The 100% delusion: about success or failure in automation

You often hear: “XY is a job that a machine can never do.” I always cringe slightly, because if I’ve learned anything, it’s that you should be very careful with the word “never”. Especially when it comes to technology. On the other hand, I often come across people who dream of a fully automated business and stand in their own way with this view.

(Reading time: 5 minutes)

“It doesn’t scale like that”

I don’t know how many times I’ve heard this phrase in conversations with venture capitalists in recent months. We were “chalked up” to the fact that we have not yet fully automated all of our company’s Accounto processes and yet we are already launching the product on the market.

If you analyze our experience figures in detail, you can see that it scales very well. Phenomenally well, in fact. Even with manual work.

The analogy underlying this way of thinking is: “What is not fully automated cannot scale quickly and highly”. This is also true for established business models and technologies. As soon as new concepts and new technology are used, you are really on the wrong track.

Many good product ideas are therefore currently falling victim to this 100% automation mania. They are discarded before they are even started, so to speak.

Creating a new product with new technologies and paradigms is a balancing act

However, when you use new technology to realize a new business model, a new product, different rules apply. You have to navigate the triangle, so to speak, of technological feasibility, economic viability and social acceptance.

All three points are directly dependent on each other. The trick now is to shape a product in such a way that the triangle, virtually speaking, maintains the balance.

Because if you only do what is technologically feasible, then you will have a problem with acceptance and, above all, with profitability. If you only do what is economically feasible, you will never achieve what is technologically feasible and if you only do what is socially acceptable, you will never build a disruptive product.

Doing one thing but not the other

So if you want to build an automated business, have the courage to fill in the gaps. The last 20% are only interesting to solve once you have, so to speak, slain the 80% that is really manual work.

What you should definitely have is a clear idea of the technological path. Prototypes are easy to realize. They give you a feeling of how quickly you can make progress. If you don’t have the technological vision, you might as well play Russian roulette. The killer will come at some point. But maybe not. No basis for building a company. There are plenty of other stumbling blocks that can mess up a start-up.

An economically driven, self-regulating prioritization of the development backlog

Data collection across all processes of such a company is absolutely crucial. We have organized this at Accounto so that we work with the most up-to-date data for every sprint. What we automate next in our processes with machine learning is dictated by profitability, so to speak. These are small, simple investment calculations: What are the costs of manual work? What are the investments in the area? When will we break even in this sub-process? The gains are still huge at the moment, but they decrease with each additional process.

At first glance, this sometimes leads to bizarre decisions that we would never make without the database. Small, unspectacular things are automated at an early stage. This approach ensures that we take the most direct economic route to full automation.

“We fix it in the mix”

The key here is to be better than all the alternatives available on the market with a product that is not fully automated. This is usually not achieved by using new technology per se, but by introducing a paradigm shift to the market.

These paradigm shifts must be “in line”, so to speak, with the underlying technological development. The paradigm shift we are driving (and proclaiming) is that we see it as the logical next step in accounting software that users no longer have to post themselves. The underlying technological development is the rapid progress in the field of machine learning.

As soon as it is economically viable, i.e. a product can be built that offers the customer fundamental advantages without being more expensive, it makes sense to launch it. All the productivity gains, and therefore margin gains, can be realized as soon as you have a cruising altitude. And thus actually only gain strategically.

I think these mechanisms are universal for building and launching new, disruptive products. Fortunately for us, there are a few VCs who understand this.

Target 100% automation

And yet: despite all the courage to leave gaps and manual processes, the goal must remain 100% automation. This is the only logical consequence based on physics. And since technology is constantly improving, the potential for optimization will never run out. It just slows down because it is economically driven and becomes more expensive because it becomes more complex.

However, what many start-up entrepreneurs today also disregard is the social acceptance of a component. I talk to people who want to automate and digitize support all the time. Of course they do. Technology-supported support processes make sense and help to reduce costs. But at the moment, the majority of people still want to deal with people when they have problems and questions. On the one hand, this is because the electronic alternatives are all still very labor-intensive for customers. On the other hand, there is a psychological component that we cannot yet replace.

As an entrepreneur in 2016, this is not an issue for me for the time being. I look at the competition, and for me the competition is not the few digital players in the fiduciary sector, but the thousands of traditional ones, and realize that we can do support fundamentally better than the existing providers with relatively little effort. It’s pretty simple. And cheap.

We therefore invest a relevant part of our marketing budget in really great, not yet automated, support and thus invest in one of the most crucial areas of our customer experience. Automating this now would be pretty stupid. At the same time, however, I’m quite sure that it’s only a matter of time before we implement 100% automation in this area too. When the time is right. And it will be – sooner or later.