Every so often I get an email from an analytics vendor letting me know that they've added new functionality to the platform. We've got great new features – would love to show them to you!. Inevitably, the fantastic new functionality turns out to be something like the ability to change dashboard themes or an additional database connector. These are great things, but fall a little short of groundbreaking.
It would be fair to say that over time, I’ve become a little jaded to these types of vendor announcements.
So, when I received a product announcement email from Periscope Data a few weeks ago, I was prepared to be unimpressed. I assumed that they had added new theming options, could now connect to some obscure database, or had received some fresh new certification.
Boy, was I wrong.
You might recall from my recent “Vendors You Should Know” profile of Periscope Data that they are the “Swiss Army Knife” of analytic platforms. Most analytic systems are designed for one of two audiences: data scientists or general users. The data scientist platforms allow the user to perform complex, highly targeted analyses but often at the expense of usability. These platforms are too complicated for the average user. The analytic applications designed for general users are far more comfortable to use and allow non-data scientists to perform pre-defined analyses and some ad hoc exploration but aren’t well suited to the types of complicated, one-off investigations performed by data professionals.
Periscope bridges the divide between these seemingly separate worlds and allows a company to implement a single platform that serves the needs of both data scientists and general users or, in my area of interest, product teams building customer-facing analytics. There’s an obvious cost-saving benefit to having a single platform for all analytics, but there is another less apparent advantage. Using a common platform — such as Periscope Data — for internal analysis and data products reduces the risk of a mismatch. When using multiple platforms, there’s a possibility that the numbers from one system aren’t the same as those from another. Time is wasted trying to run the cause of the discrepancy to ground, and when the issue surfaces in a data product, credibility is lost.
Even still, I wasn’t expecting all that much when the Periscope team highlighted two key new features they’ve recently rolled out: R and Python integration (launched in February) and “visual data discovery” (launching today).
This was an error on my part. Although the new Periscope release contains several exciting and useful new features, these two are true game changers.
Let me explain.
R and Python Integration Means No Limits to Roadmaps
When I heard the claim *R and Python integration* my first thought was that I'd seen this before. Several vendors already have this capability, don’t they? The way the integration of these core data science tools usually works is this: the data scientist gathers the necessary data and then creates the Python or R analysis outside of the data product. They platform claiming “integration” then displays a static image of the resultant R/Python analysis on a dashboard with very limited (i.e., zero) interactivity. You aren’t drilling down on these “integrated” analytics — they’ve got the same level of interactivity as that cat photo on your Facebook page.
Periscope has changed this model — radically — with their new data science platform integration. R and Python don’t just generate static images; they have the same interactivity as natively generated Periscope analytics. Data scientists can create R and Python analyses right from within Periscope. They can connect to the same data sets offered to the native analytics engine reducing the likelihood of discrepancies. The R/Python code runs directly from within Periscope and the chart, once embedded alongside other Periscope analytics, is indistinguishable from a natively generated visualization. You can drill down, filter, view the source — everything users expect from Pericope Data analytics.
For the data product builder, this is a considerable advance. It means that any analysis you need to create, any visualization you require, can be generated using the data science standards of R and Python. It means that product teams don’t need to worry about limiting their roadmap should a unique customer request come along. If you can write it in Python or R, you can seamlessly integrate it into your Periscope Data dashboard for use by a non-technical audience. And, since these integrations are done inside of Periscope and against the same data model, you don’t need to worry about data synchronization issues like you do with some other platforms.
Visual Data Discovery Brings Power to New Audiences
The second big announcement, *visual data discovery*, is equal in its impact on data product teams. When I first reviewed Periscope Data, I was extremely impressed by its ability to serve both data science and general business audiences. But, I was left feeling that the balance between the power user and the average user was skewed toward those with coding abilities. Once analytics were created, they could be easily modified by an average user, but some SQL prowess was necessary to get started.
Requiring SQL wasn’t much of a problem for the initial creation of dashboard within a data-driven application (often aided by technical resources), but it had ramification for those teams that wanted to offer “ad hoc” analytic capabilities to their customers. It meant a higher barrier to entry (SQL skills) for those that wanted to create “on the fly” analyses.
With Periscope Data’s new visual data discovery functionality, that has all changed. Data scientists can still create sophisticated analyses by writing SQL queries (or R and Python), but now the average user can build charts and visualizations via a simple drag and drop interface — no SQL knowledge required. Product teams can not only develop and modify analytics themselves, but they can now offer ad hoc capabilities to their customers. And it gets better…
Other platforms offer drag and drop analytic creation, but without much consideration to the underlying data structure. They expose data fields to the analytic builder without much consideration to helping them understand what they’re seeing or if they are using the right data. If I need the customer ID number, is it “CustID,” “CUSTOMERID,” “CUSTOMER,” or “MASTERCUSTID”? It can be tough to tell if every data element is exposed to a user who is unfamiliar with the intricacies of the underlying data structure.
Alongside the new visual data discovery functionality, Periscope has introduced features to make this problem obsolete. Now, your data team can create subsets of the entire data structure and expose only those elements to visual data discovery users. And, with clearly understandable naming. You don’t need to search through ten variations of “CUSTID” — you simply grab the “Customer ID Number” field that was made available to the visual builder. But the underlying data is still the same, still always in sync, and everything is still available to the power users who might need all of those data elements. The days of answering support calls from ad hoc analysis users confused by convoluted field names is over.
I left my meeting with the Periscope team with a very different opinion than when I entered. This wasn’t a “change colors” type of release; it was a big deal. A game-changer. The power offered by Periscope Data — the ability for data science and data products to co-exist on the same platform — had been taken to the next level. For organizations seeking a single platform for all analytic needs, Periscope Data has just made your quest much, much easier.
* Disclaimer: This is not paid content and do not I have any type of financial relationship with Periscope Data. I don't accept any type of "pay to post" content on this site. Everything I've written here is my opinion based on a pre-annoucement demo from the Periscope Data team.