How to set up a highly Influential Analytics org?

Every organization has one analytics team or many analytics teams with different models & structures. Leaders may not align on the structure and model of the analytics team but everyone knows they need one. After building the analytics team twice, I believe that I am eligible for giving my opinion on the matter. After reading this article, It may help you in building a better analytics team.

What is Analytics?

Core Responsibility of the Analytics Teams:

Deliver actionable analytical solutions/products that enable decision-making for business ( product, growth, marketing etc) by solving business problem using data.

• Business translated into Metrics: Translate all the real-world business and product efforts into metrics that replicate the real-world behavior. Movement in these metrics must affect business in real-time.
• Identify the root cause of the broken metrics: Using data and business understanding, solve the business problem aka finding the reason behind these broken metrics. Basically, what moved these metrics in the real world historically, and also come up with new ideas that can potentially move these metrics in the future.
• Optimize the metrics aka business: Deep dive into the data and business to find the levers and start taking the metrics to new heights. Businesses can always do better and this is where analysts need to start exploring the corners of data and business to reach new heights.
• Predict the future: Show the potential futures to the business and product team by using data and what it will take us to reach there. Once we are aligned on the future, the analytics team starts executing that future with business and product.

But Analytics teams rarely are able to stick to their core responsibilities. I keep hearing complaints about the state of Analytics from my friends in various top startups in India.

The problem of Analytics:

• The Business or Product Team is doing analytics work and they are not equipped for it.
• The Analytics team is not able to deliver the work which is expected of them.
• Our analytics team is always struggling with issues such as data quality, timely delivery, non-usable dashboards, and poor analysis/insights.
• The analytics team is struggling from high attrition and they are always hiring analysts.
• The analysts don’t like their work and are unsatisfied with culture and growth.

So how do we solve it?

There are multiple reasons why analytics teams are struggling and we need to solve all of them. We need to solve these issues in a certain order: data problem, org design problem, hiring problem, processes & culture.

• Analytics is supposed to solve business problems, not solve data: For the analytics team to work, they must have consumable data and tables from which they can start working on business problem-solving. In most companies, analysts spend more than 60% of their time solving for data. Sometimes data is not available, sometimes data is not correct and sometimes no one knows which data to use.
• Who is supposed to solve the data and data ops then?: I have written an article on the central data team which is supposed to solve the data problem. In some companies, data engineering is doing this and in some companies, analysts are doing this. We need to start investing in this direction if companies want to become data first. This frees the analytics team from data problems. Analytics should involve the central data team if they get to know about any new data source and make them understand the business relevance of the data.
• Analytics is supposed to solve business problems, not generate reports and dump data: A huge misunderstanding among non-data leaders is that they think that this is the analyst's job. One of the reasons everyone thinks that is due to analyst designation being used sparingly for all the roles in the data team. Let’s stop doing that. At GetMega, I used clear designation so neither business leaders nor data operators are confused with their roles and responsibilities. All such work should move to the analytics engineer/data analyst.
• Wrong Hiring for Analytics: Most of the time companies hire people who can write SQL but can’t solve business problems later they complain about it. You have to Identify the right people for the job. Analysts are business-thinking individuals with data skills and they have a desire to solve real-world business/product problems. Business impact is the core driving factor for them. Their core skill set is problem-solving.
• Core characteristics of an Analyst: Analysts are highly independent individual that works best as internal consultant for the company. Their reporting is to the analytics org but their work is aligned with their stakeholders and themselves. We need to hire highly independent individuals who can deliver projects on their own end to end. Give them a problem and let them find the answer. They have to be mentally strong as they have to upset a lot of people and especially senior business people by showing analysis multiple times that don’t speak in favor of business people. No matter what happens they shouldn’t get biased or influenced by business or product people.
• Empower analysts by keeping them close to business but not own them: Analysts need to eat and sleep business. They can’t be added at a later stage when the problem is identified. If there is an analytics POC for the strategy team, they need to be in the loop with all the strategy teamwork, discussions, and project. The analytics team needs to proactively identify the problems and projects they think will impact the business significantly and align business on it. However that being said, they shouldn’t be reporting to business or product managers.
• Centralized Analytics with Centralized Data Org: As I have written in detail about this in my article. But let me explain this again, the moment analyst becomes part of the business or product team, they become a liability. They stop thinking like analysts and try to become product or business managers. From a career perspective, analysts see their growth as business or product managers only. Their work is influenced by the people who are running the business and product initiatives. Insights and recommendations are not trustworthy then and they start working on pure business projects instead of data-driven business problem-solving.
• Give the credit where it’s due: One of the core reasons why analyst leaves their career to become product manager or business manager is that they rarely get the credit for solving business or product. Whatever analytics works on goes via some business/product guy and they almost always fail to acknowledge the role analytics played in solving a problem that they owned. Whoever executes the project gets the credit and generally, they are a business or product people. This must change if we want hardworking and smart people to work in analytics. Also, stop blaming the analytics if some of the initiatives are not working out. Data is not wrong.
• Ability to change between different data roles: A data career is complicated compared to business/product and non-linear. Data Operators may switch from various analytics roles to data scientist roles to analytics engineer roles. All those roles are relevant for a data operator and add value to them as a data operator. If they are not able to do that, then they will get bored and leave the company. We need to stop looking at these roles as separate roles(eng vs product vs business). They all are actually data roles and data operators should be allowed to switch among them.

Despite all the above things being known, we will still struggle if we don’t understand the difference in various types of analytics. Not all analytics will be the Same. Once we understand it, we can hire the right people for the right job.

Various Analytics Team and who they are:

• Business Analytics: This team will be highly business-thinking driven and focused on actionable analytical solutions. Far away from tech. Primarily takes the help of central data team for if stuck in anything tech or downstream data.
• Product Analytics: This team will be highly product-thinking driven and focused on product/feature optimization. Work closely with tech with respect to any new feature implementation and work with the central data team with respect to upstream data and downstream data.
• Fraud/Risk Analytics: This team will be highly fraud-thinking driven and focused on proactively identifying and limiting fraudulent behavior. Far away from tech. Primarily takes the help of central data team for if stuck in anything tech or downstream data.
• Game Analytics: This team will be highly game-thinking driven and focused on product/feature optimization. Work closely with tech with respect to any new feature implementation and work with the central data team with respect to upstream data and downstream data. Data is a lot here and is messy.
• Marketing Analytics: This team will be highly business-thinking driven and focused on actionable analytical solutions. Primarily takes the help of central data team for if stuck in anything tech or downstream data. Data is a lot here and is messy.
• Fraud Investigation: This team will be highly fraud-thinking driven and focused on proactively identifying and limiting fraudulent behavior via forensic analysis. Primarily works with fraud analytics team. Minimal data skills are required but high problem-solving skills are required.

Hopefully, now you may have some clarity on how to build successful analytics teams. However this building of a successful Analytics team can’t be done in isolation, whoever is building the analytics org must understand the complete data ecosystem and the complete data org. Also, at a different stage of a company different problem exists. You can refer to this article to know more about analytics at different stages of a company and its correlation with other data teams.

In summary, set clear boundaries and ownership between various data teams about the part they are solving in the Data Ecosystem.