Skip to main content

What GermainUX Watches

Portal, contact center and cases

Find the Cause

One worked example

Alert and Act

Catch it next time

Set It Up

Portal, calls, cases and markers

Overview

Calls and cases spiked this week, and the support manager asks the question every contact center knows: did self-service break? The contact center sees more work arrive. The portal team sees no outage. Neither sees why customers who used to find their answer online are now on the phone. A case or call dashboard shows that volume moved, not why. The cause can be a portal release that breaks Knowledge search for one browser, a case form that fails to submit, a product problem that makes more customers ask, or a busy day. Each one needs a different team and a different fix. GermainUX connects the portal to the cost it creates. It watches what customers do on the portal, such as searches with no results, rage clicks, errors and failed submits, next to the calls, chats and cases they turn into. It compares the period with the days before, ranks the browsers, pages and releases that moved with it, and rules out the contact center and the platform when they stayed normal. Then it shows you the sessions behind the rise.
Calls and Cases RiseAggregate AnalysisPortal SignalLeading Factor or MarkerSession Replay
This page uses a customer Help Center on Salesforce Experience Cloud, cases in Salesforce Service Cloud, calls on RingCentral and chat on Genesys Cloud as the example. The same approach works for any self-service portal or website that GermainUX monitors, next to any contact center. To act on customers who are about to leave rather than on volume, see Salesforce Cancellation Prevention. When calls take longer instead of becoming more frequent, see Call-to-Case Monitoring.

Questions It Answers

  • Why did calls and cases rise this week?
  • Did self-service break, or did more customers have a problem?
  • Is it one browser, device, page or release, or every customer?
  • Which searches stopped finding articles?
  • Is the contact center slower, or is it getting more contacts?
  • Did the fix bring case volume back down?

What GermainUX Watches

GermainUX keeps each signal as a KPI, so the analysis can compare the portal with the contact center over the same hours.

Find the Cause

The investigation starts in the contact center and ends on the portal:
  1. Open Aggregate Analysis on calls or new cases, for the hours when volume rose. GermainUX compares the period with the days before.
  2. Check whether the contact center itself changed: call duration, agents on queue, the Salesforce instance. If they stayed normal, the extra volume came from somewhere else.
  3. Add the portal signals on the Relationships chart and in Related KPIs: searches with no results, rage clicks, user-facing errors, failed requests.
  4. Read the Leading Factors of the portal signal that moved. A segment classified DEGRADED shows which customers self-service failed for.
  5. Find the release marker, rule out what stayed normal, and open one session in Session Replay as evidence.
The example below comes from simulated data, and its numbers belong to the example.

Example: A Portal Release Breaks Knowledge Search on iPhones

At 10:00 UTC on a Friday, a Help Center release replaces the Knowledge search component. On Mobile Safari, the new component sends the search without its text, so every search returns no articles. Nothing errors. In this example, the support team takes more than twice as many calls between 10:00 and 16:00 as on the previous days.
1

Spot the rise

Open the analysis of the Ring Central Call transactions for 10:00 to 16:00 on the day. The support team took 42 calls, against an average of 16 in the same hours on the previous days.
Analysis of RingCentral calls to the support team on the afternoon of the release: 42 calls against an average of 16 in the same hours on the previous days, with the hourly calls rising on the Relationships chart

Support Calls Analysis - GermainUX

Run the same analysis on new cases: the Salesforce Case Status KPI, filtered to the New status, which records one fact per case with its origin. In this example, 90 cases are opened against an average of 43, and Leading Factors splits the rise by origin. Phone cases nearly triple to 38, Web cases rise by two thirds to 45, and Chat cases double to 7.
Analysis of new Salesforce cases on the afternoon of the release: 90 against an average of 43, with Leading Factors showing 38 Phone, 45 Web and 7 Chat cases, all above the previous days

New Cases Analysis - GermainUX

2

Check the contact center

Compare the average call duration with the previous days, and check the agents on the Genesys chat queue and the Salesforce instance status for the same hours. In this example, calls last as long as on other days, the queue has its usual number of agents and the instance status is OK all day. Agents aren’t slower and nothing is down: more customers are contacting support.
3

Find the portal signal

Add the Help Center’s searches with no results and rage clicks to the Relationships chart, and look for the signal that rises in the same hours as the calls. In this example, searches with no results rise from 10:00, and rage clicks on Search appear in the same hours. The Help Center has 21 rage clicks between 10:00 and 16:00, and none on any other day of the week.
Analysis of Help Center rage clicks on the afternoon of the release: 21 rage clicks between 10:00 and 16:00, with the Relationships chart showing none on the other days

Help Center Rage Clicks Analysis - GermainUX

4

Read what ranks

Open the analysis of the Help Center searches with no results. In those hours, 110 searches found nothing, against an average of 10 on the previous days. Leading Factors don’t point to one missing article: every search term spikes at once, from installation appointment to warranty extension. Search itself is broken.
Analysis of Help Center searches with no results on the afternoon of the release: 110 against an average of 10 on the previous days, with Deployment Run markers at 10:00 and 15:25 on the Relationships chart and Leading Factors showing that every search term spikes at once

