7 February 2018
In Part 1 we started our comparison of the software packages Power BI, Qlikview and Tableau. In the first part, we covered these criteria:
In this part (Part 2) we are going to continue with:
• How easily can I share reports and dashboards with colleagues?
• How much training will my team need to be able to get started with it?
• How much does the tool cost?
• Verdict: which tool should I choose for dashboarding and reporting?
This fourth criterion looks at how easily you can distribute information in your organisation. The three tools each have their own “publication environment” (Power BI Service / Report Server, Tableau Server / Online, Qlikview Management Console) – which are often paid-for services. Whether or not you use that publication service will depend heavily on the level of maturity of your organisation. One extreme is to make a work file in the tool and sending it to your colleague by e-mail. The other extreme is to publish that work file in the publication environment, set it to automatically update it and to automatically inform your colleagues of any updates so they can see the content. We make a distinction between:
“Your message is only as good as your ability to share it”. And, more broadly, if your information is user-friendly it will have the most impact. How do our three tools perform?
Power BI generally distinguishes between two places to work in: Power BI Desktop – where you make your reports – and Power BI Service – where you upload your reports and make them into dashboards. Power BI has recently also added an “on-premises” solution (stored on a special server): Power BI Report Server. We regard both solutions as one and the same (considering their shared functions) and for the purposes of this comparison will proceed to refer to them as the “Power BI Service”. It isn’t unheard of to make the explicit distinction between reports and dashboards, but it can be a let-down if you are presuming that you can transfer a report directly one-for-one into a dashboard. Microsoft approaches a dashboard as a collection of multiple reports – where you “pin” a few visuals (like Pinterest) of each report onto a dashboard. These dashboards are made up of tiles. So each tile is a visual from one of your reports. And yes, you can pin whole pages from reports as a “live page” – where the whole page becomes one tile – but the result is not exactly the kind of report you would have built in Power BI Desktop. The scaling is different and the display is more compact. It would be premature to write off the Power BI Service on the basis, as its other functionalities are extremely user-friendly. You can edit content at the same time using an online editor. Publishing a local report to the online environment is a smooth process. Pinning visuals on dashboards is wonderfully easy too. Users get a personal corner of their own and can share workspaces with each other. You can effortlessly share dashboards with everyone in your domain (e.g. @cmotions.nl) thanks to an address book connection. If you are still too attached to your old PowerPoint file, there are options to export into PowerPoint. The result leaves much to be desired. As standard, you can only share dashboards. If you want to share the whole set of data set, report and dashboard, Microsoft now offers special “apps”: packages of dashboard material that you can share easily.
Qlikview dashboards are built within the developer’s desktop environment. A dashboard can include multiple tabs, and each tab can accommodate multiple visuals. You load data and the app (report or dashboard in .qvd format) in the Qlikview Management Console (MC). Here you easily pre-programme the tasks and, if you wish, make them conditional on one another. Running tasks in MC creates a new online dashboard that is available 24/7 for the end users. You can make prior arrangements with users about what time data will be refreshed. The user can see on each dashboard when it was last updated. As well as being able to access the dashboard via the web, there is also the option to distribute the report, or part of it, in PDF format. It is also possible to e-mail the report in .qvd format, but the recipient needs to have the required access to view the latest figures. You can save selections from dashboards as favourites to use in the future. These “bookmarks” can be shared with colleagues, e.g. with your department.
Tableau generally offers two places to work too: Tableau Desktop and their online environment. In the online environment, Tableau distinguishes between Tableau Server (you organise the hosting yourself “on-premises”) and Tableau Online where they take care of it all for you. Because Server and Online have an extremely similar user experience, we are considering them as the same environment for the purposes of this comparison. This user experience is undoubtedly the most highly developed. In Tableau you also start by making the dashboard components: “slides” which are combined later on a dashboard. You can then publish these on the online environment. Tableau distinguishes between high and low sites (environments where your data and visualisations appear), projects, workbooks and views. You can subsequently use views as your dashboards. Publishing is safe and very straightforward, and authorisations can be modified in minute detail for each individual user. Once in the online environment, sharing with other users is a matter of a couple of clicks and you can also open a compact version of Tableau Desktop. Aside from the visualisations, you can also publish the data sources themselves, which can be useful to make new visualisations from them later. Finally, there are extensive possibilities for publishing snapshots of dashboards from the online environment to a PDF or an image etc.
