What does automation mean for small businesses? And what processes can they automate today using current technologies?

Cloud computing provides small businesses with direct access to a huge range of automation solutions. But rather than automating their existing admin processes, small businesses may be better examining whether standardised processes already exist in the cloud through specialist tailored services.

Small businesses should be wary though as costs can escalate. Return on investment (ROI) analysis should be undertaken before committing, and automation services themselves need carefully managing. Small businesses must be particularly careful that they don’t end up locked-into a spaghetti like arrangement of different cloud services with escalating costs. One advantage of automation in the cloud however is that costs are linked to usage and thus scale as the business scales.

Where small businesses have existing processes that need automating, companies like AnotherMonday and UIPath have Robotic Process Automation (RPA) products targeting small business. Similarly, Low-Code providers like Mendix, Pega and even ZoHo support point and click application design. However, the skills needed to use these products are by no means trivial and the challenge is not just building the automation but maintaining and evolving it as business changes. Our own research suggests that many small businesses across Europe choose vendors based on the proximity of support staff during adoption[1].

Is the future of small business management creating entities to help owners/managers better run their businesses?

For decades small business managers have harnessed software such as spreadsheets to manage accounting or word-processors for mail-merges. The difference today is that new software is transforming processes allowing the integration of parts of the business using RPA, Low-Code development and AI. RPA automates robotic-human-processes in the sense that they are most capable of doing things which humans find robotic – the boring, repetitive, easy, precise work. Indeed since research suggests (Willcocks and Lacity 2016) companies do not fire staff when replacing their work with RPA but instead redeploy them to more value-added work, so many employees may welcome RPA as it frees them to do the exciting, interesting and value-creating work rather than the drudgery. I believe the future of small business is about focusing on core-capabilities (the stuff the business does best and adds value) and seeking to streamline the none-core capabilities through automation, cloud services and better organisation.

One way of achieving this streamlining is through collaboration. Small businesses should collaborate to automate capabilities which are not core and build ecosystems of connected services which adds value to all. For example, small UK businesses in a sector might collaborate to produce standard automation for a problem they all face (for example specific customs-related administration post Brexit). These businesses are well placed to understand the specific requirements for the automation solution for their sector and thus develop something which works for all. Government pump-priming support for such initiatives might also be helpful.

What are the key challenges facing the small and micro business owner wanting to use more automation? Isn’t AI and Machine Learning only for big businesses?

Advanced AI and automation are already available as cloud services from the likes of Amazon AWS, Google’s Tensorflow, and Microsoft Azure. While these obviously require some skills to exploit, they do not require a PhD in AI. Indeed, innovative small companies may be better placed to create innovative AI solution as they are unencumbered by legacy IT processes and can move quickly.

There are however five major issues: Data, skills, interfacing and bias:

  • Small businesses will struggle to acquire the data to train machine learning (ML) algorithms whereas big businesses often have data-lakes from which they can use ML to extract insight and value. It is these vast data-lakes which are driving the growth in AI. Small companies may struggle to have enough data to train the algorithms to work effectively.
  • Small businesses may struggle to find the skills even with todays cloud-AI framework. My own research (Venters, Sorensen et al. 2017) suggested 71% of IT departments have lost revenue due to lack of cloud skills alone. Building an AI automation solution requires cloud and AI skills as well as those of analysing the business problem being are automated. Maintenance of the application will also be needed over the long term. Small consulting practices and agencies will likely emerge which are better placed to assist specific types of small business in this work.
  • A solution to the skills and data challenges is interfacing: connecting small company’s systems together to pool the data they have, share the cost of innovation, and build system which, by connecting companies together, can complete with large enterprise. For example, firms within a supply chain might collectively (obviously within the limits of competition law) build automation which, through the collective sharing of production and usage data, improves the production processes of all of them. Managing this interfaced relationship will obviously require work (and is the focus of my own research[2]).
  • One of the significant challenges of AI is that it can only learn from historic data which may embed pre-existing biases, or fail to reflect current (and planned future) organisational realities. Keeping an open mind, examining critically and carefully calibrating AI solutions for today’s reality is vital. Given that organisations will change and evolve this final issue requires careful skills. It is easy to build a successful solution to yesterday’s problem.

Is a bot, or other Machine Learning-based entity, a small business’s next employee?

