Business Automation: Practical, Not Hype

Every few years a new wave of software promises to run your business while you sleep. The pitch is always the same and always slightly too broad: automate everything, eliminate human error, get your evenings back. The reality inside a fifteen-person company in Biñan or Cebu is narrower and far more interesting. Some work automates beautifully, some resists it completely, and telling the two apart before you spend money is the entire skill.

What automation actually does

It helps to be precise about the mechanism, because the marketing rarely is. Automation does not remove work from your company. It moves work from the moment of doing to the moment of deciding. You sit down once, write the rule — who approves a purchase request under a certain amount, what happens to a leave form after the manager signs it — and from then on the system applies that rule without anyone re-deciding it.

The second thing it does is convert memory into records. A great deal of small-company effort goes into remembering: who was supposed to send the supplier quotation, whether the new hire's documents came in, which request is still sitting with the boss. That effort is invisible on any org chart, but it is real, and it is the first thing to fail when the company grows or someone resigns.

So the honest promise is not "less work." It is fewer repeated decisions, and less of the day spent re-asking, re-typing and re-remembering.

Three traits of a task worth automating

A task is a genuine candidate when it has all three of these, not one or two. First, it repeats often enough to matter. Something done twice a year is not worth encoding, no matter how annoying it is; something done forty times a week almost always is. Frequency, not irritation, is the measure.

Second, the rule can be written down. If you can state on a single page what should happen in every ordinary case, the task is automatable. If explaining it takes an hour and ends with "it depends on who's asking," you do not have a rule yet — you have judgment, and judgment does not encode.

Third, it crosses people. Work that lives entirely in one person's head and stays there gains little from software. The payoff comes at handoffs: a request moving from a staff member to a supervisor to accounting. Handoffs are where things stall, get lost, or get done twice, and that is exactly where a system earns its cost.

Where the payoff is real in a small company

Apply those three traits to a typical Philippine SME and the same short list appears every time: intake, follow-through and visibility. These are unglamorous, which is precisely why they are usually still being handled by group chat and printed forms.

None of that is dramatic. It is also the part that quietly holds a growing company together, and it compounds because every one of those tasks happens many times a month.

Where automation does not pay

Four categories are worth refusing outright. Rare work is the first: annual, one-off or seasonal tasks rarely repay the setup effort, and by the time they come around again the rule has changed anyway. Second is work whose rules are still in flux. If your approval thresholds are being argued about this quarter, encoding them now means rebuilding next quarter.

Third, and most important, is a broken process. Automating a bad workflow does not fix it; it makes the bad outcome arrive faster and more consistently, and it adds a layer of software between you and the underlying problem. Fix the sequence on paper first — you will often find that half the steps existed only to compensate for a missing record, and disappear on their own.

Fourth is work that is really a relationship. Chasing a client for an overdue payment, telling someone their performance is slipping, negotiating with a supplier who has been with you for eight years: these are judgment and trust wearing the costume of a task. A system can remind you to do them. It should not do them for you.

The costs the pitch leaves out

Every automation carries three costs beyond the licence fee. The first is definition: someone in your company has to sit down and articulate how the process actually works, not how it is supposed to. This is genuinely hard and it cannot be outsourced to a vendor, because only your people know which exceptions are real.

The second is maintenance. Rules drift. Approvers change roles, statutory requirements are updated by the agencies, thresholds move. An automated process that nobody owns will keep confidently applying last year's rule long after it stopped being correct — and it will do so silently.

The third is adoption. A form that half the staff bypass because the group chat is faster is not a system; it is a second place to look. That usually means the new path has to be genuinely easier than the old one, not merely mandated.

A test to run before you buy anything

There is a cheap diagnostic that costs two weeks and no money. Pick the process that annoys you most and simply count it: how many times did it run, how many times did it come back incomplete, how many follow-ups did it take. Then try to write its rule on one page. If the count is high and the page is easy to write, automate it. If the count is low, leave it alone. If the page is impossible to write, you have found a management problem, and no platform will solve it for you.

Practical beats impressive

The companies that get real value from automation are rarely the ones that bought the most ambitious system. They are the ones that picked three or four high-frequency, rule-bound, multi-person tasks, encoded those properly, and left everything else alone until it earned its turn. That is a slower story than the pitch decks tell, but it is the one that survives contact with an actual December payroll run. Start narrow, measure honestly, and let the results — not the promises — decide what you automate next.