General Discussions

Discuss any topics that are not product-specific here.
SOLVED

Best practices in workflow testing

beejprado16
5 - Atom

Hello! This is for a standard workflow in Alteryx Designer. I have this project to automate tagging of opportunities and accounts from Salesforce. The workflow just adds two more fields to the data which is kind of the following:

 

1. Does the opportunity meet a certain set of criteria (yes/no)

2. If yes, what category does it fall under? (each category has its own set of criteria)

 

My question is that how do I effectively test my workflow results and show these to the stakeholders to build confidence in the model? are there best practices in workflow testing? I have ideas on my own but I also want to ask the community regarding this. Any idea will be appreciated. :)

4 REPLIES 4
caltang
17 - Castor
17 - Castor

What I would do is to take a subset of your large data from Salesforce and run a sample use case against your manual process/other process via SQL etc. 

 

You can do the following:

  • Accuracy of results - Do they match 100%?
  • Speed of workflow - How much time do you save?
  • Ease of making changes - Best way to do this live is to change your criteria live, and show stakeholders how easy it is to change. Another way would be to have a non-Tech user try the workflow and see if they understand.
  • Add comments, containers, and annotations to better show the documentation within the workflow itself. 

In fact, I wrote a whole Blog post here: https://community.alteryx.com/t5/Engine-Works/Tips-for-Reverse-Engineering-Current-Processes-in-Alte... 

 

Have a read and let me know if it helps! :)

Calvin Tang
Alteryx ACE
https://www.linkedin.com/in/calvintangkw/
fmvizcaino
17 - Castor
17 - Castor

Hi @beejprado16 ,

 

As a technical presales/sales engineer responsible for proofing the value of Alteryx to prospects, this topic was always a struggle not only for me but for all the partners across the world I would say.

 

  1. Know your audience: Are they technical, somewhat technical or not technical at all? You need to please all of them but the focus of what you will present need to be focused on the main stakeholders
  2. Build a simple PowerPoint focused on the results with KPI, reasons to use the workflow and so on and additional slides to simply explain and contextualize what your model is doing. Always using the KISS method (Keep it simple and stupid 😅)
  3. Your workflow needs to also be simple and compartmentalized, meaning you should add containers to separate parts of your workflow and also closed containers or macros to hide the complex parts.
    1. Example of containers
      1. Inputs
      2. Preparing the datasets
      3. Adding Business Rules
      4. Selecting the best category
    2. Hide all complex business rules in a macro or minimized container. You don`t have to comment or explain every tool in your workflow, but you can add a known business rule name as the container name
  4. When presenting your workflow, try to make it as easy as possible. Try the best you can not to lose yourself on very technical explanations or trying to explain a single alteryx tool. I understand that technical users usually tend to go in that direction when nervous or something like that, so keep that in mind.
  5. The structure of your presentation could be something like this:
    1. Introduction
    2. Explaining the problem - if you have KPIs, even better. 
    3. Explaining your solution with KPIs (@caltang `s suggestion)
    4. Demo your workflow and introduce Alteryx if needed
    5. Only go into details if needed. (If people start asking very specific questions in the middle, tell them you will show at the end of your presentation)
    6. Done!

Not sure if that was what you are looking for, but there you go! :) 

 

Best regards,

Fernando Vizcaino

 

 

 

beejprado16
5 - Atom

Thank you @caltang and @fmvizcaino for your inputs. I also read the blog and yes it was helpful. I was actually thinking of sampling the data and have the end users do their current process and compare their results with the results of my model. Maybe, I just need to align the expectations on accuracy and how long is the QA period going to be. Thanks for the tips as well!

caltang
17 - Castor
17 - Castor

Glad it helps! Could you mark both replies - @fmvizcaino's reply and my reply as accepted solutions? This can help others in the community in the future as well + to close the thread.

 

Thanks @beejprado16 !

Calvin Tang
Alteryx ACE
https://www.linkedin.com/in/calvintangkw/
Labels