The data pitfalls you need to watch out for on your transformation journey
- Posted on March 17, 2021
- Estimated reading time 3 minutes
The way organizations are rushing to prematurely trumpet the successes of data and analytics strategies brings to mind the horse-before-the cart idiom. Realistically, many data and analytics projects will fail due to a range of factors, notably people and processes. A cloud-based strategy undoubtedly helps organizations shed storage and computing costs, so resources can be redirected to more valuable activities. However, based on our experience working with clients globally, there are four pitfalls that need to be avoided to ensure your proverbial data and analytics cart doesn’t topple over.
- Lack of value-driven use cases
The value proposition for many data and analytics projects is poorly defined. New cloud platforms are rolled out to ingest data and generate reports (aka analytics), but use cases often end up replicating on-premises reporting solutions. This simply provides the same data perspectives, but with cost overruns. For example, a client we recently worked with had started out with aspirational use cases, but the strategy was twiddled down to the same operational reports as before. As a result, the only change mobilized was that data was moved to the cloud. This saps the business and induces innovation lethargy. Another risk of not having a clear use case proposition is that projects fizzle out because the business value is not understood by your data analysts and business stakeholders.
- Extensive centralization / decentralization
Creating a centralized repository of data and analytics capabilities can generate bottlenecks for data scientists by forcing them to wait for their data assets to be available for analysis. At the other extreme, having a complete decentralized analytics framework proliferates analytics silos across the organization, vanquishing correlated insights. An extremely decentralized framework also creates a bad data governance model, causing costs to multiply fast across the organization. Building a centralized enterprise architecture and governance model (driven by a chief data officer) makes sense, but there also needs to be flexibility for data assets serve multiple groups, such as lines of business.
- Lack of data democratization
Building a data-driven culture requires focus. Expose and liberate the data for everyone, so they can also play with the data and extract insights without waiting for outputs from your data scientists. Identifying data-driven decision frameworks within the organization helps with this. For example, at a large biotech client in Europe, Avanade devised an architecture where raw data (PII anonymized) that arrives from pharmacies and health care centers is exposed to Ninja analysts to run their data science algorithms, extracting correlations between economic indicators and prescription refills in real time.
- Insufficient investment in data talent
Investments in data assets and frameworks will be squandered if an organization fails to invest in hiring, enabling, and retaining data scientists and translators. The war for top data talent is a significant challenge, but your people truly are the difference between success and failure. Make enablement a priority, not an afterthought for your data and analytics projects. By doing this, you’ll help build a culture where your people can seamlessly take next-best-actions based on data-driven insights, which will ensure your projects have material margin impacts and differentiate your organization against competitors.