Skip to content

Automation21 July 2026

What to automate first in your SME, and what not to

Xora Intelligence10 min read
A stationery shop owner times how long it takes to process the orders

It is past nine. You have pulled down the shutter and sat down with the notebook where, over the month, you have been jotting down everything that "should be automated one day". There are eleven things. You start reading them and do not know which to pull on, so you go for the first, which is the one you wrote down on the day you were most burnt out. By the third you give up: you do not know whether that is the one to do first or how much work is behind it.

That is the most expensive mistake in automating a small business, and it is not a technical one. It is a mistake of order. Almost nobody fails by choosing a bad tool; they fail by starting with the task that shouts loudest instead of the one that moves the needle most. The one that shouts loudest is usually the one that annoyed you most on your last bad day, which is not the same as the one that gives you most back.

So this article will not tell you what to automate, because that depends on your business. It will give you something more useful: where to look so you can decide for yourself, how to rank the candidates by effort and impact, and what is best left exactly as it is.

Where do you start automating in an SME?

With what repeats and does not need your professional judgement. That is the whole sentence. Everything else is nuance.

A task is a good candidate when it happens many times, always follows the same steps and the answer to "what if the data comes in odd?" is a rule and not a hunch. There, a system made to measure for your business does the job as well as you do, without taking up your time. When every case is different and depends on what you decide as you look at it, it does not.

And in a small business this is not a luxury, it is survival. According to the INE Central Business Directory, on 1 January 2025, 54.4% of Spanish companies had no employees. In more than half of the businesses in this country, the paperwork is not passed to anyone because there is nobody else. When you are the bottleneck, order matters twice as much: you cannot afford to set up first what gives you least back, because the time you spend setting it up also comes out of you.

That is why it pays to look before choosing. And there are four places where something worth moving up the list almost always turns up.

What should you look at to know what to automate first?

Four places. Each one is measured with your numbers, not those of a brochure. None needs a study: it needs you to count what is already happening to you.

1. Repetitive work that does not need your judgement

Take a normal week and, every time you do a task by hand, write down three things: how many times it happens, how long it takes and who does it. Without rounding up to justify yourself or down so you do not have to see it.

At the end of the week you have a number of hours. The tasks that add up the most hours and need the least thinking are the first candidates: answering the same questions on WhatsApp, copying data from one place to another, turning an approved quote into an invoice, sorting documents. None of that sets you apart as a professional. You do it because it has to be done, not because anyone hires you for it.

2. What is lost by not replying in time

This does not appear on any spreadsheet, but it is one of the heaviest. Open the call log of the business phone and look at last week: how many missed calls there are, how many you returned and, of those you did not return, how many were from someone who never called again. Do the same with the messages that came in out of hours.

Each of those is a customer who went to whoever answered first. If the problem is that nobody knows who was supposed to return each call, perhaps you do not need a bot but a CRM for SMEs where each contact has its task. If that count comes out large, you have a strong candidate there, because automating the first reply does not save you a task: it gives you back work you were losing without seeing it.

3. What has to be redone because of a mistake

Count the times last month that something had to be done twice. A corrective invoice because of a mistyped tax ID, two appointments at the same time, a figure that turned out wrong three weeks later, an order that went out mixed up. Each one costs the time of redoing it, and some have a worse cost that you will never see in a number: the customer's face when you have to explain it.

What gets redone because of bad copy-and-paste is among the most rewarding things to automate, because the system does not lose focus at seven in the evening.

4. Getting paid sooner

This is the one most often forgotten and often the biggest. Getting paid sooner is not a cosmetic improvement: it is money that is already yours reaching your account earlier.

And here is a figure with a name attached. According to the CEPYME Late Payment Observatory, in its second-half 2025 edition, the average payment period in Spain was 80.5 days on average for 2025. The maximum set by Law 3/2004 on combating late payment is 60 days. In this country people pay late as a matter of structure, and an invoice that goes out on day 1 instead of day 20 starts that clock nineteen days earlier. That 80.5 is not your number (yours is in your accounts, in how long you take to get paid) but it tells you the problem is real and widespread.

That is why sorting out invoicing often climbs the list. And there is one more push, and it is not an opinion: Royal Decree 238/2026 of 25 March, published in the BOE (Spain's official gazette) on 31 March 2026, develops the mandatory electronic invoicing system between businesses and professionals. It does not tell you how much you will save. It tells you that the paperwork side of your business will have to be digital and in order anyway, so whoever prepares in good time arrives without surprises. If that is your case, here we explain how to automate invoices.

What should you not automate?