Searches With No Results Analysis - GermainUX

To see who it’s broken for, compare all Help Center searches on Mobile Safari 18 with the other browsers on the Compare page. In those hours, a search on Mobile Safari 18 returns 0.159 articles on average, against 6.475 on the other browsers, and Top Differences flags search results as lower on Mobile Safari. It isn’t more iPhone traffic: Mobile Safari 18 runs 107 searches and the other browsers 120, but 104 of the 107 on Mobile Safari find nothing, against 6 of the 120 elsewhere.
Compare page of Help Center searches on the afternoon of the release: 0.159 search results on average on Mobile Safari 18 against 6.475 on the other browsers, with search results listed under Top Differences

Search Results on Mobile Safari vs Other Browsers - GermainUX

5

Find the release

On the Relationships chart, look for a release marker where the searches stop finding articles, and hover over it to see what changed. In the searches analysis above, a Deployment Run marker sits at 10:00, the hour the searches with no results jump. Its details show the Help Center release that changed the Knowledge search component, the search box and the search controller. A second marker at 15:25 is the rollback, and searches on Mobile Safari find articles again from then on.
Details of the 10:00 Deployment Run marker on the searches chart: release Help Center 2026.11.2, deployment DEP-3184, completed, which changed lwc/knowledgeSearch, lwc/searchBox and KnowledgeSearchController.cls, with the 15:25 rollback marker next to it

Release Marker Details - GermainUX

6

Rule out the rest

Check the status and response time of the search request, and the user-facing errors on the case form, against the previous days. In this example, the search request still succeeds as fast as usual: 227 requests take 0.337 seconds on average, against 0.333 seconds on the previous days. Errors stay at their usual level. That’s why no error alert fired. Other browsers, the case form, the Salesforce instance and the contact center all stay at their baseline. The search changed, for one browser.
Analysis of the Help Center's Knowledge search requests on the afternoon of the release: 0.337 seconds on average against 0.333 seconds on the previous days, with a flat response time on the Relationships chart

Knowledge Search Requests Analysis - GermainUX

7

Watch one session

Open the session of an iPhone customer from the afternoon in Session Replay. Look for the searches that found nothing, the repeated taps on Search, and the moment the customer gave up and went to Contact Support.
The extra work doesn’t end with the rollback: check how long new Web cases wait for an agent in Salesforce Case Status. In this example, they wait about twice as long as usual that afternoon. Fix the search component for Mobile Safari, release it again, and run the same analysis on the following days to confirm that searches, calls and new cases are back at their baseline.

Alert and Act

Each step of the example points to a signal you can catch sooner:
  • SLAs and outlier detection on the share of searches with no results, by browser, so a search that breaks for one segment raises an alert
  • Percent change SLAs on calls and new cases by origin, so the support manager hears about a rise the same hour
  • Synthetic user scenarios that search for a known article and submit a test case right after each portal release
  • Alerts to the portal team when rage clicks or user-facing errors rise on the Help Center, by email, Microsoft Teams or Slack
  • Scheduled reports on cases by origin next to portal searches with no results, for the support and portal teams

Set It Up

Each step adds value on its own. Start with the portal and the cases, then add the contact center.
1

Monitor the portal

Deploy monitoring for Salesforce Experience Cloud, or RUM JS for your website, so GermainUX records sessions, searches, clicks, errors and requests. Configure privacy for what customers type into the case form.
2

Collect the cases

Turn on Salesforce statistics monitoring for Case Resolution and Case Status. Case Status needs Field History Tracking on the Case Status field in Salesforce, or GermainUX gets no status history. Add instance status monitoring for your Salesforce instance.
3

Monitor the contact center

For RingCentral, deploy RingCentral monitoring in the agents’ browser. For Genesys Cloud, deploy Genesys Cloud monitoring for the queue metrics.
4

Relate the signals and show releases

Configure KPI relationships from calls and cases to the portal signals that explain them: searches with no results, rage clicks, user-facing errors and outbound requests. Show your portal releases as markers.
5

Set SLAs and alerts

Set SLAs on searches with no results and on calls and cases by origin, and alerts to the portal and support teams.

Additional Resources

Analysis Dashboard

Aggregate and instance analysis

Salesforce Experience Cloud

Portals, journeys and feedback

Use Cases

More Use Cases

Get Help

Every support organization has its own portal, channels and case process. The GermainUX team can help you relate your portal signals to your calls and cases, and set up the alerts that tell the right team first.

Contact GermainUX Support

The GermainUX team helps you get up and running faster, empower your team, turn data into actionable insights, and optimize performance to deliver measurable business outcomes.

Can't Find What You Need?

Search all GermainUX docs with Ctrl + K, or use the feedback buttons at the bottom of this page to tell us what is missing.

The content of this document is proprietary to GERMAIN SOFTWARE.