It’s 2018, and aside from the debate on the semantics of the words ‘BI’ vs ‘Analytics’, tools have started to stabilize a bit in terms of their vision and execution. Gartner released their yearly magic quadrant for analytics tools, and I found no major surprises. The same usual Analytics & BI Platform players made it with their entrenched install base at enterprises in the 2000s (looking at you, IBM and Oracle). Really, who wants to even begin to tackle re-writing 1000’s of Business Objects and Cognos reports? (hint: you shouldn’t do this. Without looking at your environment, I guarantee 90% of those reports are either redundant with different filters, or not being used).
Comparing Analytics & BI Platforms
In this post, I’ll lay out the top 3 Analytics & BI Platforms that I recommend to my clients, which may not always align with what Gartner says. There are always niche use cases that require different solutions and tools, but these typically cover the use cases I see at mid-size to large enterprise environments today.
There are many, many posts that compare these tools as if one is better than the other — I won’t attempt to do that here, primarily because it always depends on your own needs and environment. Instead, I’ll give a digestible C-level / Director-level oriented guide of what each one excels at and where they fall short.
Several new players made their way up, and Microsoft continued to execute on its vision to become the clear leader (with a caveat).
Microsoft Power BI
Microsoft was incredibly smart and deliberate about how to market this tool. You sign up, get a free trial, and you have a working dashboard within 10 minutes. After that, you pay $10 /user /month. Why wouldn’t you use this, given the low cost and ease of starting? (not to mention, their desktop tool is completely free)
I’ve seen many clients with a growing Power BI deployment, sometimes consciously, but sometimes before a data strategy is even in place. It’s just so easy and cost-effective that Power BI often becomes the de-facto tool at an organization solely because their current analytics tools (or Excel) just wasn’t working for them.
Power BI is an awesome dashboard tool, and Microsoft has the ecosystem to support just about every reporting scenario, from standard reports (SSRS) to tabular in-memory models (SSAS) to Excel-driven analysis and discovery within Power BI dashboards. It works, and it works well, assuming you’re bought into the Microsoft ecosystem.
That’s the gotcha, but also the reason why you would select Power BI (or ultimately, Microsoft) as your analytics platform. You absolutely need to be willing to go all-in with Microsoft, because there’s more you need to do to make Power BI an effective choice. Mixing and matches tools from other vendors will be a difficult-to-maintain environment, and much more costly.
You need to be aware that the $10 price tag for Power BI is only a small part of the equation. You’re going to need a semantic layer within Analysis Services for performance and governance — doing this will avoid creating 100 versions of the same data model. You also may need formatted reporting capabilities found in Reporting Services, understanding that there is still a gap in Microsoft’s execution. SSRS only works in on-premise deployments. Yes, you can embed them into Power BI, but you need to buy into a hybrid deployment model to make this work (for now).
At this point, now that you’ve decided to go all-in with Microsoft, you’re likely going to want to take advantage of Azure’s PaaS services. Don’t worry — they’re awesome, and you will love them if you’re a Microsoft shop. As you scale, the Power BI Premium offering will remove the per-seat licensing cost for view-only users, giving insight to the entire organization without the high price tag.
You may need to accept a multi-cloud or hybrid deployment model. Make sure your database is fully supported within Power BI. If you’re not running on a Microsoft SQL Server database or not using SSAS, you may run into some issues. There’s absolutely no good reason to spend money on a powerful analytics database, only to find out Power BI either uses ODBC and/or requires the data to be imported into Power BI before using it.
Microsoft is adding new features and improved connectivity to new databases and applications on a monthly basis. It may be only a matter of time before it becomes the default choice for reporting no matter where your data lives.
Summary: Power BI is your gateway drug into Microsoft’s ecosystem. Be willing to go all-in to reap the benefits. Budget wisely and plan for both the Microsoft services fees and implementation costs. Consider Power BI Premium for larger implementations.
If you’re in Chicago-land, this might be the first time you’ve read much about Looker. It’s growing in popularity on both the west and east coasts, as well as in the UK — and there’s a good reason for that. Looker has taken a completely different approach to analytics than the other players, and it either clicks with you, or it doesn’t. If it does, you’re in for a treat, because when you deploy it within the right architecture and environment, Looker becomes a complete joy for business users and data analysts alike to use.
