take relatively little effort but offer limited benefit for the bottleneck you chose. A weekly summary email may be simple to set up, but if it does not change the bottleneck, it can wait. Easy work should not take the place of an important project just because it is easy to finish.

Defer or Skip: Low Impact, High Effort

Avoid these for now. A custom dashboard that pulls from several systems may be interesting to build. Even so, it adds connections and upkeep without moving the problem you chose. Priorities change, so revisit these candidates when yours do.

Check Whether the Process Is Ready

Before building anything, ask one question about each promising candidate. This is a checkpoint, not a third axis or another score.

Is the process clear, is the information reliable, and can a named owner handle human decisions, notice failures, and use a manual fallback?

If an important piece is missing, leave the candidate in its quadrant and mark it Prepare first. Its potential impact stays visible, but you do not build it until the missing requirement is fixed. Write down what that fix is.

For example, a reminder that fires when an inquiry has no recorded response only works if your team records responses reliably. If some replies happen by phone and never get logged, the reminder will nag about leads already handled. The preparation step is to agree on what counts as a response and where it gets recorded.

How to Use the Automation Prioritization Matrix

The whole method fits on one page. Work through it in order, because each step depends on the one before it.

1. Choose the bottleneck. Pick the one problem costing the most time, money, consistency, or customer goodwill right now. If you have not done this yet, the Automated Growth Roadmap™ walks you through choosing the bottleneck and fixing the process first.
2. List three to five bounded candidates. Write each one with a trigger, an action, an intended outcome, and an owner.
3. Record baselines and dependencies. Note how the bottleneck performs today. Then note what each candidate relies on, such as clean data, another step, or a specific person.
4. Assess impact and effort. Mark each candidate high or low on both, with a one-line reason. Use the same capacity and time horizon for all of them.
5. Check readiness. Leave any candidate with a missing requirement in its quadrant, mark it Prepare first, and note the fix.
6. Choose one pilot. Among the ready candidates, pick a manageable one that directly affects the bottleneck.

Breaking a Tie

Two ready candidates sometimes look equal. When that happens, prefer the one with stronger evidence, fewer dependencies, easier reversal, and more frequent feedback. A task that runs daily tells you sooner whether the change worked than one that runs monthly.

One exception matters. If a candidate is a prerequisite for others, such as reliable response tracking, treat it as the foundation, even when its direct impact looks modest.

Write a Pilot Statement

Finish with one sentence that a teammate could read and act on:

When [trigger], [action] will happen, owned by [person]. We will measure [outcome] against [baseline] and review after [a suitable period or number of cases].

Your Comparison Worksheet

Copy this table into a notebook or spreadsheet and fill in one column per candidate. Keep the bottleneck the same across every column.

Field	Candidate 1	Candidate 2	Candidate 3
Bottleneck
Candidate
Trigger and action
Owner
Current baseline
Expected impact and evidence
Implementation effort and rationale
Dependencies
Readiness or missing preparation
Quadrant
Proposed pilot measure
Review point

Example: Choosing the First Lead Follow-Up Automation

The business below is fictional, and every figure and judgment in this example is illustrative. These are teaching numbers, not AGL results or industry benchmarks.

The Situation

A small service business already uses a website form and a CRM. New inquiries wait too long for a reply because nobody is clearly responsible for each one, and follow-up is inconsistent. A typical stretch shows about 20 inquiries a week, with a median first human response of six business hours.

The bottleneck is clear: slow, uneven response to new inquiries. Five candidates come up, all aimed at that same problem.

The Assessment

Candidate	Illustrative impact	Illustrative effort	Readiness or dependency	Decision
Send an internal notification to the assigned lead owner	High: directly addresses delayed response	Low: an existing CRM rule may support it	Ready once the owner and response window are defined	First pilot
Create a reminder when an inquiry has no recorded response	High: helps prevent neglected leads	Low, once response tracking is reliable	Prepare first: define what counts as a response	Next connected candidate
Automate lead qualification and quote preparation	Potentially high	High: rules, data, exceptions, and review needed	Prepare first; customer commitments stay human	Plan Carefully
Email a weekly summary of lead counts	Low for this bottleneck: visibility alone does not speed up replies	Low, if existing reporting supports it	Ready	Optional Improvement
Build a custom multi-system dashboard	Low immediate impact on response time	High	Several integrations and maintenance dependencies	Defer or Skip

The Pilot

The internal notification is the leading candidate because it is manageable and directly addresses the delay. Before launching it, the business confirms ownership, the response window, and reliable response logging. Maria, the office coordinator, becomes the lead owner for every web inquiry, and the agreed response window is three business hours.

Reliable logging starts with a clear definition. The team agrees on what counts as a first human response: a personal call, email, or text from Maria or a teammate. It also confirms that each one is logged in the CRM with a time. If replies are not logged reliably, the pilot cannot be measured.

That logging also makes a second baseline possible. Over two typical weeks before launch, Maria records the percentage of inquiries still waiting for a first human response after the three-hour window: unanswered inquiries divided by total inquiries, multiplied by 100. A percentage keeps the two-week baseline comparable with the four-week pilot, where raw counts would not be. It is a separate measure and is never compared with the six-hour median.

