5 Considerations when using Tableau for survey Data

When I present Tableau to people working with survey data, I often hear complaints that Tableau doesn't support their needs. In this blogpost I want to emphasize why this is wrong and even how I save up 50 % of the time I used to spend creating a presentation on survey data.

Tableau is not an out of the box survey machine, however if the below considerations are made, then you will never again want to go thru the old excel -> powerpoint -> excel -> powerpoint treadmill.

1: Transpose Your Data

Tableau works best, when your data is inputted on the lowest possible level. This means that the classical structure of surveys where each column represents one variable is no longer the best. In this structure the lowest level is the respondent, but we need it to be the answer of a specific question. Your data should look like the example below. Each respondent is now represented 1 time pr. question. We use Alteryx to do the transformation, but you can also use this nice tool from Tableau.

2: Use Weights According To The Result Type.

When working in SPSS or SAS a weight is simply added to the data, and all our calculations from now on are influenced by the weight. Tableau is not built for weighing and we must therefore think before we calculate (Damn!). The two primary things are percentage and averages. When calculating a distribution of percentages on a specific question just use the weights as measure and do a table calculation. When calculating averages, for example average satisfaction on a 1- 5 scale, the following formula does the trick. This requires that your data is transposed as described above.

3: Do String Formulas In Advance

One of the worst things a user can do to destroy good performance in Tableau is a long string calculation. If you need to visualize all the users that mention a specific brand do the calculations beforehand, and store it as a unique variable. This is not the most sexy solution, but you will be happy about it when presenting to management or a client. While you are at it also do your NPS conversions and so forth beforehand. Even though calculations on numeric variables are much less of a performance problem it all counts towards a nice experience for the client.

4: Utilize The Power Of Tableau

It is important when building a survey presentation in Tableau to Tableauify the charts and dashboards. If you try to convert your 80 slide powerpoint presentation 1:1 into dashboards, then you will have gained nothing. Instead build filtrations into your dashboards and allow the client to do click whatever they need information about. In the below anonymized example, the client can click one of the KPIs in the top, and get information on each competitor. Afterwards the client can click the competitors of interest and the KPI development for the chosen competitors are shown on the right. 1 dashboard substituting at least 10 slides.

5: Use New Datasources

Coming from the world of SPSS and SAS, survey analysts are usually only working with survey data. Sometimes they sneak in some mediaspending but that is about it. With Tableau it is so easy to import datasets from very different datasources and blend them in the same analysis. Maybe geographical demographics can explain the varieties in our satisfaction level? Or maybe sales, spend, crm AND survey are better at describing the current state of things than survey alone?

I love working with survey data in Tableau and I believe that companies that will continue to work with survey data in old formats will have a hard time in the future. We leverage on Alteryx to speed up the process by as much as 500 % from old methods. In a blog post soon I will describe how.

Back to blog

Subscribe for updates