Dashboards are increasingly being viewed on mobile devices. Therefore it is important for the tooling to provide for both good-quality display of your publication and an easy process behind it.
Microsoft built the “mobile grid” into Power BI at any early stage, allowing you to adapt your dashboard directly for mobile use. The great thing about that is that you don’t even have to completely build a desktop-only report, and instead it provides components from your desktop version as building blocks in your mobile “grid”. With the Power BI app in the Apple Store / Play Store, it is straightforward to also display dashboards on tablets and phones.
Dashboards are Qlikview extremely accessible, 24/7 by both web and mobile. The good “responsive” display on both desktop and mobile is achieved using parameters at the design stage. Because publishing to the web is kept separate from designing in desktop and loading data, the continuity of the display is protected. There is an official for iOS (‘Qlikview mobile’), but not for Android.
When it comes to mobile publishing, Tableau again is far more highly developed than Power BI and Qlikview. It is striking that you can even edit a dashboard on a tablet. The desktop functionalities are transferred well to the mobile experience. Mobile applications are available for iOS and Android.
Making a dashboard is actually a two-step process: data processing and then visualisation. As far as data processing is concerned, in our view the data visualisation tool is often expected to do too much of the work. Whilst a database management system such as Microsoft SQL Server or Oracle is good specifically at data processing, a large part of the data processing still wrongly ends up in the visualisation tool. This is inconvenient as it results in slow dashboards, including when updating data. Now let’s look at how the three tools do in terms of updating data.
Especially in the case of business intelligence, you need to update data regularly after publication. If you have a simple data connection to an unconnected file, Power BI will retrieve the new data from the source file with ease. There is one condition: you must not change the structure of your source file. If your connection is more complex, then you can set up a “gateway” to your live database from within Power BI. In Power BI Service it is simple to create a “schedule” to stipulate when the dashboard should update itself automatically without you having to lift a finger.
In Qlikview it is advised to keep the data load (ETL) separate from the visualisation. These can be done in two different .qvd documents which are referred to as “apps”. The ETL shows the links to data and can be used with a variety of functions to model data into a final table that is saved in Qlikview format. This final table is then scanned into the visual app. Updating reports on the web is done via a separate Management Console. In the console you can easily activate or deactivate processes. You can follow logs when error notifications appear and on the basis of these you can restore and reactivate the subprocesses (for loading the ETL and visual app). You can switch off reloads or make them conditional on other (Qlikview etc.) processes, and it is simple to configure this. If you, as a developer, want to test the report on desktop, you can do so easily using links to the necessary databases in the ETL section and save the final table in a fixed place in the testing environment.
Updating data in Tableau is very straightforward. If you only work in Desktop, it is easy to refresh an underlying source. If you work in an online variant (server/online), you can simply get it to “push” your local data to the online environment. If the data is your own (i.e. not public data from Google Analytics etc.), Tableau can serve as a “proxy” between your own servers in the department and the cloud. Refreshing data using the “Refresh from source” function is user-friendly for beginners but also offers experts batch functionality to refresh online dashboards from a command line (or any other scheduling tool).
Each of the three tools offers sharing options in a different way. Workbooks are often made in a desktop version and then published to actual dashboards on a publication server. Power BI is also convenient in how it integrates with the Office package in your organisation (address book you can share) but ought to be more flexible in transferring reports to dashboard. Qlikview supports both data processing and visualisation very well. According to Qlik (the supplier), future developments of this tool which continue the artificial intelligence trend. Tableau is without doubt the most highly-advanced tool. This is evidenced by the “clean” design of the environment, but also by the details: searchability, favourites, custom start pages, tracking edits, undo/redo and making/managing “subscriptions” to particular dashboards. The degree of specification when refreshing data (incremental, cumulative, custom scripts) also puts Tableau in the lead.
