What is Tableau good at?
Great question, actually.
Tableau's main features are:
- Easy to connect to a variety of data sources
- Build interactive dashboards in minutes
- Usable by business people across the organization, with no need to write software or tangle with SQL queries
On the face of it is exactly what DevicePilot is good at too - so what's the difference?
How much data do you have?
If you google "Tableau IoT", you'll discover plenty of conference talks about how, in principle, you might use Tableau with IoT - for example, this one talks about how IoT will generate 40Zb (zettabytes!) of data next year. But if you search for actual, worked examples, they tend to be someone connecting up a single Raspberry Pi as a demo, which generates a few channels of data, which they then analyze to produce a few line charts. Fine for demonstrating the principle, but you could easily achieve the same result with almost any tool such as Excel or Google Sheets.
That "40Zb" is part of the answer: even modest IoT deployments create large amounts of data - just 1,000 devices updating 10 properties every 5 minutes are creating 86 million datapoints. So what? You can store that data hundreds of times on a humble $10 SD card the size of a fingernail.
One of the most common data ingestion sources for Tableau is a spreadsheet - but not many spreadsheets have 86 million cells, and this is just one month from one small IoT device estate. The task is to digest high-volume, relatively low-value data into key metrics and Tableau starts to struggle to digest this amount of data efficiently.
What kind of data do you have?
IoT inherently produces live streams of changing properties. Tableau can igest data from streaming sources, and Tableau dashboards can be set to auto-update to reflect the changing analysis. But IoT data isn't just live, it's actually a time series and so in most IoT applications, it's vital to be able to do powerful analysis on time series data.
IoT time series analysis
Very often in IoT, you require time-based analytics to be done at the per-device level then aggregated into overall numbers about the device estate. This isn't simply a case of aggregation across the device estate - the "roll up" that all time series databases provide, for example, average temperature, doesn't tell you anything very useful. However, average temperature over time? Absolutely crucial information.
If, say, a connected industrial fridge door doesn't close after two minutes, you may judge that the door has been left open for too long... but note that no new data arrives at this moment - it's the absence of data which causes the change in state.
And finally - most questions our customers ask of the software only make sense if it's asked over a time period, such as one week.
Your analytics tool must be able to drag a question like "how long was the fridge door open?" or "how often were no fridges available?" through all the data from the past week, for each device individually. In Tableau, this is a difficult if not impossible question to frame, whereas it's the most natural kind in IoT.
Rules, triggers, actions
Every IoT company is a service company which, as it matures, effectively becomes a well-honed set of processes for delivering and managing an excellent service to its customers.
If those processes are carried out by humans alone (even humans looking at beautiful, up-to-date dashboards), then they will be carried out slowly and inconsistently. Plus, the increasing number of people needed when trying to scale processes gets expensive. It's therefore important to automate the handle-cranking processes, which is a key part of any IoT platform. Tableau doesn't really have a solution for this.
DevicePilot, by contrast, can easily notify people when action is needed, or keep other service tools (e.g. CRM systems like Zendesk) up to date with the state of your device - for example, by flagging a ticket when a device goes offline, and then automatically closing it again if the device comes back online.
Tableau has great integrations for ingesting data from databases and spreadsheets in cloud services. DevicePilot is great at ingesting data from streaming IoT platforms in cloud services and from IoT communications providers.
DevicePilot also integrates "sideways" with services such as Slack and Zendesk so that it can trigger actions on them as above.
Perhaps the most interesting point about integrations is that DevicePilot is great to integrate with Tableau. DevicePilot is great at digesting the live stream of IoT data into a much smaller, higher-value stream of business analytics, which Tableau can then ingest and combine with other information such as sales data.
At a glance
Tableau is a great tool, but it's focused on answering business questions on business data, which is somewhat different from IoT questions on IoT data.
Is it impossible to use Tableau for IoT visualization and management? Clearly not - in software, anything is possible! You could use a spreadsheet for customer support, or use Google Docs to manage your billing. But really you shouldn't - clearly, using the right tool for the job delivers enormous benefits in productivity, business growth, and the happiness of your customers and your team.
Want to see more great IoT content?
Sign up to our (spam-free!) newsletter for industry news, updates, and advice on how to get the most out of your project: