October 20, 2026 · Kairos Advisory Partners

How to tell if a process is worth automating

Five questions that separate the processes worth automating from the ones better left alone.

Read the article

Most businesses have a list of tasks someone wishes would “just happen.” A weekly report rebuilt by hand. Orders retyped from email into another system. A reconciliation that only one person knows how to do. Some of these are excellent automation candidates. Others would cost more to automate than they will ever save, or would lock a broken process into code.

The difference is usually easy to see once you ask the right questions. Here are the five we start with.

1. How often does it happen, and how long does it take?

Frequency and effort set the ceiling on what automation can save. A task that takes twenty minutes a day adds up over a year. A task that takes two hours once a quarter usually does not justify a build on time savings alone.

Be specific. “It takes forever” is not a number. Ask the person who does the work to time it for two or three cycles and write down every step, including the waiting and the rework.

2. Does it follow rules, or does it need judgment?

Software is very good at steps that can be written down: if this field says X, put it there; if the total does not match, flag it. It is weaker where the work depends on context, relationships, or judgment calls that change case by case.

Many processes are a mix. A useful approach is to automate the rules-based steps and route the judgment calls to a person, with everything they need in one place.

3. Is the process stable?

If the steps change every month, an automation will need constant changes too. That does not rule it out, but it changes the math. If the process itself is still being worked out, settle it first. Automating a process that is not settled tends to make the confusion faster, not smaller.

If the process is broken, fix the process first. Automating it only makes the problem run faster.

4. What does an error cost?

Some manual work is slow but forgiving. Other work is fast but risky: a wrong figure in an invoice, a missed compliance date, a mistyped account number. When errors are expensive or embarrassing, automation can be worth it even for a task that does not take much time, because consistent checks are one of the things software does well.

5. Are the inputs digital and reasonably consistent?

Automation needs something to work with. Data that already lives in a system, a spreadsheet, a database, or a structured export is the easiest starting point. Paper forms, free-text emails, and PDFs in many layouts can still be handled, but they take more work and more testing. Knowing this early keeps the estimate honest.

A quick way to score it

Rate each question from 1 (weak candidate) to 3 (strong candidate). There is no magic total, but the pattern tells you a lot:

• Mostly 3s: a strong candidate. Worth scoping.
• A mix: automate part of it, or simplify the process first.
• Mostly 1s: leave it alone for now, or look for an existing tool that already does it.

Signs it is not worth it yet

• The task is rare and quick, and nobody minds doing it.
• The steps are still changing or are not written down.
• Most of the work is judgment, negotiation, or conversation.
• An off-the-shelf tool you already pay for can do it with a setting or two.
• The real problem is upstream, such as bad data coming in from another team.

Start with a written description

Before anyone writes code, write the process down in plain language: who starts it, what they look at, what they decide, where the result goes, and what goes wrong. This one page is useful even if you never automate anything. It shows where time goes and where errors creep in, and it makes any conversation with a developer or consultant faster and cheaper.

When we review a process, this is where we begin. Sometimes the answer is software. Sometimes it is a better checklist, a setting in a tool you already own, or a small change in who does what. We will tell you which one it is.

When a manual process starts to hurt, there are two broad options. Buy software that already exists, or have something built for you. People who sell one option tend to argue for it. The honest answer is that both are right in different situations, and the wrong choice costs money either way.

When an existing tool is the better answer

If your process looks like the process most businesses in your field follow, there is a good chance a mature product already handles it. Accounting, payroll, scheduling, e-signatures, help desks, and customer relationship management are well served by off-the-shelf software. Those products have been tested by many customers, they get regular updates, and someone else handles security patches and support.

An existing tool is usually the right call when:

• Your process is close to standard for your industry, and you are willing to adjust small details to fit the tool.
• You need it working quickly and cannot wait for a build.
• The vendor is established, and the subscription cost is reasonable for the value.
• You already pay for a tool that can do the job, and it just has not been set up.

That last point is worth checking first. Before building anything, find out what your current software can already do.

