working through problems
The demand for self-service analytics has soared, but with increased usage comes the need for clearer communications PHOTO: lunde2012

When my clients ask me what a successful company-wide self-service analytics collaboration looks like, I tell them to imagine a group of jazz musicians: The individual players all contribute song notes from their own instruments and styles, but it’s the collective sound of everyone playing together that creates the musical magic that the audience hears onstage.

Now, extend that musical collaboration metaphor to your organization and it’s easy to see how the self-service analytics contributions made by your individuals and teams can raise the level of your company’s business intelligence to new heights.

Demand for Self-Service Analytics Is Soaring

What’s more, thanks to the soaring demand for the data needed to power increasingly sophisticated machine learning initiatives, the popularity of self-service analytics software has soared. Just last month, in fact, Small Business Trends highlighted a Gartner report projecting that the global analytics and business intelligence market will grow to $22.8 billion by the end of 2020.

However, at the same time enthusiasm for self-service analytics is taking off, demand for the more exacting data hygiene standards necessary to run machine learning models is also on the rise. That’s because machine learning models are far more exacting in their data requirements than other self-service analytics tasks such as creating dashboards.

Data Hygiene Is Vital

Because machine learning models ‘learn’ based on the data collected, a high level of organizational commitment to data hygiene is vital. How can organizations capitalize on the growth and democratization of self-service analytics software without compromising their overall corporate data collection and deployment strategies?

Fortunately, there are plenty of solutions that highlight data hygiene while also supporting advanced analysis and graphing methods. These solutions include open source tools like Neo4j, a graph database, as well as technical programming languages like R and Python that feature data visualization libraries.

Self-Service Analytics Tools Abound

Among other options are cloud solutions that manage machine learning algorithms, while providing visualization for decision trees, clustering and other configuration results from data sets. Many excellent tools exist, including Microsoft AzureGoogle Tensorflow and Apache Singa.

Yet, even with self-service tools designed to maximize teamwork and minimize implementation problems, analytics professionals still need to be on the lookout for miscommunications that could impact how results get interpreted and reporting errors are managed across their organizations.

Using Communication to Avoid False Scope

One such risk lies in the way communication takes place within an organization’s team structure. For example, when different team members view the same data for different purposes, each perspective brings different data and support needs when they are interacting with the reports.

In such cases, so-called false scope — where a tactic or activity can seem like a connection to a larger strategy but turns out not to be — can occur. For instance, cross-functional teams might identify a data relationship but not know upon first assessment whether their results are minor or point to a more significant business problem. False scope makes it tricky to know which actions will actually enhance a product or service or inform a process within the organization.

Beware of Shadow IT Solutions

Another problem that can hamper analytics communication occurs when tech support becomes so self-contained that solutions aimed at supporting individuals don’t get showcased or shared across the entire enterprise. Such ‘shadow IT’ solutions may create short-term fixes but don’t establish long-term protocols for sharing best practices for problem-solving and productivity when preparing analyses.

Depending on a model’s purpose and how a department is aligned, identifying gaps in communicating analytical insights can take some time to wind through the entire modeling process.

Analyzing Information Sharing Processes

Self-service data preparation, for example, can take weeks or even months. The tasks usually include gathering requirements, modeling data, creating reports and arranging for distribution of those reports.

Practitioners should look to processes where information is shared easily to see where the information flow is working well and where there are potential gaps. In addition, studying each iteration of a task or process can reveal whether a communication concern is uniquely temporary or symptomatic of an ongoing process problem.

Begin With Your Dashboard Solutions

If your analytics communication process seems fragile or undermined by gaps, start by analyzing your existing dashboard solutions, as discussed in this earlier CMSWire post. Focusing your discussion on what’s working and what isn’t for a specific dashboard can be a great way to rally your team around key objectives. Then, by expanding that conversation to include other areas of your organization, you can counter any unduly narrow perspectives on how data can be modeled.      

Such a discussion can also reveal how to get past false scope. Is the data aligning with your underlying needs? Will it solve the problems you care about? And remember, too, that dashboard discussions can also address shadow IT concerns and encourage the formulation of best practices.

Educating Your Organization

As self-service analytics continue to gain in popularity and achieve even more widespread adoption, the need for organizations to coordinate their usage will become even more important.

Ultimately, success for your organization will come in the form of balancing the power of your self-service analytics instruments with the training and teamwork you provide to maximize their coordinated usage.