Power BI – Custom Visuals, the Great Abdication
Napoleon's abdication

Guest Author – Narius P

I’d like to explore the idea that Microsoft has abdicated a key responsibility within the Power BI ecosystem and left it to the developer community; I refer to their approach to Custom Visuals.

If you are familiar with Power BI then you will know that while it comes with a few built-in visuals, bar/column charts, pie / donut charts, matrices and tables, cards and maps etc, it is still woefully lacking in more advanced visuals; especially considering that Power BI has been around since 2013 (originally as Excel add-ins).

In all that time the most interesting of the built-in visuals, the Decomposition Tree, has only recently been added. There were some moments of excitement along the way, for example the Sand Dance visual but it never made it into the base product.

Instead Microsoft opted for “marketplace” style approach (currently called AppSource), pitching their own custom visuals for Power BI against creations from third-party developers, allowing “the community” to expand the visual toolbox incrementally. Conceptually this is a great idea, why limit the imagination and creativity to just Microsoft employees? There is so much development and creative talent out there, it would give Microsoft the edge over other BI tools if there was an endless selection of visuals users could pick from. At the time of writing AppSource contains 262 “Power BI visuals”.

The reality however is somewhat different; what are the incentives for a third-party developer to build a custom visual?

  1. Kudos – if there are not masses of visuals available then you’re one in a million for having your work in AppSource.
  2. Marketing – it’s free advertising to have your company name there in AppSource. This is an attractive model for IT consultancies as they use it as an entry point to new clients, rather like a fly fisherman uses a lure to catch a fish.
  3. License Fee – you can monetise your work, some custom visuals have a separate license fee to access the non-basic functions.

All well and good and from a user perspective as it should mean more choice, a wide support network and a never-ending supply of custom visuals with which to expand the reports. So, what are the problems I see with this approach?

1) Choice – Anyone who creates music on a computer will recognise the approach Microsoft have taken, you record music using the base software, but can enhance the sound etc using “plugins”, which are  third-party add-ins to allow you to tweak the sound in endless ways. With thousands of plugins available more time is spent fooling around trying different plugins than making actual music. Power BI is heading the same route, time wasted sourcing and trying out different Custom visuals instead of building business solutions. Sometimes “less is more”.

2) Support – The italicised text below is from the EULA for a custom visual developed by Microsoft:

“SUPPORT SERVICES. Microsoft is not responsible for providing support services for the Visualization. If Microsoft is the Visualization developer, it may provide support services, but is not obligated to do so under this agreement. Contact the Visualization developer to determine what support services are available.”

If your client / employer is a business that relies on Power BI reports why would you use a tool that effectively had no support? The statements above are vague and non-committal and indicative of Microsoft’s approach to the Custom Visual ecosystem, “vague and non-committal”.

In some cases, the visuals are built by one person within a third-party organisation, what happens if they leave or move onto other work? Are the third parties committed to maintaining these tools, is it really core business or are they simply viewing it as a marketing opportunity, fancy lures to attract big fish.

When a fault does occur who is responsible? What is the turnaround time for a fix? Is the visual accurate? What are the limitations? And so on.

3) Cost – Microsoft has pricing models that support both profit and not-for-profit organisations. I’m not aware that external parties that charge for custom visual licenses support the same model in the case of NFP clients.

Additional cost comes with additional payment setup / administration and in large organisations that can take a long time to get in place, which is not a necessarily attractive proposition if the third-party developer is a small business, reliant on the cash flow.

4) Standardisation – For a long time there has been little to stop developers creating non-standard user features within a custom visual, the ArcGis visual (in its early days) for example had a completely nonstandard edit pane to change the settings and attributes of the visual. Some custom visuals did not support the colours specified in custom themes and others could not render colours accurately; some custom visual just look a little too “simple” and others are just too cumbersome to learn to use.

This lack of enforcement of a standard approach costs money in time wasted learning how a single visual works when the user would already know how, had it followed the standards that other built-in visuals follow.

As you can tell by now, but to clarify, it is not that I am not a fan of the concept of Custom Visuals (especially as companies can create their own for their own use), it’s the route Microsoft has chosen to take, which at times has felt unstructured and “free for all”, rather than planned and well-thought out / executed.

This week a very public statement was issued by a third-party developer CloudScope, stating that they are removing all their custom visuals from AppSource and would no longer support Power BI. They state two issues for this, one of which is that the lead time for reviewing a submission is now 90 days which from CloudScope’s perspective makes CloudScope seem sluggish in delivering fixes and enhancements when the delay is Microsoft’s very long review turnaround.

Three months is an eternity in business terms, and I suspect it is more indicative of a large review workload at Microsoft with a few staff than 3 months of rigorous review and testing.

The full CloudScope statement can be read here: https://www.cloudscope.io/blog/view/cloudscope-ends-support-for-power-bi-custom-visuals

One of the lines in their statement that caught my eye was:

Without contributions from third-party developers, the Microsoft Power BI platform will be less compelling

This really is the nub of this article, Microsoft has put too much faith in their approach to Custom Visuals ecosystem while the base product is only delivering to the lowest expectations of what a BI tool should do. They have abdicated their responsibility to a community that is not that fully supportive or engaged with this approach.

I personally believe that Microsoft needs to rethink its approach; encourage third parties to develop, and then purchase visuals that stand out for inclusion in the base product, standardising them along the way and supporting them as part of the base product license.

About Me: I am an independent BI consultant in the UK (Milton Keynes) and have been working with Power BI since 2013. I don’t develop Custom Visuals, so my thoughts here are very much from a report developer and users’ perspective