Becoming a Data Driven Company from A to Z

An analytics expert, a data analyst, uses data science for marketing, sales, and data visibility across the organization. That’s exactly my role: I help to better understand and define the environment in which we work to make the most of it. Previously, I had the chance to work for several great companies, such as Pratt & Whitney Canada and L’Oréal Canada. They had something in common: they were quite advanced in the field of data analysis, or as I like to call it, they were data driven. But when I was lucky enough to start from scratch in a young startup to create a data-driven culture, I jumped on the occasion! The adventure begins here. I will give you a brief overview of the techniques and pitfalls that we have experienced over the last few months, to become a more data driven company from A to Z.

From A to D — Demonstrate the Value of Data Analysis

When I started at the company two years ago, I was impressed to see how close the teams are to each other and to see how attentive employees are to customers. Yes, most of the decisions seemed to be based on intuition, in order to be customer-centric. In fact, as in most companies, the accounting team was already working very hard to measure and communicate the company’s financial indicators. My role was more to demonstrate the utility of non-financial data to better adjust corporate strategies. When I first arrived, my first challenge — and not the least one — was to prove that the non-financial data collected, that is to say the usage data of our software sold, that of our customers, the marketing data, the data sales, support team, etc., were essential to make informed decisions.

Step 1: Clean and create databases
In the first place, it is important to revise databases to ensure the quality of data. Even more important, by doing the exercise agile, we can ensure that teams have full confidence in the information collected.

Step 2: Demonstrate that these analyzes are important in answering some key questions
With its own data, it is possible to answer simple questions, but which bring a lot of value to the company. For example :

• Why there are sometimes slowdowns in marketing activities
• What features of the software we sell are the most popular
• Which marketing initiatives helps the most to acquire new customers

So, the goal is to demonstrate that it is possible to bring out general trends, since the discipline of business intelligence allows us to visualize the forest, not the tree, as we do when we speak only with certain customers.

From E to H — Making data ubiquitous in the business

When data analysis finally gained traction in the teams, the company began to develop in-house analytical capabilities. I was finally able to hire a BI developer to better extract, transform and democratize a variety of data. All the work done by the developer has enormously facilitated my work and has also allowed the company to really benefit from the data. With plenty of data, we were able to initiate slightly more advanced projects to bring more insights to the teams. startups should normally give a lot of flexibility to these projects to promote innovation. If they do not, try to get them onboard! Here are some examples:

1. Creating a tool that integrates with Slack messaging technology. In a competition open to all called #SlackathonMTL, we met, a team of 4 people including myself, to create a bot analyzing the performance of a website and automatically sending a short summary to users. Out of 50 teams, our project went to the finals!
2. Creating a solution to link analytical software to a CRM solution. For 2 weeks, I have formed an interdisciplinary team to work on our project, which was to imagine a response for sales and marketing teams wondering how each customer uses our product to better guide their approach during a software license renewal.

From I to M — Train the teams

Educate, educate, educate … It’s not an easy task, and yet it’s so important! I understand, somewhat to my own expense, that it is vital to decentralize the power of data analysis and understanding.

Obstacles to come

Whether it’s for special projects or our day-to-day tasks, our main effort has been to create performance indicators and corporate dashboards so teams can constantly improve. That being said, I must admit that we have fallen into some traps.

Obstacle 1: Performance indicators are not always clear

The imagination can have its limits! In order to avoid teams limiting themselves to traditional performance indicators, which are often financial, and stick to a more conservative framework, our role is to offer good suggestions and an environment that encourages exchanges with these teams, and highlight additional innovative key indicators. After a little work with some teams, we were able to establish metrics that are currently the main indicators of the company: the WAU (Weekly Active Users, or weekly active users of our software) and the Road to 1 million ( the goal of reaching one million active users for our product).

Obstacle 2: Teams do not know what to do when indicators are down

It is not always easy to know whether, first, the decline of an indicator is important enough to do anything, and second, what to do concretely to restore the situation. We learned this at our expense. How to resolve it ? By providing teams with access to more knowledge of statistics and business intelligence.

Obstacle 3: It’s too easy to be selfish
Too often, experts tend to keep the technical aspect for them. It’s easier than communicating it to everyone, and it takes less energy. Then, it ensures that the teams remain dependent on experts (I also learned that at my expense).

To avoid this, we decided to decentralize the team. In fact, recently, product data analysts are now referring to product teams and customer data analysts now report to marketing teams, who are in charge of the customer journey. As the leader of the data science community, I am more in charge of training and coalescing business intelligence and artificial intelligence strategies.

Destination: from M to Z — Where the magic happens! When data analysis meets automation

That’s really the direction we should all want to take. It is about using machine learning technologies and automating them to be able to better predict customer behaviour and thus automate our actions. By developing these artificial intelligence projects, we have almost no limits:

We are able to almost perfectly personalize all our interactions with our prospects and our customers, and remain constantly relevant.
We can do more with less. In this case, we can limit the time wasted doing paperwork and spend more time making sure people are well accompanied through their customer journey.

In conclusion

In recent months, we have made excellent progress in terms of its capacity to analyze and make data-driven decisions.

As I recently left this company to join another startup, I am very proud of the progress made and that they are now ready to move forward with super-stimulating projects that aim to maximize the use of customer data and automate certain activities through artificial intelligence.