Data Analytics – Evaluating Visualisation and BI Platforms

There is a bewildering array of data visualisation and business intelligence platforms on the market today. At Sonalake we review the players regularly and during our latest refresh we stopped counting when we reached 70! This makes the job of deciding which platform meets your needs an even tougher task.

This guide is not intended to recommend a specific platform, but rather to help you understand how best to evaluate these tools for your particular circumstances. It includes several relevant criteria which will allow you to weigh and rank platforms accordingly.

Let’s start with the core features of Data Visualisation and Business Intelligence Platforms.

Data Sources

Your data most likely comes from a variety of diverse sources, and any tool considered should easily allow end users to integrate custom or metadata for inclusion in reports and dashboards. Structured data formats will also need to be supported, and you should also consider whether a streamlined ETL flow will enable the automation of data-loading.

Storage Technologies

Traditionally visualisation tools work with relational databases. While relational database are well suited to many types of problem, there are alternate technologies that may be more appropriate to specific types of data. For example, graph oriented data, time series data, and real time streaming data are, in many cases, better served by technologies other than a relational database and this is an important consideration to bear in mind.

Visualisations

The fundamental principle of any data visualisation tool is to quickly and intuitively represent and display data in a comprehensive variety of formats. Any tool considered should allow the user to generate a wide variety of visualisations and charts that are capable of processing information from multiple datasets. As a minimum these should include, for example, histograms, barcharts, line graphs, annotated time series, pie charts, tables, gauges, geographical map views, and treemap visualisations. Customisation and design aesthetic are two more important criteria for any platform, allowing for standout visualisations and unique infographics.

Customisation can also be used to mask the workings of the tool, ensuring your customer is presented with a user interface that prominently highlights your theme and brand rather than the product itself. You may wish to use the tool to promote regular collaboration and there may be increased levels of client interaction and thus a neat, professional interface is an important consideration.

Dashboards

Data Analytics platforms allow users to create and maintain a collection of dashboards that offer tangible insights into business processes. However these tools only realise their full potential when end-users can easily and intuitively create these dashboards. If specific skill sets are required, clients may soon become frustrated by intermediary stages which can also create workflow bottlenecks.

Dashboards should be fully interactive, allowing users to drill through to individual elements and filter information using context-sensitive options to explore data. More advanced capabilities such as ad-hoc querying and multiple dashboard integration (while maintaining individual context) are also important considerations.

Reports

Another use case for these toolsets is in report production. Reports can be in PDF, Excel or within the body of an email or presentation. If this is a key requirement for your data analytics tool then you need to understand whether or not these reports can be generated. If so, is there a web based or offline tool to help you do this? If the report-generation tool is available offline, is it available for the platform that you use?

We recommend that you download and try out the software before you complete the evaluation. The questions you need to ask here are how easy is the software to use, are there good online guides to help you do what you want? Are there templates provided or do all reports have to be generated from scratch? Can common headers, footers and styles be set up and reused?

Component-based reporting enables sub-elements of reports to be drawn up, and then incorporated into a variety of reports as required, while parameterised reporting automatically prompts users to input data before the report can be generated. Linking or chaining reports allows users to drill down to another report where the context of the first is preserved. The availability of these functions within the tool is the difference between presenting your client with a qualified, interactive report or something basic and static.

Push

Having designed and built your reports and dashboards, it is likely that you will have a requirement to disseminate this data to your users. Items to consider include distribution schedules, report file formats, and distribution list management. Also query whether reports can be archived centrally and can be readily retrieved to provide historical views.

Customers may also be interested in receiving notifications when thresholds are breached. In this case, consider which notification channels are supported, for example SMS, Slack, HipChat.

Operational & Commercial Considerations

Once up and running, the tool will require either inhouse or external support and certain constraints surrounding this should be considered carefully. At some level the tool will need to be compatible with your existing IT infrastructure and allow for the growth and scale of your business.

User management support and integration with Active Directory or LDAP to support central authentication are important considerations for any analytics platform. Administrative users as well as role-based permissions will both need support, but on a deeper level options such as restrictions on specific reports, data and functions, and custom dashboards based on preassigned roles will make for a far more intuitive and efficient platform. Data level restrictions also extend to multi-tenancy instances and a useful tool will provide this functionality while simultaneously supporting audit logging of user actions.

Lastly you will need to consider the commercial model. How is the evaluated product sold? Is it available in an open source, freemium or licensed version. If licenced, what is the annual license cost and what are the annual support costs. Are there additional costs for scaling up users?

You should also consider other costs that may not be immediately transparent, for example, some vendors will charge a premium to facilitate hosting data in different geographic regions. Given the growing concerns and regulation surrounding data protection and international safe harbour privacy principles you may unwittingly incur additional costs to support a dual location data storage policy.

Finally, aside from the software, what is the availability and cost of specialists in the product to support evolution and growth of the solutions across the organisation?

Weighing it all up

We recommend that you use an evaluation matrix that lists all available features and functions and then allows you to score and weight these features according to your individual needs.

Unfortunately there’s no silver bullet when it comes to evaluating software. The evaluation itself is very useful in highlighting the features that you need most but you will likely have to compromise features against available budget. You may also take non-tangible factors into account including the quality of interactions you’ve had with the providers and whether or not you have existing experience with these tools within your organisation.

Sonalake’s experience

At Sonalake we develop and deploy a deep level of data visualisation functionality within several domains, ranging from large data analytics platforms in multinational organisations to reporting and dashboarding modules for niche software vendors. We regularly carry out detailed hands-on evaluations of leading and challenging BI and PM (Performance Management) tools on the market, both commercial and open source, and we contribute to a number of open source environments.

A domain-specialised solution of our own is VisiMetrix – optimised to enable end-to-end monitoring of network and service performance for telecommunication network operators. The solution takes performance data from multiple complex network platforms and is configured to provide end-users with visual KPI dashboards in a context that is optimised for their function, from CxO to customer care agent.

If you have any questions about implementing BI solutions or want to learn more about VisiMetrix, do not hesitate to contact us at info@sonalake.com.