Looker checks off most of the boxes in terms of dashboard capabilities and data discovery use cases, but also opens up capabilities that you may have never known you even needed — like collaborative sharing of your data. Unlike Power BI or Tableau, it requires a full semantic model for storing all your business logic and metrics without having to add multiple versions of a slightly-different metric to your database tables.
This means you cannot just take Looker, point it at a database, and get your visualizations in minutes. It requires upfront definition, thought, and development of what your data means to the business. This is a double edged sword — you now have single sources of truth across the enterprise, allowing for financial models, sales models, etc — but you need to put in that upfront work before you can start creating reports. From an enterprise governance standpoint, this is exactly what you should be doing.
Self-service BI should not be defined as business users spending their time writing calculations and joins — this is an IT role, with business providing the definitions and requirements. The real value of self-service BI comes when IT has made it easy to drag and drop data elements to quickly gain insights and drill into the details. Looker enables this, along with sharing, scheduling, and even integrating into Slack.
From a technical standpoint, Looker puts the processing 100% on the database. You will need an analytics-based database, such as Snowflake, Azure DW, Redshift, or BigQuery.
More savvy data analysts and developers will either find Looker difficult to use, or love its paradigm shift. Clunky drag and drop development (again, looking at you, Cognos) has been replaced with a language called ‘LookML’. It’s SQL-like, but the development is pure text. Joins are not defined by dragging lines visually, but instead by coding. Metrics such as ‘Net Sales’ are defined once, but can then be referenced within other metrics, such as ‘% Profit’. Git integration makes version control and team development a breeze. Traditional BI developers will likely struggle with this approach, and if you select Looker as your tool without a developer-oriented team, know that change management processes within IT will be something you want to incorporate into the project.
Summary: Looker is different, but provides a full semantic model for governance and a single source of the truth, as well as powerful sharing and exploration capabilities. Looker requires an analyics/MPP database for optimal performance, and you will need to budget this into your project. Forward-thinking organizations will need to embrace the developer-focused way of defining metrics and joins — or decide Looker is not the right fit.
This section is going to be a bit shorter, because a quick google search will give you just about everything you need to know about Tableau.
First off — there is no better tool at producing incredibly sophisticated and well-designed visualizations. The UI/UX for business users is why Tableau has been so highly adopted within business units. Tableau’s ability to connect to just about every data source makes it easy for users to point at their data and produce visualizations.
…and that’s the rub. There are two situations where I would recommend Tableau:
You have a well-defined data warehouse already in place, and you do not need to create an enterprise-wide semantic model on top of that. Ultimately, your data warehouse is your semantic model, and you will need to keep your ETL up-to-date to keep business users happy.
You need a tool for data discovery across your various source systems and applications, and the team using Tableau is data-savvy enough to understand the joins and the business logic within that source.
Performance is not Tableau’s strong suit, which means you need to invest heavily into your ETL. Develop your complex calculations and joins into the ETL and database as much as you can before it hits Tableau. You do not want to do this within Tableau, given both the degradation of performance and the devaluing of a single source of the truth.
Price has often been a major factor against Tableau, but costs have come down a bit with their adoption of subscription-based models. Given Tableau’s dominance in the market, as well as its ease of use, you will have no problem finding resources and developers that can help you with implementation.
Summary: Tableau blows the competition out of the water when it comes to visualization and ease of use. At the end of the day, however, Tableau is only a visualization tool — not an analytics platform. Your business users will likely ask for Tableau (if they haven’t already). So long as you know its weaknesses (performance issues, no semantic layer) and address them properly with well-thought and well-maintained ETL, Tableau will be exactly what business wants and needs. If you already have a large Tableau deployment and are having trouble maintaining it — take a step back and focus on your ETL and data warehouse environments — it will save you some time and future troubles.
Closing the Loop on Analytics & BI Platforms
In summary, the platform you select is going to be highly dependent on your future vision, your data strategy, and your business users’ needs. Before any tool selections, I recommend defining that future state with a data strategy. By putting the upfront thought into a well-defined data strategy, you will have a phased road-map on how to execute all layers of your data architecture, as well as a better sense of cost and timeline to deliver value to your customers.
This post was originally posted on Medium.
Fred Bliss is the CTO at Aptitive. He brings over 15 years of experience solving complex business problems through data solutions including cloud integration, data warehouse modeling, ETL, and front-end reporting implementations.