Skip to content

Director, do you have your data in order?

AI is only as smart as the data it uses – and that’s why data quality is an increasingly important competitive factor, reminds Sofigate’s Heidi Valleala.

Dealing with a customer service bot is one of the most typical everyday situations in which a typical consumer encounters AI. For example, when you ask an airline service bot for advice when planning a trip, AI will provide the answer.

The bot processes the airline’s vast database of learned customer service patterns and best practice answers and uses them to provide you with a solution to your problem. It may not have occurred to you that you could fly to your final destination via Amsterdam instead of Frankfurt, but the bot has the capacity to take this into account.

Sometimes the bot can’t answer your question, and sometimes it gives mediocre answers. If you don’t end up buying a ticket or are frustrated by the quality of customer service, it’s easy to blame the AI. However, the fault almost always lies in the data and its format that the AI is processing. Even if a lot of data is available, there may be gaps or it may have been poorly classified and structured in the database. In this case, the AI will not be able to process it correctly, and the quality of customer service will be radically reduced.

This is a fictitious example and in reality, airline service bots can significantly improve service efficiency. However, the example illustrates well the overlooked problem of AI projects. Companies do not always understand that AI is only as intelligent as the data it uses.

An AI project is a data project

Modern technology platforms are able to bring AI capabilities to customers at a dizzying pace. However, AI cannot always be deployed directly unless the right data is available.

For a company planning to deploy AI, an analysis is usually conducted to identify the company’s capabilities and identify the gaps that need to be addressed. The analysis often reveals three things that come as a surprise to the company.

  1. Many are already on an AI journey without realising it

As many as more than half of companies have business platforms that can quickly integrate AI functions that offer at least the minimum benefits. Many are simply unaware of this and therefore not taking advantage of the investments they have already made.

While management is asking to develop the company’s AI capabilities, the same meeting may approve a project to upgrade the existing ERP system and increase the marketing budget. In reality, AI capabilities are likely to already exist in all of these technologies, but they are not known, and no one is supporting the company in analysing them.

  1. Insufficient information destroys important data

Even if AI capabilities are understood, one challenge remains: data. Insufficient information means that companies are not exploiting or storing all the data that is useful for AI. For example, customer service conversations used to be as useful to a company as old post-it notes. Conversations may have been saved for quality control or training, but after that they were thrown away like paper scraps.

Now, however, interaction data can provide AI with valuable raw material that, when processed, can help a business in many ways, such as automating responses to customer questions, analysing customer needs and providing a basis for marketing planning and product development. All of this is important raw material for AI to learn about the company’s business models.

Previously irrelevant information has consequently become worth its weight in gold. But the unfortunate reality is that if you start from scratch when collecting interaction data, it takes time to build up a usable knowledge base. In other words, if there are plans for an in-house database, you should start collecting it right away.

  1. No one owns a data strategy

Data is still used in silos within companies. Even when data is aggregated, each unit of the organisation is still only responsible for the quality of its own data and what data is stored and what is not. Sales uses customer data for its own operations, finance uses financial data, and so on. Data that seems less important may be recorded inaccurately or not collected at all.

But the power of AI lies in its ability to quickly combine and analyse a huge amount of data collected from different sources. This requires an enterprise to have a holistic, cross-silo data strategy that is centrally managed.

Do you destroy important data every day?

As data has become more important with the advent of artificial intelligence, more and more companies are now realising the problems associated with data quality. They have built data roadmaps and mapped out the steps involved in collecting data. They have also launched data cleansing projects to make the best use of existing data gaps.

Above all, they have approached data and its management as a holistic process. They understand where relevant data is generated, how it is distributed and where it is used. In this way, they ensure that data that is useful to the business is no longer destroyed on a daily basis.

The result of this process is that companies that invest in the quality of their data win. Their ability to deploy AI is simply higher than that of companies whose AI operations are based on incomplete or inaccurate data.

While AI can improve the efficiency of a company’s operations at the bare minimum, the benefits can be multiplied when the data available is of the highest quality. It is therefore first and foremost a competitive advantage. At the same time, high-quality data ensures that the company can keep pace with future developments in AI technology.

Read more:

Wanted: The three heroes of the productivity leap

AI will replace humans – and three other myths busted in 2023


Heidi Valleala is Senior Business Executive for technology projects in Sofigate’s Salesforce business. She works on AI and data projects with clients, with the goal of improving organisational efficiency and service through AI capabilities in technology. Heidi has strong experience in developing sales, customer service and marketing projects.