The problem with modern analytics is that it overpromises and underdelivers.
We can even quantify the disappointment:
This begs the question: why even bother with analytics?
Well, when analytics is done right, it pays back $13.01 for every dollar spent. In fact, data-driven companies outperform their competitors in almost every conceivable way. We have broached the topic of extracting value from data before, from how to set up the right data strategy to building a data-driven culture.
Here, we focus on a lesser-known topic: how to build the dream analytics team to drive the home run for your company.
When it comes to your first analytics hire, you might be tempted to go for the Stanford data scientist who has a couple of Kaggle medals under her belt, a double degree in computer science and statistics, as well as a customized Kafka MVP in her side projects portfolio.
She is undoubtedly amazing.
But if you place technical expertise at the center of your hiring decision, you’ve missed the point.
Your first hire will need to be a generalist who is willing to wear many hats. From establishing an engineering data pipeline to creating BI reports and data visualizations to deliver value, they will wield the wide-ranging skillset necessary to cover all aspects of your analytics pipeline. As the pioneer of analytics at your company, there will be little room for specialization.
Instead of searching for a master of all trades (unicorns are rare), place less emphasis on technical skills and more on their character.
The first hire will have to be a curious individual and a people person. As the culture champion, they will introduce data-driven decision-making into the company, thus drastically altering the way business decisions are made. Every change is likely to cause friction, so the amicable personality and endless enthusiasm will be the drivers that help your company to transition from opinion-based to data-driven decision-making.
If your budget permits hiring more analytics talents, opt for two data engineers and one data analyst/business intelligence expert to begin with. The 2:1 ratio reflects the intrinsic needs within data analytics teams, i.e. engineering the data infrastructure (setting up data models, configuring databases, etc.) takes more time than analyzing data (the so-called ‘hierarchy of data needs’).
A company’s success is tightly coupled with its organizational structure and processes. Sure, sometimes a hot disorganized mess does succeed, but that’s mostly down to luck.
If you want to achieve your goals, you need to organize your team members according to the organizational structure, which increases their chances of delivering the right value to the right stakeholders.
Data analytics teams appear in three different organizational units:
The team organization is a reflection of resource allocation and who they answer to.
A fully centralized team operates as a department within the company. It has its own KPIs, strategic goals aligned with company goals, and takes care of the data engineering, analytics, and science within the team.
A centralized team also works with other departments, but you can imagine the collaboration as a consultancy. The data team is called upon to consult on data matters on a project-by-project basis. The primary stakeholder for the centralized team is still the leader of that team.
There are great advantages to a centralized team:
Centralization does come at a cost, though.
Centralized teams are less bound to other departments, so a lot of business knowledge is lost. For example, a team of data scientists might develop the best recommender algorithm in a technical sense, but if that algorithm fails to help Marketing achieve their strategic goals, the endeavor was futile.
Many larger organizations have suffered as a result of this, with analytics talents failing to understand the business needs of another department. Centralized teams sometimes employ a ‘translator’ role, with a team member fully or partially devoted to translating between the business requirements of other departments and technical specifications of the data department. They are also tasked with translating the technical results to the general non-technical public.
Another organizational approach was established to avoid wasting analytic labor: the embedded or decentralized team structure.
In this organizational principle, each department within the company gets its own in-house data specialist(s). For example, Finance might get a business analyst who generates reports and metrics to measure financial success, as well as two data engineers who automate data collection from ERPs and CRMs to generate those metrics.
The advantage of decentralized teams is obvious - non-data departments get their own personalized data-driven recommendations, which help them to achieve their specific goals faster.
But the disadvantage is also clear - decentralized teams often duplicate work. The same data sources can be built and used across different departments, which wastes analytic work.
Additionally, the lack of a unified (central) data organization and governance policy causes many small differences within the data itself. You might end up with three different definitions of revenue, with as many, if not more, differing monthly numbers. This causes confusion and loss of trust in the analytic capabilities across the organization.
To get the best of both worlds, hybrid team organizations were invented. Hybrid teams have a center of excellence - a unified data strategy and governance policy which specifies all ETL processes and data assets across different departments. Not only this, their functional teams serve as in-house consultants across different departments.
This organization is much more flexible, as it covers both the technical needs of the data team and the insight needs of various departments.
As your team delivers its first successes and gains approval from leaders and peers across the company, there will be a push to get more data in front of decision-makers.
This is the chance to grow your team further by adding employees with more diversified analytic roles:
Don’t hire diverse roles if you don’t need them. You might need additional data engineers who can sort your data warehouse or some more analytics professionals who will deliver reports faster, but you might not need a data architect just yet.
Retention is rarely talked about, but it’s a cornerstone of a highly functional data analytics team.
When you’re unable to retain your team members and employees start to leave, this causes several disruptions:
Retention is a complex topic, but we can say that loyal workers either need to be happy at their current level or feel like they can achieve happiness (growth) in the future.
The latter in particular is often neglected in data teams. Data professionals don’t know what their career growth or professional development could look like at their current company, so they move to another organization to chase a better future.
Setting a clear career plan in advance prevents retention issues. When drafting your career progression plan (in line with your company’s culture and processes), keep in mind that data professionals usually seek one of three possible paths:
The various possible career progressions showcase that there are multiple paths for a satisfying career, and companies seldom formalize the different ways in which an employee can grow.
Keboola won’t be able to assist you with hiring the best person, but it will help to unburden them of technical busywork.
Keboola is the all-in-one data platform. It enables you to automate your data engineering and analytic pipeline, giving your analysts and engineers more time to think and solve hard data problems.
Did we mention that it’s free? Try Keboola’s feature-packed platform without spending a cent.