Here is the pilot statement:

When a new inquiry arrives through the website form, the CRM will notify Maria, the office coordinator, who owns the response. We will compare median first human response with the six-business-hour baseline, and the percentage of inquiries still unanswered after three business hours with its pre-launch percentage. We will review after four weeks.

The business owner also sets a provisional target: a median first human response within three business hours. That target is an illustrative business choice, not a benchmark or a promise.

Two details keep the measurement honest. First, an automatic acknowledgment email is not a first human response. It can tell a lead their message arrived, but it does not count toward the response-time goal. Second, Maria records the time she spends monitoring the notification, because that upkeep is part of the effort.

Four weeks is a starting point, roughly 80 inquiries at the illustrative volume. If inquiries drop, or the period includes something unusual such as a holiday or a promotion, keep observing before judging the result. The example ends here, before any result, because its job is to show how to set up a fair test.

Why the Tempting Alternatives Wait

The reminder has strong potential, but it depends on response tracking that is not yet reliable. It stays in its quadrant, marked Prepare first, as the next connected candidate. The logging work done before the pilot is the same preparation it needs, so it can be reassessed once that tracking holds up.

Lead qualification and quote preparation could matter a great deal. It also involves rules, exceptions, and customer commitments that a person should make, so it belongs in Plan Carefully as a later, staged project.

The weekly summary is easy, which is why it is tempting. Seeing the numbers more often does not make anyone reply faster, though, so it waits. The custom dashboard adds several integrations and ongoing upkeep without touching the response delay, so it is deferred.

Common Mistakes When Using the Matrix

Most problems with the matrix come from what goes into it, not from the grid itself. Watch for these six.

• Ranking unrelated tasks together. A reporting tweak and a lead-response fix are not competing for the same goal. Choose the bottleneck first, then compare only the candidates that serve it.
• Guessing at impact. “This would really help” is a hope, not evidence. Point to recent delays, misses, or errors, even if your count is rough.
• Ignoring maintenance. An automation that takes an hour to build may still require substantial upkeep. Count the actual monitoring time before labeling it low effort.
• Automating a broken process. Automation speeds up whatever process it runs, including the parts that do not work. If ownership or steps are unclear, mark the candidate Prepare first.
• Overlooking dependencies. A reminder that relies on data nobody records will fail quietly. List what each candidate needs before you judge its effort.
• Launching several projects at once. When three changes go live together, you cannot tell which one helped or which one broke something. One pilot keeps the cause and the result connected.

How the Matrix Fits With AGL’s Other Frameworks

Each AGL framework answers a different question. The Automation Opportunity Matrix™ sits between choosing the problem and committing to a build.

Framework	Question it answers	When to use it
Automated Growth Roadmap™	What should we fix and automate first, and in what order?	Before the matrix, to choose the bottleneck and fix the process
Automation Opportunity Matrix™	Which automation candidate should we build first?	When several candidates compete to improve the same bottleneck
Automation ROI Calculator™	Is this one automation worth its full first-year cost?	After the matrix, before you commit time or money
90-Day Automation Plan™	How do we implement it over the next 12 weeks?	Once the pilot is chosen

Two other frameworks help only in specific situations. If a candidate depends on software you already use, the Tool Overload Test™ can show whether that tool is earning its place. If you want to see where a candidate fits in your wider system, the Automated Growth Stack™ maps the parts of a small-business stack.

Frequently Asked Questions

How do I prioritize automation projects?

Start with the bottleneck that matters most, then list a few specific automations that could improve it. Judge each one on business impact and full implementation effort, check whether the process is ready, and choose one pilot with an owner and a measure.

What counts as high business impact?

High impact means a meaningful improvement to the bottleneck you chose. That could be recurring time reclaimed, shorter delays, fewer missed handoffs, fewer errors, or a better customer experience. A task that happens often but does not touch the bottleneck is not high impact for this decision.

Should I choose the easiest automation first?

Only when it also helps solve the right problem. An easy automation tied to the bottleneck is a strong first pilot. An easy automation unrelated to it can wait, even though it would be quick to finish.

What if my best opportunity is not ready?

Keep it in its quadrant, mark it Prepare first, and write down what needs fixing, such as a clearer process, reliable data, or a named owner. Do not build it until that gap is closed. Often the preparation work itself becomes the right first project.

Do I need AI or an automation platform to use the matrix?

No. The matrix is a decision method, and it works on paper or in a simple spreadsheet. Many good candidates can run on tools you already have, and some problems improve with a clearer manual process before any automation is added.

Choose One Automation to Test

Pick three candidates connected to your most important bottleneck. Fill in the worksheet above, check each one for readiness, and write a single pilot statement. That sentence is the whole output of this exercise.

Before you commit time or money, use the Automation ROI Calculator™ to estimate the full cost and potential return of the candidate you chose. It separates the value of freed time from actual cash savings, which keeps the decision honest.

If the numbers hold up, turn your chosen project into a 90-day plan with a baseline, a careful build, and a clear review point.