Telecom Customer Support 
Process Automation 
erwin.de.ley@isencia.be
Setting the scene 
We work with Belgium's largest telecom provider 
Traditional telecom services : 
• voice, internet, IDTV, mobile 
Nomadic services : 
• personal cloud, streaming, video-on-demand, gaming, … 
iSencia Belgium NV 2 
Millions of customers
Setting the scene 
Technical support & repair 
• 15K calls / day 
• 10K field interventions 
• 700 call center operators 
• 2500 field technicians 
After-sales, billing support 
• 30K calls / day 
• 1500 call center operators 
iSencia Belgium NV 3
Setting the scene 
The network is the service. 
Geographically distributed network, multitude of street cabinets, devices, … 
Many variations of complex products and line technologies 
Rapidly evolving products and services 
Long half-life of old products and devices 
iSencia Belgium NV 4
What was in the past 
iSencia Belgium NV 5
Old-style repair support process 
Rem.intervention 
Customer calls 0800-... 
Gets an operator (after a while) 
Operator tries to understand what the customer's problem is 
And starts consulting all kinds of BSS/OSS support systems, launching 
network measurements, retrieving equipment states... 
Manual analysis of the information 
On-the-spot diagnosis and resolution decision 
iSencia Belgium NV 6 
Call 
Manual 
Collect 
data 
Manual 
Analyse and 
diagnose 
Field intervention 
Solution
This doesn't help either 
Large organization with many separate teams 
Each historically had own tools and methods 
Customers get sent around between teams 
Operators and technicians are often subcontracted 
=> Volumes i.o. customer satisfaction 
BSS/OSS systems owned by many application teams 
Mix of legacy systems, new technologies, 
packages, in-house dvp ... 
iSencia Belgium NV 7
Issues 
Customer often fails to correctly describe problem 
Manual data retrieval and line measurements are slow 
Customer and operator are blocked waiting for results 
Approach and results depend on individual skills 
Long average call handling times 
Risk for errors & inconsistent support decisions 
Technician's tools inconsistent with operator decision 
Low first-time-right ratio 
Low customer satisfaction 
Huge learning curve and high recurrent training costs 
iSencia Belgium NV 8
Improving the support processes 
iSencia Belgium NV 9
Where do we want to go? 
iSencia Belgium NV 10 
Customer Experience 
• Consistent support answers and decisions 
• Improved FTR 
Simplification 
• Integrated view on customer, services, network 
• Automate steps where possible 
• Support the interactions with the customer 
Efficiency 
• Always access to up-to-date best practices & formalized knowledge 
• Shorter handling times 
• Less rework, training, knowledge transfer
How to get there 
DARE : Diagnosis And Repair Engine 
Single system for automated and interactive support processes 
• Call Center Operators for retail and professional customers 
iSencia Belgium NV 11 
• Field Technicians 
• OLO Operators and Technicians for regulated and commercial products 
• Customer Self Repair 
• Proactive repair, automated line profiling, ... 
Consolidated, up to date business logic with 360°customer view 
• Frequent business logic updates 
• Data collection and remote repair in the network, even in home equipments 
• Integrated data views and interactive guidance
How to feed it 
Call center coaches and field intervention experts extract 
support&repair cases and best practices 
iSencia Belgium NV 12 
Organize support tasks : 
• Automated data collection 
• Interactive forms 
• Analysis modules 
• Diagnostic decisions 
• Advises for (automated) repair actions
How to feed it 
Define and build a library of : 
• Data connectors to OSS/BSS systems 
iSencia Belgium NV 13 
• Analysis modules 
• Reusable sub-processes 
Define domain-specific support processes : 
• reusing connectors and modules 
• specializing diagnostic decisions
Example 
Refactoring the end-2-end process for repair support 
iSencia Belgium NV 14 
X 
DARE 
IVR 
1 
2 
3 
4 
4
Pretroubleshooting & 
troubleshooting 
While customer on queue : 
• automated pre-troubleshooting 
• data collection 
• rules-based problem analysis 
• root-cause diagnosis 
• identify and prioritize troubleshooting processes 
• advise on possible repair actions 
iSencia Belgium NV 15
Pretroubleshooting & 
troubleshooting 
When the operator gets the customer call : 
• diagnostic report available 
• detailed underlying data 
• guided interactive troubleshooting 
iSencia Belgium NV 16
Diagnostic report 
iSencia Belgium NV 17
Interactive guidance 
iSencia Belgium NV 18
Decision examples 
iSencia Belgium NV 19 
IVR 
Filtering and routing decisions based on 
• Global outages, planned works in the area ? 
