Earlier in this series we looked at some of the history of the earliest days of the computerized automation industry. An understanding of the past and the future of this industry will be greatly aided by understanding the economics that form the fundamentals of the industry. Since automation is essentially done as a way to have capital expenditure replace costs of labor, the economics (and politics) of automation will be a big factor in any developments in the industry. Cost gravity will have a large influence on the future of automation products because any reduction of capital expenditure required to replace labor via automation shifts the balance on what gets automated. Reductions in the cost of automation will therefore deeply impact the economics of the automation industry by driving up demand1. It seems highly likely that the technology of automation will continue to get cheaper for at least a few decades due to the huge amount of effort that's being poured into developing this industry worldwide.
The high minimum cost of RPA products that are sold in long and expensive enterprise sales pipelines will very likely be the downfall of those platforms for anything but the biggest projects if organizations start to directly engage technical talent. This is because smaller projects just won't be able to absorb the cost of many months of sales salaries from external firms that is embodied in the price of one of these vendor solutions. Despite this many firms will not want to employ automation specialists on their staff for a number of reasons and they will pay a premium to vendors and consultants as a result (or perhaps just miss out on automation). Just like how many legal firms will want to benefit from developments in legal tech far fewer law firms will want to hire technologists since these job roles are substantially outside the area of comfort and expertise of most law firms.
While automation has been getting dramatically cheaper over recent years it will take some time for this cost gravity to reflect in the cost of tools for automation. Changes won't necessarily happen immediately at the point where spending internal staff time gets cheaper than buying products from vendors because of many of the political and organizational concerns that were talked about earlier. But there will be a reckoning when enough awareness exists of the multitude of free and open source general purpose automation tools combined with enough staff who are able to use them without a large amount of training. I can see this being similar to how the knowledge of how to use spreadsheet software got widespread enough the accounting industry changed dramatically. Now spreadsheet software is a widely used tool for things like financial modeling where many people in that industry have learned enough about the software to be able to automate parts of their own work without having to engage outside consultants or vendors. There's an upper limit on how much will be spent on educating staff in areas outside their core competencies due to opportunity costs. Because organizational factors and education are significant barriers to more complex automation projects it would seem that there's likely going to be space for many vendors in this space provided what they offer is sufficiently advanced enough for firms to not attempt such work on their own.
When technical literacy with open source tools become widespread this will start to put pressure on vendor products that differentiate themselves only on their technical merits. This is because the costs of the potentially long and expensive enterprise sales pipelines have to be paid by the customers. Cost aside eventually free and open source products can become substantially better than any of the paid options, showing that "you get what you pay for" is a dubious term in the software and IT industry, especially when naively taken at face value. We saw this happen in the databases space with SQL implementations when the free software options like PostgreSQL started to be significantly better than the closed source offerings from some database vendors and then eventually became the industry preferred technology. The closed source vendor options that couldn't keep up with the open source projects have mostly disappeared with the legacy holdouts mostly just surviving due to high switching costs of their customers. But just like the developments with databases there will be a long tail of organizations that keep using vendor products and new niche vendor products in the automation space will likely continue to be created over time.
Once the technical literacy and automation competency of the average professional starts to rise the deliberate frictions that are made by the various RPA vendors might start to become much more of an annoyance. For example the lack of APIs for many automation products, which ironically prevents many automation avenues, may force people away from buying off the shelf automation products if their needs get more advanced. If this becomes enough of an issue that it drives a large number of people away from platforms we may see a widespread response to this by RPA vendors that starts to provide more advanced scripting tools to end users. Giving more general purpose tools to power users seems to have been the approach taken by Microsoft's office suite when they introduced macro scripts. Eventually these initial scripting languages were themselves too simplistic and eventually general purpose programming was allowed.
I suspect what will happen in the automation space will mirror what happened with SQL, with a convergence towards a middle ground happening over time. I can see people from the RPA space getting more adept at the technical aspects of automation and the more traditional automation industry getting more adept at understanding and managing the organizational side of their work.
Given the extreme effort people in the automation industry place into value based pricing I think a drop in the cost of providing automation services will take a long time to reflect in lower pricing to the customers. By even if there was a drop in costs I think the industry would still grow due to the effects of Jevons Paradox. ↩
This post is part 4 of the "AutomationAndRPA" series: