How ADEM Calculates Experience Score
The Experience score gives you an indication of the health of a Mobile User,
Application, or Remote Site.
| Where Can I Use
This? | What Do I Need? |
|---|
Autonomous Digital Experience Management (ADEM) uses
Experience Score to report application
performance so that you can gauge the experience quality of your users and take action
if the score falls below a certain level.
How is the End to End Experience Score Calculated
For both mobile users and remote sites, ADEM calculates the
experience score for an app domain using synthetic metrics collected through
application tests.
For mobile users only, ADEM can also calculate the experience
score for an app domain using Browser-Based Real User Monitoring (RUM) metrics.
Synthetic Metrics Score:To calculate experience score from
synthetic metrics for both mobile users and remote sites,
ADEM uses two key metrics:
Time to First Byte (TTFB) and
Availability.
Availability:
TTFB, composed of:
DNS Resolution Time
TCP Connect Time
SSL Connect Time
Server Response Time
When
the experience score drops below
70,
ADEM identifies
the root cause by analyzing the following segments synthetic test scores:
ADEM calculates the synthetic metrics
score every 5 minutes. If ADEM detects poor performance, it
will associate a single root cause with the application for the next 5-minute
window, until the issue is resolved.
Browser-Based RUM Metrics Score
ADEM evaluates mobile user experience with the
following RUM metrics:
Largest Contentful Paint (LCP): Measures the time taken
to render the largest visual element in the viewport.
Interaction to Next Paint (INP): Assesses interface
responsiveness by measuring how quickly the app responds to user
interactions (e.g., clicks or keypresses).
Availability:
Browser-Based RUM Experience Score Calculation
If Availability = 0: The application score is set to
0.
Otherwise, the score is the minimum of LCP Score and INP
Score.
Combined Experience ScoreWhen both Synthetic and RUM metrics are available, the
experience score uses only RUM metrics.
When RUM metrics are unavailable, the experience score uses only
synthetic metrics.
Experience Score Calculation for Remote Sites
For remote sites, ADEM monitors paths based on
application forwarding policies configured on Prisma SD-WAN devices.
ADEM runs synthetic tests on all active and backup paths per
application and calculates the experience score per path, per application, and per
site, which rolls up from all remote sites as the organization experience score.
Remote Site Experience Score
The remote site experience score is an average of all test sample results that are
collected from individual applications monitored for that remote site.
Organization Experience Score
The organization experience score for remote sites is an average of all test sample
results that are collected from individual applications monitored on all remote
sites.
Experience Score for Each Path
For each path (active and backup path) that is discovered by ADEM based on the
application forwarding policy configured on
Prisma SD-WAN devices,
ADEM calculates the experience score using
synthetic metrics.
Application Experience Score
The application experience score per remote site is an average of test samples
collected from all active paths. For example, if an application for a remote site is
monitored on two active and two backup paths, the average of test sample results
from all active paths will be considered as the experience score for that
application monitored on that remote site.
How is the Segment Wise Experience Score Calculated
When the experience score drops below 70, ADEM identifies the root cause by
analyzing the following segments:
For network segments such as LAN, ISP and secure access gateway, ADEM uses
dynamic baselines for score computation. ADEM independently baselines three
fundamental metrics across the core segments of the service delivery chain
Baseline Generation and User Clustering
ADEM dynamically baselines network segments by capturing the normal
historical behavior of each network segment (LAN, Internet, PA Gateway) for clusters
of users with similar variables like location, ISP (ASN), and Prisma Access gateway.
These baselines dynamically adjust based on factors such as geo-location and service
providers. The following are the clustering variables for each segment:
| Segment | Clustering (Grouping) Variables |
|---|
| Internet (ISP) | User Location (City) + ISP (ASN) + Prisma Access
Gateway |
| LAN | User Location (City) + LAN Gateway IP |
| Prisma Access | Users accessing the particular Prisma Access Gateway |
Baselines for latency, jitter, and packet loss are automatically computed
using 30 days of historical data, segmented by different variables for various
network segments. These baselines are then refreshed every 24 hours to account for
changing network conditions or patterns.
ADEM flags deviations from the dynamic baseline rather than fixed
thresholds. These deviations for latency, jitter, and loss are computed and marked
as "Fair" or "Poor" to indicate segment health. ADEM dynamic scoring is percentile
based.
Minimum Data Requirement and Fallback Mechanism
The dynamic baseline requires a minimum of 20 agents per cluster. If this threshold
is not met, ADEM uses a fallback hierarchy.
