How to make better automation decisions

“The craft of making farsighted choices—decisions that require long periods of deliberation, decisions whose consequences might last for years, is a strangely under-appreciated skill,” the book How We Got to Now muses. It is true that making careful decisions is hard and the stakes are high. The biggest missteps in business become notorious. Think Blockbuster vs. Netflix, or Kmart vs. WalMart.

Today, enterprise automation is the most consequential decision facing businesses and the stakes should not be underestimated. Companies can be like WalMart, embracing digital technology and automation on their path to dominance, or like Kmart, who missed every digital opportunity and floundered into oblivion. Automation is powerful, making quick wins easy, but companies that embrace instant gratification put themselves at risk of becoming the next notorious business case study.

The best companies are not built from a patchwork of quick wins and short-term gains. They are built from rich systems thinking, that equally considers the value, complexity, and the costs involved with any given process. For that reason, every process in an organization should be considered for automation with a decision model and reassessed over time.

A decision framework is needed

Despite mounting pressure to automate over the last few years, 65 per cent of organizations are still either piloting a single process or have not started their automation journey. At best, companies embrace task automation, mimicking human work with bots with the goal of efficiency. Gaining efficiency is a quick win, but automation can offer so much more.

This leads to the fundamental question that must be asked when automating every process: do we retain and increase efficiency, or do we reconsider and rebuild it? The discussion is not whether to automate, but rather what to automate and when.

For any given process, it is worth considering the following questions:

1. Why automate?
2. How much money will it cost to automate?
3. How significant is the business impact and value?
4. How complex is the process we are automating, and how does the process change?
5. What changes need to be embraced organizationally and culturally to make this work?

Once you have a sense of the answers to these questions, you can qualitatively map them to a simple chart:

The Workato automation matrix

Using this matrix, you end up with four outcomes for any given process in the organization:

1. High-value and low cost, total automation: This is the no-brainer, slam-dunk option. It is cheap and powerful and there is no reason not to proceed.
2. High-value and high cost, consider automation: If it is expensive but valuable, it may be worth doing, but it also may be worth postponing.
3. Low-value and low cost, semi-automation: If it is cheap to automate but also not particularly impactful, it still may be worth a low-effort automation plan.
4. Low-value and high cost, avoid automation: Stay away. Do not waste organizational resources on something that is expensive with very little output value.

This is a useful framework, but it is easy to think about these outcomes in this theoretical context with a basic 2x2 quadrant. In light of this we will take a look at a few real-world examples. We will consider what quadrant the real-world process sits in today and where it should sit in a perfect scenario.

New hire onboarding

New employees start with a ton of energy to make an early impact. Phrases like, “I’m so excited to be here” or “I’m ready to get cracking” are common in the first few days on the job. It is up to the employer to empower new hires to be productive from day one, but to their own detriment, most companies fall short. For example, Gallup noted that only 12 per cent of employees think their company handled onboarding well.

The average onboarding experience involves waiting around or requesting access to key systems. Cloud applications and automation platforms have made total onboarding automation easier than ever and companies really have no excuse to not have a fully automated onboarding process. Employee onboarding, or offboarding, is one of those key processes that belongs squarely in the ‘total automation’ quadrant.

Back office in a bank

A friend of mine recently moved and updated his address with his bank via a website portal. He later stopped for gas, tried to use his credit card, and was declined at the zip code authorization. Frustrated, he called his bank and discovered that the web portal update only applied to his checking account, but not his credit card.

This is the perfect example of a semi-automated process. The information only moved across a few systems at the bank, but not every system. Should this be fully automated? We are talking about customer experience here, so it is certainly valuable, although it also might be complex and expensive. Many financial institutions have old-school legacy systems, which makes automation difficult. In this case, it falls in the ‘consider automation’ quadrant, where the deliberation is centered around which links in the chain of process are to be automated first and in what sequence. You can make excuses for delaying this type of automation, but eventually it should be done.

Customer success

Imagine a B2B tech company that is proud to have built an automated process that generates a usage report for every customer, emailed to the related customer success manager (CSM) every month. This seems like a useful account health tool, but when the CSM has to join an emergency call with a customer 25 days after the last report, they are vulnerable to being surprised by any changes that occurred in those prior three weeks.

This is a great example of what is currently a semi-automated process employed by a large number of organizations that should be fully automated. Customer retention is high value, perhaps the highest value of all. This business needs to give CSMs the ability to refresh and check the data any time. Weekly emails would still be helpful, but the ability to have account health at the click of a button is a life saver for any customer-facing role.

Time to upgrade our automation decisions

An article published in The New Yorker observed that: “One of the paradoxes of life is that our big decisions are often less calculated than our small ones are. We agonize over what to stream on Netflix, then let TV shows persuade us to move to New York. Buying a new laptop may involve weeks of internet research, but the deliberations behind a life-changing breakup could consist of a few bottles of wine.”

Maybe the average company’s automation decision-making is not quite as bad as some of the big life choices the article observes, but they could certainly be improved. If we start by thinking about them with a simple evaluation quadrant, we will certainly get closer to making automation something that differentiates our organization, rather than something we see as a quick win.

Today, organizational automation decisions are just another strangely under-appreciated skill, however as more companies begin to automate more processes, it is not likely to stay that way forever.