Business automation and data science

Between labor costs, tax wedge and standardized labor costs for productivity, Italy’s SMEs are really not doing well, especially the last one. And if it’s true in the private sector, let alone in public administration…Automation makes it possible to increase competitive advantage, value added per employee and even their job satisfaction, possibly by fostering growth in real wages. Data science, or data science, fortunately has various domains that enable of business automation and digital transformation, one among all: data engineering.

Entrepreneurs have limits set by law for the amount of interns they can hire, not just for low-value-added tasks, e.g., data entry, moving data from one part to another for analysis, visualization and more.

When to do automation

In lieu of spending no less than €300/month for an intern, useful automation comes from synchronizing the database of prospects and customers (CRM, although it does much more than that) and a destination where basic analysis, called exploratory for example, takes place. The possible costs of these services are to be subtracted from the tax base for taxes (deduction, a word I do not find intuitive). So they have a double benefit if the cash flow allows for this possible expense.

Instead of having monthly reports done, which are then little read, you can create notifications that send a message to the business owner and/or executive only when certain important business metrics touch certain values, e.g., positive and/or negative anomalies.

Rather than hand-pulling data from competitor sites, third-party sites related to supply, it pays to have a continuous integration process (for data in this case) that runs code on a daily, weekly, customized basis in order to always have fresh data in a desired destination, such as spreadsheets, servers.

Rather than copying the data in supplier invoices by hand, to spreadsheets such as Google sheets or excel, and then seeing it aggregated on dashboards, you can do several things: get that data programmatically via API, if the ERP you use for invoices allows it. Or use services that offer this integration directly, such as Qonto.

When to not do automation

You can automate the creation of statistical models, to explain a business variable of interest, through specific services or through certain libraries. Which I categorically advise against even in 2024.

One can automate personnel selection by using algorithms that understand resumes. However, they can be too rigid and/or penalize those without experience, give too much weight to those with more experience. That is why I suggest surveys as a tool to evaluate candidates from a skills perspective, and then continue the evaluation with an oral interview, although it is not in all cases convenient to have that order of evaluation.

More importantly, I imagine we’ve all had at least one experience with automated customer service, first with chatbots, then with agents using linguistic statistical models, with accompanying security oversights: these things boggle the mind, better to have automations that filter and channel customers to live operators, or rely on someone like indigo.ai. There are also operators who behave like machines because of questionable business procedures, in these cases perhaps a limited agent becomes better.

There is also automation for decision-making, which on paper reduces some human cognitive limitations, and there are many approaches, such as decision trees, rule-based learning, process mining, etc. However, in 2024, I find this class of tools too immature or applicable for micro-decisions.

 

If I were in your shoes, what data-related aspects of the business would I automate? You can find out through a free call of about 30 minutes.

Leave a Comment

Your email address will not be published. Required fields are marked *

Privacy Policy