• Customer info : Bad payer? SLA? … 
• Customer's products 
• Language
Decision examples 
Pre-troubleshooting process 
• Check customer's ordered products 
• Measure the physical line characteristics 
• Read network configuration, device status 
• Reach to customer modem and other CPEs 
• Check service configurations FIA, IDTV, VOIP, … 
• Compare configuration in the network and service management 
systems with expected values 
(e.g. activation of TV packages, VOIP features, ...) 
iSencia Belgium NV 20
Decision examples 
Troubleshooting process 
• Is the PTS diagnosis what the customer is calling for? 
• Can the issue be solved by the operator? 
iSencia Belgium NV 21 
• Must a ticket be created? 
• Is a field intervention required?
And now a bit more technical... 
iSencia Belgium NV 22
DARE system overview 
IVR 
Service 
Gateway 
BIQ 
E2E testing 
SALY UTS/eTS 
Callback 
manager 
Pretrouble shooting 
ABR 
Technical situation 
iSencia Belgium NV 23 
Backends 
NPS 
TBF 
Procodac 
Hlog & XDSL 
parameters 
SAFE 
expected attenuation 
ROSY 
(SALY GUI) 
DSLAM 4TEL 
BAS 
ABD 
FH 
NFH 
PSTE 
Customer facing 
serv/Assigned prod 
LAM 
Local Address 
Management 
SGW 
Modem Interaction 
Homer 
Skynet accounts 
SOCCABIS 
Skynet 
login/password 
DARE 
DARE Rules Engine DARE Process Engine 
DARE Manager 
Syst. 
Admin 
2 
STB 
HGW 
IMS 
NGN 
UMS 
… 
Product catalog PILA 
LLD 
OFA WFM 
Operator Cockpit 
CLI/CLE 
1 
Troubleshooting 
Console Advisor 
DATAMINER 
STB Info
DARE modules 
Built on iSencia's 
Passerelle EDM 
Orange boxes : extensions 
Process 
editor 
iSencia Belgium NV 24 
Passerelle 
process engine 
Jboss Drools 
rules engine 
Asset 
repository 
Application 
modules 
Backend 
adapters 
Task form entry 
raw data views 
Role-based 
security 
Process 
scheduler 
Application 
Preferences 
Execution 
traces 
Support 
modules 
Web 
Services 
(SOAP & REST) 
Operator 
Console 
Admin 
Manager 
Interface 
layer 
Backend 
simulator 
Diagnostic 
processes 
Analysis 
rules 
End-user Console 
Task form entry 
Web 
Services 
(SOAP & REST)
Asset repository 
• Process models, rules packages, images, translations … 
iSencia Belgium NV 25 
• 2-digit versioning schema 
• Packaging/grouping of related groups of assets 
• Root-level : Project 
• Simple preparation, activation, roll-back of new project releases 
• Export/import of project packages between test envs & PROD
Backend adapters 
iSencia Belgium NV 26 
• >150 adapters 
• Typically slow tasks (2s – 2min) 
• Asynch/synch bridge with internal buffering, resource pools 
(minimize internal resource blocking) 
• Configurable throttling settings to protect backends from overload 
• New/updated adapters can be uploaded without downtime
Application preferences 
iSencia Belgium NV 27 
• Stored in DB 
• Web UI to import/export, CRUD, … 
• DARE modules & services can be reconfigured “live” 
• Used for analysis thresholds, refdata, 
backend connection settings etc
Some usage scenarios 
iSencia Belgium NV 28
Collaborative process definition 
Workflow 
editor 
Task form entry 
raw data views 
iSencia Belgium NV 29 
Take advantage of : 
• Web-based access 
• Role-based security 
• Fully-featured graphical model editor 
• Asset versioning, personal workspace, 
model locking 
• Activate,update,rollback 
sets of related assets atomically 
• Test & simulation modes 
Passerelle 
process engine 
Drools 
rules engine 
Asset 
repository 
Application 
modules 
Backend 
adapters 
Role-based 
security 
Process 
scheduler 
Application 
Preferences 
Execution 
traces 
Support 
modules 
Web 
Services 
(SOAP & REST) 
Operator 
Console 
Admin 
Manager 
Interface 
layer 
Backend 
simulator 
Workflow 
processes 
Analysis 
rules 
End-user Console 
Task form entry
(Don't) work at night 
Workflow 
editor 
Task form entry 
raw data views 
iSencia Belgium NV 30 
For automated/pro-active/batch runs 
• Schedule runs day-and-night 
• Not just remote in location, but also in time ! 
