What to Know About Data Before Implementing Automation
How we can ready our companies for automation? Data scientists Sarah Haq and Pablo Pardo led us through the most important things we need to know about data, first.
- Know the importance of your data
- Store your data
- Secure your data
- Organise your data
- Understand your data
Data. The very word sends shivers down many a spine. It makes us cringe as we picture computer screens packed with numbers and disconnected facts. It causes our palms to sweat as we conjure painful memories of 11th-grade algebra. But data, believe it or not, can be beautiful. And, organised correctly, it can also be stunningly useful for improving our business.
So while some people are saying data is the new oil, consider perhaps that data may actually be the new soil out of which vibrant, eye-catching flowers -- in the form of graphs and infographics -- grow and blossom. And all of us would do well to become our own data detectives, working our data (rather than being scared of it) to discover insights that can have a huge impact on our bottom lines.
In fact, becoming our own data detectives isn’t just a good idea -- it’s necessary that we start to get up-close and personal with our data as we navigate the Automation Age, where the organisation of numbers and patterns is becoming more important than ever.
In order to find out how we can ready our companies for automation and all that accompanies it, I sat down with data scientists Sarah Haq and Pablo Pardo, who led me through the most important things we need to know about our data, first. Let’s take a look. (And don’t worry -- there won’t be a test.)
Don’t make the mistake of placing more emphasis on things like your traditional KPIs to measure your customer service team’s success than they deserve. Elements like the number of tickets opened, and the rate at which those tickets are closed, are of course important. But there is an even more valuable element to be found in your customer service, Sarah says: "the wealth of data your customers generate -- particularly from their complaints."
Your customer service agents are expected to summarise their experiences with the customers they serve and share them with your company. And despite the fact that these examples are usually positive and often suffer from selection bias, Sarah points out that this is exactly the type of business data your business should be mining to discover valuable insights about your customers.
Creating and inspiring a data-driven company culture, where every employee is empowered with the knowledge, skills and tools necessary for data analysis, is key if you’re going to start collecting and analysing your company data.
But before you start fanning the flames of this data-driven company culture, be sure to make everyone in your company understands the potential for bias in data. Decisions aren’t always successful simply because they’re based on data.
Numbers and facts can be misinterpreted, and decisions made on the analysis of these misinterpretations -- or based on a mistake made in the analysis itself -- can cause challenges. So, proceed with caution. But proceed nonetheless.
Of course, none of the above will even be possible until you have somewhere to house your data. Data storage is a big deal these days, and it seems that cloud-based storage is the way to go. One of the main advantages of storing your data in the cloud is that it can then easily be accessed by anyone in your company -- from anywhere.
Having a single data storage point that’s accessible by your entire company will stop data silos being worked on by individual departments and not shared with your wider company. Such instances can lead to your teams missing out on incredibly valuable data, just because they don’t have access to it.
There are all kinds of cloud-based data warehouses out there. Sarah recommends checking out Snowflake, Amazon Redshift and Google BigQuery and Microsoft Azure SQL Data Warehouse. These data lakes, as they’re called (when all data is stored on a single interface), allow your company to share customer service insights that could prove valuable for your marketing or product management teams.
Unless you’ve been living under a rock for the past decade, you’ll know that data security is a huge issue these days. Not only that, but automating your data has the potential to introduce even more vulnerabilities into your system. The first step is to consider just how sensitive your data is -- or, how much damage could be done if it got into the wrong hands? Are you storing your customers’ credit card information? How about their medical histories? Or worse (remember the Ashley Madison data breach scandal?)?
Logically, the more sensitive your data, the more emphasis you should put on the security of that data.
Still, Sarah emphasises that even if you’re not storing information that could cause huge problems for your customers if it was leaked, data privacy should still be of utmost importance to your company. Any sensitive data should be anonymised and protected. In the EU, regulations for such steps are already in place, under the GDPR. All companies in Europe must abide by its guidelines, which put the control back in the customer’s hands over their personal data and how it’s used. Similar guidelines are likely to crop up across the globe.
The biggest problem with customer service data is the lack of understanding and mismanagement of it.
“In my experience in e-commerce businesses at least,” Sarah explained, “I have seen groups of exceptional individuals in customer-facing roles who really understand the needs of a customer, but do not have the skills or support to quantify the work they do.”
The solution? A customer service platform like Salesforce is an ideal tool for your company to start collecting and organising its data. Many of these platforms were designed to integrate with cloud-based data warehouses (we’ll get into data storage and security later). And if that weren’t enough, there are products that do the heavy data lifting for you -- downloading your data from your CRM to your chosen data warehouse automatically.
But, before you even purchase your selected CRM, it’s integral that your data is cleaned and enriched. Pablo Pardo, Data Researcher for EyeEm, explains:
“By clean, I mean make sure that the fields are properly validated, uniform formats, uniform encoding etc. By enriching, I mean to add as much extra information on top of what’s been provided by the user as you can (relevant for your problem), from internal sources (which browser the user was using, language, etc.) or external (regional holidays, zip codes, weather forecast, etc.).”
Still, the biggest challenge with customer service data is that it’s not easy to understand as a whole. It comes unstructured, often across multiple platforms, and it can feel like there’s no rhyme or reason to it. Either that, or, the task of organising and structuring it into something that could be useful for your company, feels incredibly daunting. Sound familiar?
Luckily, these days, there are plenty of solutions that can make this process not only tolerable, but pretty quick and satisfying. Check out Mixpanel, which allows you to not only gather, organise and analyse insights from your data, but also run experiments with your data. Mouseflow is another great data collection and organisation tool that allows you to see visually how your customers interact with your website.
Using such tools, your company can discover invaluable insights from your data with AI-powered functions like NLP and sentiment analysis. Indeed, Sarah points out that NLP can now read and understand large chunks of texts (which is how customer complaints generally take shape), and translate the data into comprehensive analysis.
This analysis of the long-form customer feedback allows customer service teams to “quantify the customer ‘piss-off’ factor,” -- to quote Vishal Morde, who penned an excellent article on how companies can humanise customer complaints with NLP algorithms.
As you can see, data doesn’t have to be scary or cringe-inducing. Understanding its basic principles, and arming yourself with some excellent data tools for mining, organising, and analysing, and ensuring it’s all stored in a convenient and secure database, are the first steps towards data domination status. From there, you’ll no doubt discover insights about your customers that can add a huge amount of value to your business...and your profits.
What a beautiful thing.
About our experts
Sarah Haq is a Data Scientist at upday. upday is a personalised news app that is exclusive to Samsung devices. She is working closely with a team of Data Scientists and Data Engineers to improve the news recommendations system.
Pablo Pardo is a Data Scientist at EyeEm (Berlin). He is very interested in bringing machine learning solutions to real world problems, particularly in complex scenarios involving multimodal data.
Stay in the loop
Subscribe to our email newsletter. No spam, just occasional insight from our experts.