Simply put, it's the process of translating raw data into meaningful information, via reports or dashboards.
The demand for insightful reports and dashboards has never been greater. Organisations are all on a data journey, striving to become more data-led, and the volume and range of the data they have access to is growing rapidly year-on-year.
This creates a challenge for analysts and data scientists. Requests for new reports come thick and fast. Plus there's a constant need for information to be presented in a way that allows insights to be gleaned - and action to be taken - quickly.
The challenge is exacerbated for in-house analytics teams, as the underlying raw data is often widely dispersed, poor quality, or both.
The best dashboards are a blend of art and science, where information is both accurate and beautiful, and the key metrics leap off the page. Achieving this takes time and patience, even for the best analysts, and even with access to the latest reporting tools.
Our process for data visualisation projects has three steps:
1. Gather, cleanse and transform data
The most crucial part of any data visualisation project is data preparation. Also known as 'data wrangling', it's the process of finding, gathering and often transforming the raw data.
Where the data is highly complex, big, or coming from multiple sources, we recruit the support of Optima's data engineering team to speed up this step. Data engineers are experts in manipulating and moving data, and their toolkit is perfectly suited to extracting big data from source and transforming it into the best possible format for reporting. They also achieve this far quicker than any analyst or data scientist could.
In addition to making all the required data available, data engineers are trained to focus relentlessly on data quality. This ensures the reports are as accurate as possible.
2. Visualise data
This is where the fun starts. With the data prepared, our data scientists begin creating data visualisations that highlight the important pieces of information. This is both an analytical and a creative process, and we always take a user-centred view. This means understanding our audience and considering what they need to know to take confident action.
During this step, we take account of things like:
- Interactivity of information, including ability to click-through and slice-and-dice results
- Text vs graphical representations
- Use of colour & labels
- Signposting and cross-referencing
- Density of information
We also place great importance on allowing the data to tell its own story. This means working with the report's end-users to understand how they will consume the information, and arranging visualisations accordingly. Typically, this means building reports is an iterative process - starting with an minimum viable product (MVP), then incrementally developing the outputs in line with our end-users' feedback.
3. Automate data updates
To deliver maximum business impact, it’s important that our stakeholders are viewing up-to-date reports. This means our data visualisations must refresh and update at an optimal frequency.
Refreshing reports manually creates unwanted overheads in data teams, or unwanted costs if using a third party.
We address this problem by automating the data refreshes wherever it is economical to do so, again drawing on the specialist skills of our data engineering team. Data engineers are experts in data automation, and will ensure that the pipelines feeding data to the reports are accurate and require minimal maintenance.
