Power BI, Azure Analysis Services & Storyboarding

Power BI, Azure Analysis Services & Storyboarding

Most BI Delivery teams will recognise the weekly or monthly grind of producing the “Executive update dashboard” powerpoint deck. The new bookmarking option in Power BI can accelerate that process in a secure & flexible approach.

This weekend I came across a comprehensive walkthrough of Power BI’s new bookmarking features by Ian Horne (@olijoshcom), it’s this walkthrough that mostly inspired this post.

The Exec Storyboard Process

Most Execs have neither the time nor inclination to access BI or Analytics tools but they still need insights and therefore many organisations employ teams of people to create weekly or monthly exec storyboards. These can go by different names (Exec Dashboards, Scorecards, KPI updates) but essentially they equate to a collection of preformed charts & diagrams to illustrate key data points over the previous period, exported as images and collected into lengthy (perhaps 40+ slide) powerpoint decks.

This method is common as it’s the simplest way of getting multi-dimensional viewpoints across several data sets into an easily consumable form for a non Analytics tool user. However it has numerous shortcomings:

  • Data Security Risk – the figures often constitute “insider” level information however they are presented as images in a deck with no inherent authorisation or authentication (except perhaps at the document level).
  • Collation of Data – Data often comes from multiple places & data sources, and often in incompatible formats. A common request is a “Customer 360 view”, a summary of all major activity (across all sources) pertaining to key accounts/customers.
  • Static & Stale – The figures cannot change and cannot reflect recent updates.
  • Time to Market – as these charts are static all major combinations of interest have to be produced. 5 divisions across 4 major markets = 20 sheets.
  • Manual & throwaway – The process is intensely manual and largely throwaway. Some Excel or similar automation may be implemented but generally the data collation & formatting is carried out in manual steps which must be repeated each cycle.

The preference would be for the Exec to open up a data discovery tool for themselves and to pull the underlying information & insights directly however this is often not feasible. Apart from the fact that most Execs don’t have time to open analytics tools and go searching a key part of the storyboarding process is the focusing in on key data points of interest across the entire reporting domain. The storyboard provides a narrative, a “guided analytics” approach to highlighting issues and opportunities. Also, there’s no performance lag in a static image, the data is available immediately. The original charts may have taken several minutes to produce.

Enter Power BI, Bookmarking & Azure Analysis Services

Azure Analysis Services – foundation for storyboard

This blog post covers this in more detail however to summarise an Azure Analysis Services Tabular model can be quickly developed by non technical people to provide a collated & conformed view of data across the organisation.


Once the model is deployed it can then be refreshed periodically & secured via Azure Active Directory on a column by column basis so that only authorised users can view the underlying data.

Note: Although model creation & deployment can now be handled by non IT users via the new Web interface, security and particularly column level security currently still requires IT intervention.

Storyboarding via Power BI

The Power BI tool is then implemented against the underlying model with the following features providing storyboarding capability:

  1. Bookmarking – the principle feature, as described in the above video blog, allows for the saving of individual tabs & their selected filters.
  2. Highlighting – individual elements on the dashboard can be highlighted, bringing out charts or filters of particular relevance.
  3. Element hide/show – individual elements can be shown or hidden as needed, this allows for narrative comments to be added to the storyboard which can be toggled within a bookmark or between bookmarks.
story 1
Story 2
Comment toggled to add narrative with link

Using these 3 features a full narrative can be put together which steps through the storyboard at the User’s own pace. There are several advantages to this approach over the traditional Powerpoint deck method.

  • Greatly reduced Data Security Risk – The data is live & governed by the user’s given authorisation. No copies of the data are held on the User’s environment – this may be of particular benefit for GDPR considerations.
  • Collation of Data – Azure Analysis Services (or Power BI itself) can simplify the collation of data across multiple sources
  • Interactive & Live – The charts are fully functional & built on the latest copy of the data. Although the storyboard is a guided process there is nothing stopping the user delving deeper. This in turn leads to:
  • Reduced time to market – Without the need to export and collate images, and especially without the need to create an image for every combination of possible points of interest the storyboard can be created in a fraction of the time
  • Reuse – Once built, the same framework can be used the next cycle.

Note – Images, offline copies & GDPR: As mentioned the Data Security & particularly GDPR considerations may be of particular importance. A static powerpoint containing customer data in images is difficult to track or manage, whereas in this approach data is pulled fresh each time from the source, if a customer has requested the “right to be forgotten” they will disappear automatically on the next refresh. Although Exec dashboards most likely will not contain PII (personally identifiable information) the general point of accessing data from source rather than from some off line copy is particularly relevant when it comes to customer data.

To summarise, Power BI Bookmarking support (in preview) + Azure Analysis services provides a powerful and flexible option for the classic Exec Storyboard delivery process. By greatly reducing manual work & time to market, insights can be delivered more quickly in an interactive, secure platform.

Footnote 1 – Why Azure Analysis Services?

Most of the above functionality applies to Power BI and it’s certainly true that almost all of the above could be delivered with it alone. However there are several advantages to using AAS in the stack. Firstly, there is a question of scale – a power BI imported dashboard can contain significant data but eventually it will run out of space and this in turn may lead to the issue of a user drilling down to find root causes but finding themselves limited by the data scope in the workbook. AAS can scale in data volumes up to 400 GB compressed which provides far greater data scope.

The second issue is around security – it is far easier to enforce security in a central AAS model than across multiple stand alone workbooks and finally there is the benefit of operating over a single version of the truth, centrally defined business metrics which are the Enterprise standard.

Against this has to be set the flexibility & agility of a “pure” Power BI solution however there have been some significant advances which now make it far simpler to start with a Power BI model and then to scale up to a full AAS implementation. This blog post on Azure Analysis Services Agile Delivery covers these advances in more detail & this post covers the 3 separate mechanisms for accessing data via Power BI.

Footnote 2 – Data Security & Data Freshness

One key issue of self serve BI is that data can be extracted from corporate data warehouses & marts and then copied around the business, hopefully to people who are authorised to the underlying data but possibly not. This activity is common, a default action is to get Analytics data into a spreadsheet, play with it and then reformat it for downstream users.

Linked to the above is data freshness, once the data is extracted into Excel or some other analytics environment it represents a point in time & is no longer connected to the underlying Data Warehouse/Data Mart. These 2 problems are particularly relevant to the common analytics workflow outlined above, the “Exec Storyboard”.


Leave a Reply

%d bloggers like this: