customer usage data trending & customer health - looking for advice

Zohar Gillai

Hi all,
We are embarking on a project to collect customer usage data via Pendo and other systems.  We would like to create customer health indicators using the data that would be automatically updated, with historical data kept.

Has anyone done this? and if so, where do you keep the data?  I was thinking of a time-series database, and also what do you use for visualization? 

Thank you, Zohar

Brian Hartley

Hey Zohar - I would recommend speaking to @Brian LaFaille .  He and I had a very similar conversation last week about pulling data from many systems under one hood.

Bri Adams

Hi Zohar,

Great question and a very worthy endeavor!  We also collect our customer usage data and specifically tie it into our ChurnZero instance where we can collect and analyze that data (and make it actionable). In particular, when we are tracking and trying to proactively determine our customer's health, we have a few specific usage indicators we look for to calculate our health score. After all, we are firm believers that if a customer is not engaging with our platform, that is a massive risk.

Full disclosure, I work for the company whose software I use for this (ChurnZero), but the only reason I even know about Customer Success software tools is because about three years ago I was managing a small customer success team and we needed something to help our own team with our renewal process, determining the health of our clients, and we needed somewhere to house all of the data streams we were collecting so we could make sense of them. After some due diligence, ChurnZero was selected as the software of choice, and after a bit of a whirlwind, I left a year later to join the team here. 

Depending on your data streams, role, and the needs of your CS org, it might be time to start considering a dedicated CS tool - especially one that integrates with Pendo so you can reach all of that rich data. I could not be happier that we have one, as it allows me to track our health scores, usage data, CRM data, engagement (emails and logged tasks, etc), in one place where all of the data can be actionable.

Or - are you specifically asking more for business intelligence type tools for the product team to have insight into usage trends?

Thanks - Happy to dig in anywhere if you'd like more information!
Bri

Jeff Breunsbach

Hey @Zohar Gillai - we're working on pulling together a PowerBI dashboard with several data sources that map back to a customer.

Some of the data files we'll be using...

• Product data (from their tenant)
• Salesforce data
• Community data
• Training & Education data

This will be a start but starting via Excel pulls and PowerBI until we can find the right mechanisms to automate over time.

James Conant

That is exactly what we did. Ive found that starting with down and dirty and then improving the process and tools thru gen 1, gen 2, etc is a practical way to develop a thorough understanding of data sources, its usefulness, its state, and how to best analyze, report, and act on it.

Zohar Gillai

Hi Bri,

 

Thank you,  we are primarily looking at usage indicators as a measure for customer health.  It would be good to expand it to BI for our product teams but that's secondary at this point. Our customer success team uses Gainsight at the moment and this effort I am embarking on is really for their benefit. But I'd like to be able to do trending over time and have signals based on deltas in usage and other indicators outside the product.

 

I like that ChrunZero integrates with Pendo and I'll need to check it out.

 

Thanks, Zohar

 

Zohar Gillai

SVP Customer Enablement & Support | Agari

zgillai@agari.com | 408.930.6274

Trust Your Inbox™

 

Zohar Gillai

Thanks Jeff,

 

We are starting with google sheets (I would prefer excel) for the collection.  We are looking at:

 

• Product data (usage and log-in)
• Salesforce data
• Zendesk data  (tickets)
• Emails (from an engagement point of view – are they communicating with us

 

I think the best place to store such data is a time-series database and to put a BI tool on it, but we are just starting to think about which tools to use. I will share as we progress.

 

Thanks, Zohar

 

Zohar Gillai

SVP Customer Enablement & Support | Agari

zgillai@agari.com | 408.930.6274

Trust Your Inbox™

 

 

• Product data (from their tenant)
• Salesforce data
• Community data
• Training & Education data

This will be a start but starting via Excel pulls and PowerBI until we can find the right mechanisms to automate over time.

Zohar Gillai

Hi James,

 

100% agreed.  Are you still at that stage? Or have you chosen tools yet?

 

Thanks, Zohar

 

Zohar Gillai

SVP Customer Enablement & Support | Agari

zgillai@agari.com | 408.930.6274

Trust Your Inbox™

 

 

Todd Ilberg

@here - this is an awesome thread and an adventure that we are actively navigating today. Would love to get some insight from the group on the sync between a CS platform/tool and SFDC.

Thanks,
Todd

Todd Ilberg
VP Customer Success & Solutions, stensul
t: 646-897-6052 | e: todd.ilberg@stensul.com
150 West 25th Street, 3rd Floor, NYC

Reimagine What's Possible with Email.

Dan Balcauski

Hi Zohar,

I'm curious about your insistence on a time-series database for this use case. You should be able to handle this fine with a SQL-based data warehouse (e.g. Amazon Redshift). While Pendo will need to collect and present very granular time-series metrics, for your use case, summarized monthly or weekly customer data points are sufficient.  

Does that help?

-Dan

Ed Powers

@Zohar Gillai, I suggest you do some basic statistical analysis before you spend a lot of time, money and effort integrating data sources. Factor analysis is often sufficient to screen which data are useful for your health dashboard and which are not. Then you can more systematically collect only the data you need. You can later refine your health dashboards using regression or more advanced predictive analytics to improve accuracy. 

Assumptions that usage, login frequency, trouble tickets, email response rates, or any number of other factors (x1, x2, x3, ...xn) impact customer behaviors (y) are mathematically testable hypotheses. I teach a course on how use statistical tools in Customer Success that describes how to do it by just using Excel. This might be of interest to you. I'm also happy to help you analyze what you have now so you can construct better health indicators much faster. If that's of interest, email me at ed@se-partners.com and we can set up a call to talk about it.