Another important consideration is the extra investment in training required to use the tool. I.e.:
Ease of use is the number one prerequisite for good software. It is obvious that all three tools had this in mind. Although here, again, there are differences between them.
Users of other Microsoft products are sometimes surprised by the simplicity of Power BI. The builders have succeeded in keeping the work environment clear. Generally speaking you have three tables (worksheet, data environment, identifying relationships between data sets). In your worksheet, you have constant access in the fields to your possible visuals. The data environment has a superficial layer in which you make basic changes (field names, data types, sorting) and a deeper layer (Advanced Query Editor) in which you do more complex operations (pivoting, transforming, filtering data out). For novice users, the fastest route is to simply try it out. Power BI will suggest a visual when selecting one of the columns of your data, or automatically show a light-grey display of the eventual result when you click on a visual.
Before you make your first dashboard in Qlikview, you can already get results after just two hours of introductory training. Qlikview is very fast at visualisation, you can create a simple dashboard in less than fifteen minutes. To develop ETL in more detail and for calculated variables, you will require more in-depth knowledge. A good basis for a developer is to gain a knowledge of SQL.
You can easily find your own way in Tableau too. This is the most intuitive of this set of three tools. In Qlikview and Power BI you still see the options that aren’t any use, but Tableau intelligently makes the options you don’t need disappear. You end up having an impressive dashboard in no time and will often think of what visual suits best after you have already clicked on a visual. Tableau often advises you on what to do in this situation.
The quality of a community goes more or less hand in hand with the number of active users the community has. This point isn’t exactly a level playing field, considering the age of each tool.
More than in any other respect, it is when it comes to the community that it is most clear that Power BI is watching its competition. It would appear that it’s better to copy something well than think of something original badly. The way in which Tableau designed up its community is clearly recognisable in the Power BI community. It clearly distinguishes between good and less good answers and the winning solution to the issue is highlighted. What’s more, Power BI and Tableau both have their own company moderators monitoring the community, which is great for maintaining the level of quality of the information.
Qlikview has a global online community too, with a log of problems or questions and solutions, often proposed by community members. This means you can get answers to the questions asked in a very short time. Materials that introduce you as a new developer to Qlikview are also available here, along with a “Resource Library”.
With more than a million posts, the Tableau community is the unquestioned leader and the example to the rest. As with Power BI, here users also share example workbooks so that another user with a question can instantly try out the solutions for themselves. Online forums are supplemented with User groups, in which users can even schedule meetings in real-life. A large number of on-demand webinars are provided on the Tableau website.
None of these tools are difficult to start using. With Tableau you can get cracking a bit more quickly than with the two alternatives. When there’s something you don’t know, each tool has a community you can consult. Once again Tableau is in the lead – not least because of its maturity. The overlap between the Power BI’s DAX programming language and Excel formulas is convenient but doesn’t match up to the logical language used in Tableau and Qlikview. You can learn to drive well by getting to grips with any of these tools, but only with Tableau will you actually get your driving licence.
The final, but often deal-breaking criterion, we are going to look at is how much the tool costs. The tools each have a different payment model. Let’s zoom in on each of these models and look at your “total cost of ownership” to use the tool:
How easily can you take an exploratory look at what you get, and when (and for what) you will have to pay? That is often one of the first questions that come up when selecting a data visualisation tool. The three competitors have structured their payment models in strikingly different ways.
Power BI offers a free version of Desktop (the local work environment). Anyone can download and continue using this version free of charge. As a customer, you will only start paying when you want to start using Power BI Service – the publication service of Power BI. The package of Desktop and Service, known as Power BI Pro, offers great convenience when it comes to sharing and maintaining your dashboard. It is surprising how many functions and data connections you already have access to with only the Desktop version.
QlikView describes its pricing model online as ‘a combination of server-, user-, document- and application-based licensing’. But to find out specific prices you have to contact the organisation first. However, there is a free developers’ version available for desktop.
