The Use Case

The customer wanted to update several fields on the related Case that tracked whether the person responded to the latest survey, and then update the Case status appropriately based on a specific question’s response. 

They also wanted to send an in-app alert to a Queue if the person responded extremely negatively to any of the numerical questions being asked in the Survey. So if they responded with a 0, 1, or 2 to the CSAT question, they wanted to know. 

Using Custom Metadata / Custom Settings to Automate Specific Questions.

In the previous article we talked about the Survey data model and a sample scenario with one question and 3 choices. We’ll use that in this new flow. Remember how I said it’s important to understand the Survey data model? That was no joke – you will likely be lost if you don’t take time to understand it. 

Just a caveat – this solution assumes you do not have Feedback Management, which has the super helpful Data Mapper solution which lets you do some of what Flow does, but without the Custom Metadata complexities introduced here. Feedback Management and Flow can live in tandem, however. If you need basic automations done in your system, Feedback Management is a great way to do that… However, if you need to do things like sending an in-app alert to a queue, or re-use automations across Surveys you’ll need some help from Flow.

I will say generically that if you have the same automation that needs to run for every survey in your system or is shared across many surveys, I recommend a Flow-first approach. This allows you to make changes to your automation in one place vs. doing it on a per-survey basis. Again, though it may be worth sacrificing 

Flow Entry / Early Conditions

I won’t go element by element unless demand arises, but I do want to call out a few checks you can make.

In the rare scenario where a Survey isn’t complete, we likely want to check that the survey is actually completed:

Next up we check that the Question Responses’s survey name matches what we’re expecting. That can be done by traversing upward: $Record.SurveyVersion.Name (or $Record.SurveyVersion.Survey.Name). We can’t do that in our entry criteria (yet), so we have to do it here in a decision node.

In my example below I actually have a custom lookup on Survey Invitation called ‘Related Case’, and I plan to update some fields on the Case, so I ensure there is actually a value there. That can be done by traversing upward as well: $Record.SurveyInvitation.Related_Case__c

Set Case Fields Based on the Survey Response

Next up is where I actually set some values on the related case to track the survey completion at the Case leve (not the Invitation/Response level). These fields on Case track that an evaluation was completed and when it was completed.

Notice I don’t do an Update here – I use an assignment to save on DML as I will likely be making more changes to the case downstream in the Flow. Efficiency!

{!recordCaseToUpdate.Id} Equals {!$Record.Invitation.Related_Case__c}

{!recordCaseToUpdate.Date_Evaluation_Complete__c}  Equals {!$Flow.CurrentDate}

{!recordCaseToUpdate.Evaluation_Status__c} Equals Complete

Set Case Fields Based on a Specific Response

As Anakin Skywalker once said, “This is where the fun begins”… and hopefully not “I’ve got a bad feeling about this”. 

The first thing we want to do is grab all of the Custom Metadata Mappings related to this survey.

Step 1

Do a null check on the results just to make sure you’ve got some records, then things get a bit more interesting and purely based on your use case and what you’re trying to do. 

In my Flow I looked for any questions that were specifically Choice questions (so no CSAT/numerical questions). This was done by checking that the QuestionChoiceId field (the lookup to the actual Choice in the question) is not null:

Step 2

My survey only had 1 possible question with a choice, so you may want to consider adding more criteria to future proof things.

Step 3 & 4

Next up is where I wish we had more collection processing power without assistance of invocables, but alas, this was my solution. I looped over the CMT mapping records I grabbed in Step 1, then I did another loop (GASP!) on the collection of responses above. I felt comfortable doing this because I was only ever going to have 1 response record x 3 mapping records, so there was no risk of hitting the dreaded 2,000 iteration limit. 

Note: This is needed because we can’t reference the DeveloperName of the responses’s related question record on the left side of a Get records element yet. Another added complexity is that we cannot directly reference CMT records like you can in Process Builder, so we have to retrieve them with Get elements.

Step 5

In the first decision we match up a Custom Metadata record that stores the Question Choice DeveloperName with the response’s Question Choice DeveloperName. Once we get that match, we move forward.

{!Loop_Over_CMT_Records.Question_Choice_DeveloperName__c} Equals {!Loop_Over_Found_Choice_Responses.QuestionChoice.DeveloperName}

Step 6

In the loop I check to see that the Response record’s related Choice Matches a particular outcome stored in the Custom Metadata field Question Choice Text. This is where we route the flow based on the automation we want.

{!Loop_Over_CMT_Records.Question_Choice_Text__c} Equals Request_Met

Step 7

From there, the world’s your oyster.  In my flow, I had 3 outcomes that did different things to the case. Remember how we used an assignment element for the Case earlier instead of an Update? That’s because we’re doing another update to it here:

Summary (In Words)

1. Check that the Response’s Survey ($Record.SurveyVersion.DeveloperName) is the one you want to automate.
2. Get all of the Survey Automation Mapping custom metadata records for the particular survey.
3. Get all of the SurveyQuestionResponse records that have a related Choice record.
4. Loop over the CMT Mapping records
5. Loop over the found SurveyQuestionResponse that are choices (loop within a loop only if the expected responses with choices is tiny)
6. In the loop, do a decision to match up the looped Custom Metadata record that stores the Question Choice DeveloperName with the 2nd loop’s Question Choice DeveloperName
7. In the next decision, create your decision paths for each of the results you want to handle using the Question_Choice_Text__c field on the custom metadata record
8. Do your thang!

Send an Alert to a Queue on a Poor Responses

Whew! 

Okay this one’s a bit easier. A lot easier, actually.

One nifty thing I’ve done is setting up an in-app alert to go to a specific queue when somebody responds poorly (1,2, or 3 out of 10) to a CSAT question. 

Step 1 

Grab any question responses that have a specific numerical value tied to it. Note that I use the Id of the record that triggered the Flow (the Survey Response’s record id)

Step 2

Do your null/empty check on this step ^

Step 3  (Optional)

Here we do some stuff to the case if needed.

I actually set the case status to ‘Problem’ if we found any matching results.

Step 4 (Optional)

Loop over the questions to store the question text

We Loop over the question responses to create a text variable of all of the question names for the alert we’ll make. This will go into the alert body.

The left side is a Text Collection variable of your choosing, the right side is: {!Loop_Over_Question_Names.Question.Name}

If you’re a big USF fan like me, you’ll have the handy ExtractStringsFromCollection action to do this for you, but if you don’t want to use that, you can go this route. 

Step 5 – Send the Alert

Next up is standard fare for using the Alert standard action in Flow. I made a post about it a year and a half ago on LinkedIn before I joined the UnofficialSF world: https://www.linkedin.com/pulse/lessons-learned-flows-send-custom-notification-action-adam-white/ 

Your Notification Body can be a text template that references your text collection variable of the questions that were responded to poorly.

Step 6 – Do your Last Updates

We finally update the Case with all the various updates made in various assignment elements throughout the Flow:

And you’re done! I’d love to hear everyone’s thoughts on this – I know it’s a bit complex but once you get it going it can really scale quite well and you won’t need to make edits to your Flows if you make text changes to your Surveys.