Tableau ≠ Portfolio Intelligence
Skip the cost and headache of a DIY-solution that was never intended to support the specialized use case of an institutional investor.
Business intelligence is undergoing a renaissance. With so many options on the market, like Tableau, Sisense, Looker, Qlik, and Microsoft Power BI, these tools become cheaper and more user-friendly by the day. True data analytics have arrived on the average user’s screen at long last.
The trend has not gone unnoticed, and big data companies are securing their strategic foothold in the space, as recent large-scale acquisitions clearly testify. This year alone, Tableau was acquired by Salesforce for $15.7bn, and Looker was acquired by Google for $2.6bn. And those are just two of the most prominent.
But what is the best strategy for institutional investors? Should you subscribe to one of these common tools to get your dashboards up and going? The institutional investor has a choice: classic business intelligence like Tableau, Sisense, Qlik, etc.; or specialized portfolio intelligence like Novus. ("Portfolio Intelligence" is a term we actually coined a decade ago to describe how business intelligence applies to institutional investor portfolios.)
Many investors fail to anticipate the cost of making classic business intelligence (BI) tools work for their unique needs. They may purchase something like Tableau, and proceed to tear their hair out for as long as it takes until they achieve Jedi status with the tool, only to eventually build something that almost meets their needs. Granted, we are biased, but it's worth stating emphatically—there is a much better way!
Limitations of Classic BI
I recently spoke with a client who was an early adopter of Tableau. Their firm quickly realized that in order to build insightful and usable dashboards for their business, developer-level skills would be required to organize all their vital data and information. They also felt the pain of Tableau’s SQL limitations, the siloed experience of limited collaboration features, and the lack of regression tools, which meant tedious manual checking of complex calculations.
This client's experience aligns with that of many other frustrated investors who have tried or are trying to use tools like Tableau to extract investment insight. For the benefit of any investor who might be currently considering a classic BI solution, let me save you precious time. Here is a summary of the shortcomings investors commonly experience with tools like Tableau, This is directly based on my conversations with current or past users.
1. Slow Responsiveness
Classic business intelligence was built to visualize easy-to-handle, fairly simple data sets, like pharmaceutical sales figures by product, sales rep, and region. When applied to the complex data cubes we have in asset management (plot me the rolling alpha by security over a 6-month lookback period against a benchmark made of 3,000 positions while ensuring that the sum across securities matches the portfolio rolling alpha), you run into a few problems, one of which is speed. Classic business intelligence does not parallelize calculations. That is, it won’t split the job into easy-to-handle tasks for which several CPUs are recruited in parallel. You’ll be waiting forever, which is a huge user experience downer in an industry whose practitioners have little patience.
2. Missing Calculations
Portfolio intelligence, and finance more broadly, requires specialized analytics, which are not embedded in classic business intelligence software. Aggregating sales figures within a classic BI tool does not require variance-covariance matrices, or GIPS-compliance smoothing algorithms to ensure that single-position compounding preserves additivity at the portfolio level. Sounds Greek? Well, that’s just the basics that you have to go through to run simple contribution analysis over a period of time. It can get uglier than that, especially when benchmarking is added to the picture. So what happens then? If you chose classic BI, you’ll find yourself lost for months trying to squeeze in add-ins and other specialized routines to make it work.
3. User Imprisonment
Because of the previous point, you’ll end up with an organization where users will be held hostage to a few individuals who can maneuver an archaic Frankenstein of classic BI. Want to change the order of elements in a report? Sounds fairly simple. And yet, you have to call someone. You manage your life by yourself through apps on your iPhone, and yet you can’t change the graphic in a report without calling someone. The only people who end up happy in this situation are the few individuals whose jobs are secure because you have to ask them for help every single time.
4. Static Navigation
Compared to other industries, institutional investing is far less straightforward. It does not take much analytical work to see if a pharmaceutical is selling in a region, and whether you should consider replacing the rep. Determining if there is erosion in your capability to generate alpha requires a lot more work, as underlying factors may be numerous, and numerous are the ways in which you’d have to test them out. Therefore, you need multiple concatenated analytics, sequential drill-downs, and fluidity when moving from one analysis to the next. To a large extent, classic BI works with static reports, where drill-down and independent navigation possibilities are limited. With a tool like Novus, every click is a move into another part of the platform, which offers the deep-dive investors need. And again, this is a journey you can navigate yourself.
5. Hidden Costs
Classic BI has become dirt-cheap, with some packages available for free, at least at the beginning. In a classic freemium offering, at some point, whether because you’ll want more users or more usage, you’ll have to pay. And still, it all remains cheap, at least on the surface. The problem is that, pretty much like SAP models, you’ll soon have to spend vast amounts of money for internal developers or external consultants to come in and configure the system to your liking. And that’s when your costs balloon in a very pernicious way—they are hidden within the employees you’ve hired and their time. Some people think configuration costs are transitory and only tied to the set-up phase, destined to vanish after the set-up job is done, never to return. That is also wrong, as maintenance, updates, change requests, training, and other recurring needs may consume as much as 40% of the initial setup. And what’s more, this will happen every year.
Roadblocks to Adopting Portfolio Intelligence
Faced with indisputable evidence, you’d think institutional investors would make the obvious choice of opting for a portfolio intelligence tool—the specialized solution built just for them. Yet, some don’t. Why does logic not prevail? Well, the answer is in the question itself: logic is not what’s driving these decisions.
Here’s what might be driving the decision instead:
First, so long as technology investments are delegated (or relegated rather) just to CTOs, a principal-agent problem may ensue. CTOs may optimize for their own job security, and get organizations down a path which creates dependency on them remaining employed. Most CEOs cringe at the thought of disentangling their organization from the architecture that their CTOs have built. How to remedy the problem? Simple: CEOs must see technology as a critical part of their business. Like other functions, be it distribution or your investment office, technology engagements possess their own P&L, ROI, and similar considerations. CEOs no longer have an excuse for saying “I don’t understand the matter, I am fully removed from it.”
Second, margins in our industry are still too high for rational cost optimization to occur. Decision makers rarely bother with rational cost optimization, and I can’t blame them. $1m more in AUM pays for any and all abundant inefficiencies. If you’re managing close to $1bn, why bother? While this opinion may buy us some enemies, it may be your fiduciary responsibility to consider this reality.
Third, an institutional investor’s job is often misunderstood. The real job is decision making (and that is the intellectual property which one must protect). Data, analytics, and infrastructure are the support functions aiding your decision making. As such, you are better off outsourcing the pieces not directly tied to your core purpose.
The Tableau user I mentioned earlier ultimately concluded that Tableau worked fine for their firm's generic data visualization needs—things like number of investors per fund, investor flows by fund, or operating metrics for the business not related to portfolio analytics. Today, Tableau continues as an embedded part their process, but in their words “with a plethora of creativity to get things working well.”
They were not, however, able to create satisfactory portfolio analytics of the more sophisticated (and crucial) variety; think multi-factor risk modeling, hybrid benchmarks using variable time frequencies, or skillset analysis, to name a few. Tableau was heavily embedded in their business, but the process of moving to a true portfolio intelligence tool has been extremely beneficial for the firm, in their words.
Adopting specialized portfolio intelligence tools from the beginning allows investors to get the data and dashboards they need, without the cost and headache of attempting a DIY-solution that, frankly, was never intended to support their specialized use cases. If you'd like to learn more about how true portfolio intelligence can help you surface insights while streamlining your workflows, just drop us a line.