In the era of Machine Learning, there is still an uncountable number of organizations that rely heavily on systems of the 80’s – excel spreadsheets. The problem is not the tool per se, but the strategy across the organization, where data is imprisoned in different silos, full of mistakes and where data governance is nonexistent. Creating an environment where the main strategy is disconnected to daily operations and is extremely difficult to track target and goals.
The current business landscape shows there is a huge interest in using data for driving actions. The statistics say that 74% organizations want to be data-driven but only 29% of them actually achieve that [¹].
Implementing data-driven approaches could sound expensive for professionals outside the field. The usual thinking could be that software and hardware costs are outside the limits of the budget so even mention the idea in a meeting, would be automatically dismissed. That is far from reality because each approach is tailored to the size of the company and the desired level of integration between action and data.
The journey starts with an assessment of 4 steps.
1. Data Strategy
The first step of the assessment would be to establish a data strategy aligned with the main strategy of the business.The work frame would depend on the industry but usually, a REAN framework [²] fits in a majority of the cases. REAN is an acronym for
- Reach: Methods to drive online presence.
- Engage: Interaction with online elements.
- Activate: Action taken in predefined events.
- Nurture: Encourage users to come back and consume
According to analyst of Quru Analytics [²], this is the results of using the framework
In the 18 months since January 2013 Quru has made approximately €50 million in returns for its
client base using the methods described. Our goal is to continue until we have made €1 billion,
and we believe that while that’s hard, it’s possible.
The strategy would mean to write down the goals of the business into Key Performance Indicators (KPI) and target for each KPI.
|Increase unique visitors per month||Comparison visitors this month – last month||5% increase by month|
|Increase Sales 1||Monthly Revenue||£50,000 per month|
|Increase Sales 2||Profit margin||30% minimum|
2. Quality & Data Models
The selected strategy will outline the level of quality required. The outcome will lead to different approaches: leave the data as it is or generate new processes that will lead to enrich customer/user data.
These new processes are defined into two categories: transformations and additions. The former refers to make aggregations on selected data creating smaller datasets used on dashboards, tracking target among other reasons. The latter means to join dataset from different silos, bring data from outside, for example, Google Analytics or GDPR reasons.
3. Data Accessibility: Visuals & Dashboards
Organizations struggle at this point because it is possible they have the right data but they are not pushing the data to key stakeholders. Delivering the right data could make more productive employees but at the same time frees data/analytics teams to do more in-depth analysis, rather than compile quick reports.
The strategy described in the first paragraph would create clear guidelines to outline the tactical plan of creating dashboards and reports. Requirments would be discussed with stakeholders for usability purposes.
Depending on the structure is highly recommendable to have linked dashboards where data is separated by topics, department or similar. That would create different roles between users, leading to a more clear data governance between on hierarchal positions and data would be secure.
4. [Optional] Push Data: On-demand consumption
The final step – that very few organizations follow – is creating a set of alerts tailored specifically to the responsible users that would be taking action after the alert is received and understood. Frequency could vary, from days, weeks or real time. The key idea is not to drowned stakeholders in useless emails where they cannot see any value.
The alerts could create as well a more proactive environment where users are pushed to the data rather than obligated to access.