A bot or Machine-Learning based entity is not like an employee. But, for this very reason, they need carefully managing by human employees. Bots won’t question their work; they hold no ethical compass; they cannot easily explain how they arrived at a decision; and they cannot understand the biases they might be applying. Further, they work so quickly that the problems they cause can rapidly scale out of control. For this reason, managing bots requires frequent checks to ensure they are working in the company’s best interest and delivering a clear ROI.

As significant may be that the a small-businesses’ future customers may be AI-led bots that negotiate and place orders automatically with complex market analysis. This may change the way small businesses transact their businesses and derive profit.

Will some freelancers and micro business eventually be competing against AI-based services?

It’s easy to overplay the success of AI-services compared to a human. AI suffers from the frame problem (Boden 2016) in that they cannot understand what it is to be human and make human-decisions – they can only apply robotic action or make inferences from existing limited data. The most likely emerging reality then is not that a bot or AI will replace employees, but that employees’ capabilities will be drastically improved by working alongside AI and technology. Seeing human and machine as entwined rather than opposed is a more productive perspective.

What’s somewhat more interesting is how many micro-businesses today are becoming human-intelligence based services (forms of mechanical turks[3]) provided to large AI-reliant platform companies. Consider Uber; for the “micro-business” person driving the car their “company” is entirely controlled by the Uber algorithm which allocates their piece-work contracts to drive. All traditional business functions – marketing, quality-control, advertising, sales etc. that a small mini-cab company might have provided are subsumed into the Uber platform leaving the drive-business reduced to just the intelligent driving skill. Ubers seem closer to self-driving cars than to mini-cab-firms in that the human only provides the driving-intelligence. Similarly, eBay and Amazon Marketplace have reduced retail to mere product acquisition.

What does small business automation look like in 2025?

I think many small businesses will prove highly efficient as their administrative overhead reduces allowing them to complete more easily and efficiently with larger companies. As small business automation matures so the cost of doing business will be reduced allowing more exciting and innovative businesses emerge which are both efficient and agile. The rise of Harry’s Razor for example shows how a plucky start-up harnessing online tools in marketing can build a business to complete with global multinationals.

It would be great to see government departments focused on supporting this innovation. For example, if they used next five years to focus on producing standardised APIs that allow small businesses to easily comply with “red-tape” rather than through mountains of form-filling. This would require considerable change in the way Government built and used their IT[4].

We should also not underestimate the growth of blockchain alongside automation. By providing mechanisms for establishing trust the technology could allow small businesses to collaborate more fluidly with greater safety. Smart contracts, for example, might ensure that payments happen immediately and automatically once a product is shipped.

Physical robots are also rapidly improving and may form part of physical-automation processes alongside digital ones. For example, robots, like the early Baxter, can now safely work alongside humans and are easily trained for repeated activities. If coupled with a process automation application one can envisage such robots undertaken key tasks (e.g. picking, boxing and labelling) unaided. Further out, as robotics takes the place of humans, so the economics of manufacturing may shift in favour of producing closer to the consumer rather than overseas – a move which may benefit small business.

Finally, once automation becomes accepted within the small business community, I think we will see the rise of ecosystems of small businesses working together to take on larger enterprise. By drastically reducing the transaction costs involved in collaborating there is little reason such ecosystems cannot outcompete.


Boden, M. A. (2016). AI: Its nature and future, Oxford University Press.

  • Venters, D. W., C. Sorensen and Rackspace (2017). The Cost of Cloud Expertise, Rackspace and Intel.
  • Willcocks, L. and M. C. Lacity (2016). Service Automation: Robots and the future of work. Warwickshire, UK, Steve Brookes Publishing.

[1] Polyviou, Pouloudi and Venters, (Forthcoming),” Sensemaking and proximity in cloud adoption decisions” (Working paper).

[2] https://binaryblurring.com/2017/12/04/win-of-6-million-to-research-digital-interfacing/

[3] https://en.wikipedia.org/wiki/The_Turk also https://en.wikipedia.org/wiki/Amazon_Mechanical_Turk

[4] Fishenden, J., M. Thompson and W. Venters (2018). Better Public Services: The Green Paper accompanying Better Public Services, A Manifesto. Launched at the Institute for Government on 27th March 2018.

Feature image by James Pond on Unsplash. (With thanks).

Written by Dr Will Venters