As important as knowing where to start is knowing what to leave alone. Some things, even if they can technically be automated, cost a lot to set up and end up worse than before. These four, leave as they are.

  • What needs your professional judgement. The diagnosis, the made-to-measure quote, the decision on whether a job can be done or not. That is why they hire you. A system can prepare the information so you decide faster, but the decision is yours and that is where your value lies.
  • The customer dealings that matter. The conversation that closes a big project, the call to calm someone who is angry, the one with the customer of a lifetime. An automatic reply in the wrong place at one of those moments does more harm than all the time it saves you on the others.
  • What changes every time. If every case is different and no two are alike, there is no pattern to teach a system. Automating something that does not repeat is set-up work with nothing at the other end.
  • What you do once a quarter. Even if it is a chore, if it happens rarely, the time to set it up and maintain it is never recovered. Volume rules: lots of small repetition is worth more than one big, rare nuisance.

The healthy split is simple: the software takes the repetitive and predictable, and you keep what needs a person thinking. Not the other way round.

How do I rank what is worth doing by effort and impact?

You now have two lists: the candidates (from the four places above) and the untouchables. Now you need to rank the first. It is done with two questions per task, nothing more.

How much impact it has. This is what you got from counting: hours a week times what an hour costs you, plus whatever you can measure in unanswered enquiries, redone work and days to get paid. It does not need to be exact; it needs to be yours. If you want to refine that number, here is the whole method for calculating the return with your own data.

How much effort it takes to set up. The most reliable sign is how many things it has to talk to. An automation that lives on its own (answers frequent questions and that is it) costs little. One that has to check your diary, look at your till system, get into your invoicing program or talk to your accountant's system costs more, because the work is in the connections, not in the part you can see.

With those two, the order sorts itself out:

  1. High impact and low effort. Start here. These are the ones that show most and take least work to set up. The first reply to customers, which is lost today, usually falls here.
  2. High impact and high effort. Worth it, but they are planned, not improvised one afternoon. Invoicing connected end to end is usually here.
  3. Low impact and low effort. They are fine, but they are not urgent. Once the ones above are running smoothly.
  4. Low impact and high effort. No. Not now and not later.

And the step everyone skips: measure before setting anything up. That week's count is the only thing that will let you know, three months from now, whether you got it right. The "before" exists only once; as soon as you automate, there is no way to get it back. Without that starting figure you will not know whether it worked, you will only believe it.

Start with one. Just one. Once you see the effect of the first on your day-to-day, expanding is easy and the next one picks itself. If you want to score each candidate with fixed criteria and a matrix, the full method is in the guide how to choose your first AI project.

Frequently asked questions

I don't even have time to make the list. Where do I start?

With the task that interrupted you today while you were working. That one. You do not need the full inventory to start looking: take that one, count it for a week (how many times, how long it takes, who does it) and you already have your first measured candidate. The long list can wait; the count for one task cannot.

Do I have to automate everything at once?

No, and that is exactly what you should not do. You pick one, set it up, measure it and, if it has worked, move on to the next. Trying to automate everything at once is the surest way to set up several things badly instead of one thing well.

What if I pick the wrong first automation?

You will find out soon, and that is why you measure beforehand. If after three months the task you automated still eats the same amount of time, you chose the wrong one. It is not a drama: you have learned which one it was not with a single test, which is exactly what starting with one instead of eleven is for.

Do I need to be a big company for it to pay off?

No. The factor is not how much you invoice or how many of you there are, it is how much of what you do repeats. A small business with a lot of repetition has more room than a big one where every job is different. That is why the calculation is done on the process, not on size.

If you want us to sort it out with you

You can do all of the above on your own, and it is written for that. You make the two lists, count for a week and the order appears.

If you would rather we look at it together, we sit down, go through your processes, see which ones repeat and what they have to connect with, and out comes an order and an idea of what each thing costs. It may turn out there are three clear tasks to start with, and it may turn out that for now it is not worth touching anything: if so, we will tell you that too.

Tell us what you need and we will say whether we can do it, how we would approach it and what it would cost. If it is not for us, we will tell you that too. Write to us and we will look at it.

Found it useful? Share it with someone who needs it

Xora IntelligenceAI consultancy and custom software in Los Barrios (Cádiz). We write about what we build for businesses in the province. If you have a question about your case, write to us.

Keep readingRelated articles

The whole blog

Bay of Gibraltar · photo by Cabeza2000 (CC BY-SA 4.0)

Shall we talkabout your business?

Tell us which tasks take up most of your time. We will tell you whether they can be automated and how we would do it. If we can't help, we will tell you that too.