User analytics vs Active customers vs User segments

Hi @ryanne.perry - I think one major factor in the discrepancies you’re seeing is that the Engagement dashboard is looking at users at a company-level.  So that will always be different from what you see in the User overview and in Segments.  For any other discrepancies (outside of synch gaps related to what’s in Salesforce) I think we’d need to get specific about the data points and their definitions.  As you say you’re looking to measure ‘Active Accounts and Users’, what I would suggest is to take the number from the Engagement dashboard as your ‘active accounts at company level’ and then I’d take the ‘All unique visitors’ from the Audience dashboard as a count of total active users.  We can get creative with the Moderation view, the User overview and Segments for reporting, but they’re not really built for that - so personally I try to stick to the analytics dashboards whenever possible.

@Kenneth R there are some real discrepancies in the reporting that I have raised. This is largely due to how users are recorded (from what I have seen). It appears cookies are used to store a visitor_id for each user. However, this can change. People actively block cookies, they use different devices and incognito mode will also create different visitor_ids.

I am taking part in the BI Connector beta and put this query together to demonstrate that the visitor_ids are not 1 to 1 or even the likely number of devices used to 1. I have some users with hundreds of different visitor_ids. The query is below. Try it out on here and you will see what I am talking….

 

with usersPlusVisitorIds as (
select u.user_id,
u.user_name,
pv.visitor_id 
from "AwsDataCatalog"."{community_id}"."user" u join "AwsDataCatalog"."{community_id}".pageview pv 
on u.user_id = pv.user_id
group by u.user_id,
u.user_name,
pv.visitor_id )


select user_id,
user_name,
count(*) as visitor_ids
from usersPlusVisitorIds
group by user_id,
user_name
order by visitor_ids desc

With the BI Connector I can accurately report on everything our logged in users do, so I am using this now. We don’t have anonymous users at the moment, but this will require a different way of identifying a unique user.

On a point related to anonymous users, it should be pointed out that the reporting suggests we have hundreds of anonymous “users”….even though they are not able to get anywhere beyond the login page. I have raised this as well since an entity (computer or human) getting to the login page and no further is not really a user. However, they are classed as users in terms of the users that usage limits are measured against.  
 

Thank you for sharing @rhall !  You’re stretching my personal technical limits slightly, but I get your points.  My understanding from over the years is that quite a lot of web analytics is an inherently imperfect science.  In the old days we used Google Analytics for everything, and I know that GA (and other similar tools) also has many challenges with cookies getting blocked, etc…  When utilising cookies, a ‘unique user’ can never mean an actual unique human - for all the reasons you mention.  It’s a helpful (and imperfect) approximation.  I’m not trying to say that the way we do it is perfect or can’t be improved, by the way.

And it is awesome that you’re already using the BI connector!  It’s exciting to be able to see exactly what specific logged-in users are doing.  And there’s so much more that we will be able to do with the data lake.  

Also, I hear you on the anonymous users - it’s a fair point!

@Kenneth R I guess what I was suggesting is that since I can identify unique logged in users with the data Gainsight is making available to us, it might be nice if the reporting were to be updated to use this as well. The anonymous users (as I said) will be harder to identify, but they are not so important when it comes to granular data. However, the total number of unique users (since it is a metric that is in our contracts) really should be tightened up so that we don’t see what I am seeing with my closed community. 

If I had a grudge against another Gainsight community user (I don’t and I wouldn’t do this anyway) I could quite easily set up a system to repetitively load their login page again and again, with different visitor_ids over several weeks. I could blow their unique user limit in no time.

…..I now suspect that if anyone does do this, I have set myself up to be blamed for this 🤦🤣

Thanks @Kenneth R - maybe the better question is, what is the difference between active customers in the dashboard vs Salesforce? When I look at Active Accounts via the Salesforce integration dashboard the number is WAY higher than the Active Accounts in Community dashboard?

Are you able to break down what counts as active in Community dashboard vs Salesforce?

@ryanne.perry unfortunately we are facing issues with SFDC integration. We are actively working on a robust solution that will be applied across all our Gainsight products. Hopefully, in the near future, we can offer a much better experience with our in-app dashboards.

Thanks @Aleksandra - can you then confirm what counts as activity in the Community dashboard?

Hi @Aleksandra, are you saying that the integration from Community <> Salesforce is completely unreliable (meaning it shouldn’t be used)?

@ryanne.perry Hi, we checked the integration for your community, and everything seems to be working fine. If you still see any issues with data, please raise a support ticket with more details.