> I never really understood tableu, it just makes fancy graphs from your data doesn't it? What's new about it?
It's really easy for business users to point it at a database and get started on their own, exploring the data. It feels like it has a much lower barrier to entry than many other reporting tools.
And once you win the hearts of executives, that's kind of the end of that discussion. It's a really sticky product.
From personal experience, Tableu has saved my team countless hours adding reporting features to our products just because some middle manager on another team got a hair in their butt, so I can see the value.
I once accidentally spawned a five-person department at a Fortune 50 company because I threw together a Django app that integrated with Active Directory and dynamically displayed a few reports that were relevant to the logged-in user's role in the organization.
I was in Corporate Strategy, and one of the C-level execs saw it. They realized that I could add new reports in a few days by myself instead of going through IT, and within a year they'd hired a new manager to run my little group and three others developers - who weren't technically developers, because we weren't part of IT.
For another ~3 years, the company's primary reporting tool for front-line operations employees lived on a repurposed desktop being used as a server that was plugged in under my desk. We did get a nice, beefy SQL Server and a decent UPS, so it wasn't a complete shoestring project.
I left shortly after IT discovered they'd been cut out of the loop on a critical business process and demanded that the whole system be rewritten in C#.
"I left shortly after IT discovered they'd been cut out of the loop on a critical business process and demanded that the whole system be rewritten in C#."
been there, done that. you have to get them on the phone early and do the whole "oh you're so great" song and dance to prevent ego bruising.
when i scope projects for clients if I have to engage with their IT department I add 6-8 weeks of implementation minimum. Anything from getting a CSR signed to SSO integration is always an excruciating experience. Going over the SOW and hitting on that item always results in a slow, exhausted, but accepting sigh from the client.
I think you might be surprised at how many people do know what a join is, they may call it another name, but it's not a hard concept to understand. It's a very common thing I've seen non-technical people do and use.
Yeah, one of my old coworkers was ecstatic when he learned about joins. He'd been downloading two tables and then using vlookup to join them in Excel. Using actual joins sped up his workflow a lot.
Thats why their non-technical user abstraction is a worthwhile investment for businesses who want their devs to be doing real work instead of pandering to the latest whims of a product director / marketing exec
At my last job I had a nontechnical boss, and I would have been ecstatic if he had a) enough of a clue to try screwing around with SQL, and b) the wisdom and humility to ask for help.
Yeah, I occasionally have to visit folks and walk them through extremely basic things like sorting data in Excel and creating pivot tables (very much not my job, I'm general purpose BI development mixed with statistical analysis & ML) but I never begrudge those meetings because it means the user is trying to become empowered to work more independently. No matter the specific skill level, that's something I can get behind.
At my place, somebody a while back drank the Tableau koolaid and quickly designed some nice looking graphs and sold the commercial team on its value, and the president signed a check for a license.
The software team was then tasked with putting the fancy graphs that guy did into our web application with the following constraints:
- the Tableau server can't be exposed on the public internet
- our UI can't indicate that it's using Tableau anywhere (i.e., use an iframe or something)
This turns out to be tricky. Tableau doesn't really like to be embedded in 3rd party applications, it leaks information about itself in a number of places. It requires that every person looking at the graph be a user according to Tableau's definition of user. Synchronizing authentication to Tableau server and workbook authorization gets tricky.
The next task coming up is that they want users to be able to ad-hoc schedule an email to themselves with the fancy graphs attached as a pdf, so we've got that to look forward to.
I worked on a data dashboard site that embedded Tableau workbooks. I found their JS API to be decent for embedding workbooks[0]. I used what Tableau calls "Trusted Authentication" to handle the authentication to Tableau Server[1]. It's harder if they are using Tableau Online.
>The next task coming up is that they want users to be able to ad-hoc schedule an email to themselves with the fancy graphs attached as a pdf, so we've got that to look forward to.
This is the endgame of every OLAP tool. But execs really love this feature.
1) An effective tool for people to explore data (with a relatively low barrier to entry - some training required).