When a custom build makes sense

Custom work earns its place when the fit is poor and the gap matters. Common signs:

• Your process is unique. The way you price, schedule, inspect, or report is part of what makes your business work, and no product fits it without awkward workarounds.
• Your systems do not connect. You have good tools that do not talk to each other, and people copy information between them by hand.
• You are paying for features you do not use. A large platform does a hundred things, and you need three.
• You have outgrown a tool. It worked when you were smaller, and now the limits, workarounds, or per-user costs get in the way.

A custom build does not have to replace your existing software. Often a good result is a small piece of software that connects the tools you already have, or fills the one gap they leave. We will recommend the off-the-shelf tool when it fits.

The simplest solution that fully solves the problem is usually the right one.

The hidden costs of buying

• Subscriptions that grow. Prices change, and plans get restructured.
• Per-seat fees. Costs rise with every person you add, including people who only need occasional access.
• Lock-in. Your data and processes live inside the vendor’s product. Leaving can be slow and expensive, and export options vary.
• Workarounds. When a tool almost fits, people fill the gaps by hand, and that time is a real cost even if it never shows up on an invoice.
• Paying for what you do not use. Higher tiers often bundle features to get the one you need.

The hidden costs of building

• Upfront cost and time. A build takes weeks, and you pay for it before you see the full benefit.
• Maintenance. Software needs care when the systems around it change: a vendor updates its interface, a password policy changes, a new field is added.
• Ownership responsibilities. If you own the code, someone has to know where it lives, how it runs, and how to change it. Good documentation and handoff matter.
• Key-person risk. If only one developer understands the system, you depend on that person. Insist on code in your own repository and plain-language documentation.

A simple way to compare

Write down what each option would cost over a realistic period, such as three years. For an existing tool, include the subscription for every user, setup, training, and the time people will still spend on workarounds. For a custom build, include the build, the time your team spends on it, and a realistic allowance for maintenance. Then compare that against what the process costs you today.

The numbers will not always decide it. Control over your data, the fit with how you work, and the risk of a vendor changing its terms all matter too. But putting both sides on paper makes the trade-offs clear.

How we approach it

When a business brings us a problem, the first question is whether it needs custom work at all. If a tool you already own can do it, we will show you how. If a well-known product fits, we will say so. When a build is the right answer, we build it around your process, document it in plain language, and train your team to run it.

AI is now built into email, word processors, spreadsheets, and many other business tools. Some of it is genuinely useful. Some of it adds cost and risk to work that ordinary software already does well. For a small business, the question is not whether to “use AI.” It is which jobs it is good at, and which it is not.

Where it tends to help

Current AI tools work well with language and loosely structured information, where a good first draft or a fast sort saves a person time, and a person still checks the result.

• Drafting and summarizing. First drafts of routine emails, summaries of long documents or meeting notes, and rewrites for a different audience. A person reviews and edits before anything goes out.
• Pulling information out of documents. Reading invoices, forms, or contracts in different layouts and extracting the fields you need, with checks on anything that looks wrong.
• Sorting and routing. Reading incoming messages or requests and suggesting a category, priority, or owner.
• Searching your own material. Finding the right answer in a large set of internal documents, policies, or past projects.
• Helping people write formulas, queries, or small scripts, with review by someone who understands the result.

Where it does not

AI tools produce answers that sound right. That is not the same as being right, and in some kinds of work the difference matters a great deal.

• Exact, rules-based steps. Calculations, reconciliations, moving data between systems, and applying fixed business rules are better handled by ordinary software, which gives the same answer every time and can be tested line by line.
• Decisions with legal, financial, or safety consequences, unless a qualified person reviews every result.
• Work where you cannot explain the result. If a customer, auditor, or regulator may ask why a decision was made, you need a process you can explain.
• A process you have not defined. AI does not fix an unclear process. It makes the unclear parts harder to see.

Most of the time, the right tool for a repetitive business process is ordinary, predictable software. AI is one option among several, not the default.

Risks to plan for

• Confident errors. AI output can be wrong in ways that are hard to spot. Build in review where mistakes matter.
• Data privacy. Know what happens to the information you put into a tool. Read the vendor’s terms on data retention and training, and keep sensitive client or employee data out of tools that are not approved for it.
• Hidden review time. If every result needs careful checking, the time saved may be smaller than it first appears. Measure it.
• Changing tools. Models and pricing change often. Avoid designs that depend on one vendor’s product behaving exactly the same way forever.

How to test before you commit

1. Pick one narrow task with a clear definition of a good result.
2. Measure how it is done today: time, error rate, and who does it.
3. Run a small trial on real examples, with a person checking every result.
4. Compare the results against your baseline, including the time spent reviewing.
5. Expand only if the trial shows a clear improvement, and keep a person responsible for the output.

How we approach it

We use AI where it fits the job and ordinary software where that is the better tool, and we will tell you which is which. Many repetitive business processes are better handled with predictable code that follows your rules exactly. Where AI does help, we design the process so a person stays in control of the result and your data stays where it belongs.

An automation project should end with a clear answer to a simple question: was it worth it? That answer is only as good as the numbers behind it, and the most important numbers are the ones you collect before the work starts.

Start with a baseline

Before anything changes, measure the process as it runs today. Ask the people who do the work to track a few cycles. Keep it simple, and write down:

• How long each cycle takes, including waiting, checking, and fixing mistakes.
• How often it runs: daily, weekly, per order, per client.
• Who does it, and roughly what an hour of their time costs the business.
• How often something goes wrong, and what it takes to fix it.
• How long it takes from start to finish, from request to result.

Two or three weeks of honest tracking is usually enough. Estimates from memory tend to be off, in both directions.

What to measure afterward

Hours saved

Compare time per cycle before and after, multiplied by how often it runs. Count the time people still spend on the process, including checking the automation’s output.

Errors and rework

Track how often mistakes happen and how long they take to fix. For some processes this matters more than time saved, because one error can cost more than a week of manual work.

Turnaround time

Measure the time from request to result. A report that used to arrive Thursday afternoon and now arrives Monday morning can change decisions, even if the hours saved are modest.

Capacity freed

Hours saved only matter if they go somewhere useful. Note what the team does with the time: more clients handled without new hires, faster responses, or work that used to be put off.

The most important numbers are the ones you collect before the work starts.

A simple payback calculation

Payback is the time it takes for savings to cover what the automation cost. The basic formula is:

Payback period = total cost of the automation ÷ net savings per month

Net savings means the value of time saved and errors avoided, minus any ongoing costs such as maintenance, hosting in your own accounts, or subscriptions the automation needs.

Hypothetical example. These figures are invented for illustration only. They are not a quote, an estimate, or a typical result.

Hypothetical payback example

Time spent on the process today	6 hours per week
Time spent after automation, including review	1 hour per week
Hours saved	5 hours per week, about 21 hours per month
Assumed cost of an hour of staff time	$40
Value of time saved	about $840 per month
Assumed ongoing costs	$40 per month
Net savings	about $800 per month
Assumed one-time cost of the automation	$8,000
Payback period	$8,000 ÷ $800 = about 10 months

Use your own numbers. Be conservative about savings and complete about costs, including the time your team spends during the project. If error reduction matters, add a line for the cost of the errors you expect to avoid, using your own records.

When to review

• After the first few weeks. Confirm it works as specified and catch problems early.
• After about three months. Compare against the baseline once the team has settled into the new way of working.
• Once a year. Check that the automation still fits the process, the systems around it still work the same way, and the savings are still real.

Also review whenever something around the process changes: a new system, a new vendor, a change in volume, or a change in who does the work.

How we approach it

We ask for a baseline during scoping, and we write down what success looks like before we build. That way, at handoff, you can see for yourself whether the work did what it was meant to do.

Who we are

Kairos Advisory Partners LLC is a Texas limited liability company based in Flower Mound, Texas. Contact: info@kairosadvisorypartners.com. Mail: Flower Mound, Texas 75022.