With Tableau, you pay – after a trial period of 30 days – per user, per month for the basic Desktop version or the more advanced Professional Edition. In you additionally require the functionality to share, then you pay per user, per month to use Tableau Server. If you don’t want your own server but want to outsource maintenance to Tableau, this is possible for a small supplement in addition to the price for Server: Tableau Online. But Tableau also has a free version: Tableau Public. As with Power BI, this is free of charge for unlimited use too. However, it doesn’t give you access (which you do get in Power BI) to live connections to your own (SQL / Oracle) databases. You also can’t save your work files locally. The degree that many organisations handle is often so sensitive that Tableau Public is therefore much less suitable than Power BI Desktop.
Obviously it is a crucial concern to consider what it will actually cost. To be more specific, the total costs of ownership (TCO) of such a tool. This is a difficult thing to calculate because you also have to include the costs associated with incidents, and the work required for maintenance and continuity.
At first glance, the prices for Power BI Pro are the ones that capture your imagination: USD 9.99 per user, per month for up to max. 10 GB storage per user – i.e. when rounded up, USD 120 per user per year. That sounds cheap, but the competitors are convinced that the resulting complexity with Microsoft when you try to get “more” out of the products ultimately ends up being more expensive. This could include extra costs for training, extra time to make a dashboard and support in the event of incidents due to unforeseen situations. Plus there is a supplement to pay if you end up saving more than 10 GB of data per user. If you buy a hundred licences, you get 100 x 10 GB = 1,000 GB available to you, which may potentially all be used by a single user. If you make clear internal arrangements and buy a larger number of licences, you can avoid exceeding this limit.
Qlikview doesn’t give anything away about the prices for the Enterprise edition on its official website. Our enquiry to a reseller didn’t uncover any more information either.
With Tableau, the basic Desktop version costs USD 35 and the Professional Edition costs USD 70 per user, per month; which equates to USD 420 / USD 840 per year. With the Professional Edition you get more advanced data connections and publication options: which are essential when you want to use direct connections to “live data” in your databases. If this is required, you pay USD 35 per user, per month to use Tableau Server functionality, a further USD 420 per year. The Professional Edition & Server costs annually, when rounded up, from USD 1,260 per user, per year – not including any additional costs for your own server. If you choose Tableau Online (so not on your own server but all on Tableau’s servers), than you pay a further EUR 42 per user, per month in addition to the Professional Edition. In that scenario, the total for Professional Edition & Online becomes USD 1,344 per user, per year. Finally, the Tableau Reader – to only view the dashboards – is free of charge.
It is impossible to compare these three tools like-for-like with objective measures. The final costs remain to be seen, based on their ease of use, how much additional training is required and how many incidents occur. After all, the definition of what is “cheap” depends on an organisation’s needs. In an objective comparison between the direct costs for licences, Power BI is the clear winner. Qlikview can’t even be included in the comparison on costs (because of its lack of transparency). Tableau claims to be a better player than Power BI etc. when it comes to its level of convenience and scalability. By the same token, in our experience we are yet to see with Microsoft’s visualisation tool any of the signs of extreme complexity after upscaling – except for being slightly less easy to use. On the contrary, the user-friendliness of Power BI beats many of Microsoft’s earlier products. When Tableau costs ten times more than a comparable product from Microsoft, it looks like a risk worth taking.
Which tool should I choose? You’ve already read the conclusion between the lines: it depends. More important than picking a single winner is identifying how much weight you give to each criterion and then examining that criterion in detail. You do this firstly by looking at what needs have to be met in your organisation. This comparison is a criterion-by-criterion guide and not an exclusive comparison to whittle them down to a single winner. These tools choose all three directions along the spectrum between “BI” and “stand-alone data visualisation” and therefore also the degree of data preparation required. It is reassuring that all three of the tools have a highly respectable basis level and are future-proof: it is in fact the choices of Microsoft, Tableau and Qlik that dictate the ideal environment for each of these tools. Of course, we will combine the separate scores below. So, most importantly, you should concentrate most on the criterion that has the greatest weighting for you.
21 February 2022
Plan your calendar free and make sure you have plenty of party clothes in your closet,... read more
7 April 2021
In these boring lockdown-times we are all desperately looking for ways to still interact with our... read more
1 April 2021
Almost every day we go online we encounter recommender systems; if you are listening to your... read more