2) An effective dashboard authoring tool (i.e. to make small specialized data reporting apps) which are simple enough to be used by anyone without training. These dashboards typically give some sort of situational-awareness for key performance indicators (KPI) such as sales, inventory, etc and are highly specialized for a specific use or role.
Imagine how useless an airline website would be if it were non interactive reports of all the data for the day loaded onto one page. Good dashboards are interactive, not passive reports.
>This change in approach could be equated to going to individual airline websites to check routes, dates, times and fares of flights as opposed to just going to a website like Orbitz or Travelocity – punching in where I want to go and when, and it pulls in a report of all the flights that meet my criteria. I can then narrow those results down by a number of criteria – time of day, number of stops, price, etc. It’s self-service reporting in the truest form.
Making fancy graphs beyond standard spreadsheet templates can become time consuming / labour intensive very quickly. More enlightened companies will not hesitate ot pay for Tableau if it relieves their 6-figure salary data scientists from reporting/dashboarding work.
I think PowerBI compares with Tableau quite favorably. For the most part PowerBI is a clone of Tableau, but it goes farther and adds in much more powerful scripting(M, Python, R), a library of custom visuals, Power Query which is a nice GUI for data transformation, and deep formula language (DAX). I think it's easy for an Excel or Tableau user to pick up in a day, but it scales much more as you have more data analysis expertise.
Also MS has been improving it like crazy over the past few years. They clearly want to maintain their dominance in the data analysis space, but see the winds shifting away from Excel and are not waiting to become irrelevant.
The model isn't that a business user would use Python or R. Rather, a data scientist shares live-computed analytics to business users through a report / dashboard.
Microsoft has PowerBI which is already almost magic for many business users with it's ease of connectability with different databases and SaaS services. Transformations, parsing, mixing data from different sources and reporting capabilities are also easy to start with and quite powerful in the right hands. It's not integrated into Excel directly but sits quite well imho in the MS ecosystem.
PowerBI wants to be Tableau based on the number of times Microsoft sales has demoed and whatnot.
But I haven’t met databiz people who switched from tableau to PBI. I’m not sure why as they look pretty similar on paper.
I think Tableau’s mental model is about exporting and storytelling while PBI is about reporting.
Personally, I find it harder to work with PBI because it’s the only license model more confusing that Tableau. And there’s no Tableau Public equivalent.
This - we had power pivot in excel - then for some licensing reasons we didn’t. I know how to buy excel and firms buy it blindly - tie power bi into that sales channel
I'm not the OP but having dealt with QuickSight it really feels super flaky and difficult to create graphs with. I can't remember the exact problems my co-worker had problems with it, but if I recall correctly one of the problems was that QuickSight expected all JSON data keys in S3 to be in the exact same order in every file thus making changing the schema a very huge pain in the butt. Also the UI itself is not good and creating eg running monthly average was very non-intuitive. A lot of small stuff that makes you not wanting to deal with it if possible. But it's cheap and has good AWS integration so I guess it has its purposes.
Coming from an enterprise background in BI using traditional tools like Cognos and Business objects, Tableau is head an shoulders above these. It's not only significantly easier to build complex interactive dashboards with Tableau, Tableau enables a level of rapid data exploration that simply isn't possible in the traditional tools. In those, you pretty much needed to know exactly what you wanted to build, making it difficult to find new insights.
It's like Access for business people and in many cases gets driven by the business, not IT. In many cases, it cuts out the IT SMEs that govern access to something like business objects.
They also do a good job at seeding it in visible places. Alot of newspapers are using Tableau for infographics, etc.
The primary value is making interactive data visualizations with minimal effort. I can go from nothing to having something in the user's hands in less than a day, complete with things like tooltips, dynamic filters, parameters, etc.
I love Redash, but Tableau has a much larger set of features. Redash is pretty good at queries and basic charting. Tableau is pretty good at queries and basic charting, but also offers a bunch of extra visualization types, predictive analysis, native cohort tools, etc, etc.
(That being said, I think a lot of teams are quick to reach for heavyweight tools like Tableau when Redash would be perfectly good for what they need at far less cost)