What we collect

When you use the contact form or book a call: your name, email, company, phone (if you give one), and your message or booking details. The companies that host and protect this site also process basic technical data, such as IP address, browser, and device type, to deliver and secure it.

How we use it

To respond to your inquiry, schedule calls, run a project if we work together, keep ordinary business records, and keep the site secure. If you give a phone number, we use it only to reply to your inquiry. We do not send marketing calls or texts.

Who we share it with

Only service providers that help us run the site and reply to you: Microsoft 365 (email and Bookings), Cloudflare (website hosting), Formspree (contact form processing), and our spam-protection provider. We may also disclose information when the law requires it. Data is processed in the United States.

Spam protection

We plan to protect the contact form with Cloudflare Turnstile, which checks signals such as IP address and browser details to tell people from bots. Cloudflare’s Turnstile Privacy Addendum describes how it handles that data.

We do not sell or share your data

We do not sell personal information, share it for cross-context behavioral advertising, or use it for targeted advertising. We do not sell sensitive data.

Cookies, analytics, and Do Not Track

This site does not use advertising or cross-site tracking cookies, and no third party tracks you across other sites through it, so we do not show a cookie banner. The site keeps one small preference in your browser for the current visit so the opening animation plays once; it is not used for tracking. If we add analytics, we will use a cookieless service and update this policy. Because we do not track you across sites, browser Do Not Track signals do not change how the site works.

State privacy laws

Based on our size and activities, comprehensive state privacy laws such as the Texas Data Privacy and Security Act and the California Consumer Privacy Act do not currently apply to us. We still honor reasonable requests to access, correct, or delete the information you sent us.

Retention and security

We keep inquiries only as long as we need them to respond and for our business records, use reasonable safeguards to protect them, and securely delete what we no longer need. If a security breach affects your information, we will notify you as the law requires. Please do not send passwords, account numbers, health information, or other sensitive data through the form.

Children

This site is not directed at children under 13, and we do not knowingly collect their information.

Your choices

To review, correct, or delete the information you have sent us, contact info@kairosadvisorypartners.com.

Changes to this policy

If this policy changes, we will post the new version on this page and update the date at the top. If a change is material, we will also note it at the top of the page.

Information only

Content on this site is general information. It is not legal, tax, accounting, investment, or financial advice. Kairos Advisory Partners LLC is not a law firm, an accounting firm, a broker-dealer, or a registered investment adviser, and we do not give advice about securities.

No guarantee of results

Results vary from business to business. Nothing on this site is a promise of specific savings, outcomes, or timelines. The terms of any engagement are set out in its Statement of Work.

Client stories

Any client story or case study we publish describes real work and is shared with the client’s written permission. We do not publish invented reviews or testimonials.

Copyright and trademarks

Site content is © Kairos Advisory Partners LLC. Kairos and the Aperture logo are trademarks of Kairos Advisory Partners LLC (™; not federally registered). Other names on this site belong to their owners, and mentioning them does not imply endorsement.

Copyright concerns

If you believe something on this site infringes your copyright, contact info@kairosadvisorypartners.com with a description of the work, where it appears on this site, and how to reach you. We will review and respond promptly.

Links to other sites

Links to third-party sites, such as LinkedIn or Microsoft Bookings, are provided for convenience. We are not responsible for their content or practices.

Limitation of liability

The site is provided as is, without warranties. To the extent the law allows, we are not liable for losses arising from use of the site.

Governing law

These terms are governed by Texas law. Venue: Denton County, Texas.

Changes

We may update these terms by posting a new version on this page with a new date.

Contact

Questions about these terms: info@kairosadvisorypartners.com.

What we do

Visible focus, full keyboard access, sufficient color contrast, support for reduced-motion and reduced-transparency settings, and labelled forms with plain-language errors.

Known limitations

Online scheduling will use Microsoft Bookings, a third-party tool. If it does not work for you, contact us and we will schedule with you directly.

Report a barrier

If something on this site does not work for you, contact info@kairosadvisorypartners.com. We aim to reply within one business day and will offer another way to get what you need.

Review

We review this site for accessibility when we make significant changes. This statement was last reviewed on the date above.