• Integrate automated self-diagnostics 
in your workflows 
• Automated notifications when needed 
• Role-based security for secure support 
and collaboration 
• Connect to your sysadmin monitoring tools 
via adapters, web-services, event logs etc 
Passerelle 
process engine 
Drools 
rules engine 
Asset 
repository 
Application 
modules 
Backend 
adapters 
Role-based 
security 
Process 
scheduler 
Application 
Preferences 
Execution 
traces 
Support 
modules 
Web 
Services 
(SOAP & REST) 
Operator 
Console 
Admin 
Manager 
Interface 
layer 
Backend 
simulator 
Workflow 
processes 
Analysis 
rules 
End-user Console 
Task form entry
Execution monitoring & support 
Workflow 
editor 
Task form entry 
raw data views 
iSencia Belgium NV 31 
The platform enables : 
• Browse active workflows 
• Integrated execution tracing, 
provenance info 
raw data views, graphical execution 
traces, 
Task form replay ... 
• Secure pause/stop/resume 
• Collaboration between business 
analysts 
and IT support 
• Testing options : mock testing, 
replay etc 
Passerelle 
process engine 
Drools 
rules engine 
Asset 
repository 
Application 
modules 
Backend 
adapters 
Role-based 
security 
Process 
scheduler 
Application 
Preferences 
Execution 
traces 
Support 
modules 
Web 
Services 
(SOAP & REST) 
Operator 
Console 
Admin 
Manager 
Interface 
layer 
Backend 
simulator 
Workflow 
processes 
Analysis 
rules 
End-user Console 
Task form entry
Traceability approach 
• Each process execution gets unique ID, 
communicated to client apps 
• Each task's lifecylce is stored 
iSencia Belgium NV 32 
• With backend data, 
analysis results etc 
• Internal monitoring, 
timing & error statistics
Execution browser 
& raw data views 
iSencia Belgium NV 33
Graphical execution trace 
iSencia Belgium NV 34
Finally, log files! 
• All those application features are nice but... 
• In the end sys admins want log files! 
• Errors, timeouts etc get logged with specific error codes 
• Monitoring tools have their own rule-based decisions to react on error 
frequencies, assign tickets to application teams... 
iSencia Belgium NV 35
Lessons learned 
iSencia Belgium NV 36
Graphical tools: Sales pitch vs facts 
• Graphical process models, rules-based logic … 
often presented as a way to let non-developers build 
business solutions 
• We've noticed that this works for contained scopes 
• But for larger contexts, with long-term maintenance, 
performance requirements etc …? 
iSencia Belgium NV 37
Graphical tools: Sales pitch vs facts 
What works best in our situation : 
• Close collaboration 
• Process experts formalize the definitions 
• Functional experts turned developers maintain : 
• Interactive processes with simpler decision rules 
• Threshold values for complex decision modules 
• Reference data 
iSencia Belgium NV 38 
• IT team maintains : 
• Complex, high-throughput flows, modules 
• Builds adapters 
• & coaches the business developers
Agility and Testing 
• Built for fast implementation & maintenance cycles. 
But errors have large impacts. 
• Gradually evolved to traditional release-bound test cycles 
for non-minor changes 
• Live changes still allowed for : 
iSencia Belgium NV 39 
• Threshold changes 
• Ref data 
• Urgent rules fixes
Data issues 
• Data distributed across many applications, databases... 
• For repair processes we need live data from 
network, devices, ongoing provisioning flows, … 
• Confronted with many intermediate layers & conventions 
• Backend systems all have different backgrounds, data 
structures 
• Data quality issues become visible in your projects 
(as in all integration solutions) 
• Where to find the authorative data source 
(batch replications, plain duplication, ...) 
iSencia Belgium NV 40
Data issues 
• We store lots of data per diagnostic run 
(approx 150M data items / day) 
• Maintain 6 weeks of results in the operational DB 
• Older weekly partitions off-loaded to DWH 
• Decision analytics done there 
iSencia Belgium NV 41
Organization issues 
• Meeting culture, budget allocations etc all slow down the 
real implementation work 
• Many different teams are impacted 
• They may see your integration work as a threat 
• End-users will see your work as threat 
(your decision automation project is paid for 
by cost-cutting programs) 
iSencia Belgium NV 42
Thank you 
iSencia Belgium NV 43

Decisions Management use case : Telecom Customer Support Automation

  • 1.
    Telecom Customer Support Process Automation erwin.de.ley@isencia.be
  • 2.
    Setting the scene We work with Belgium's largest telecom provider Traditional telecom services : • voice, internet, IDTV, mobile Nomadic services : • personal cloud, streaming, video-on-demand, gaming, … iSencia Belgium NV 2 Millions of customers
  • 3.
    Setting the scene Technical support & repair • 15K calls / day • 10K field interventions • 700 call center operators • 2500 field technicians After-sales, billing support • 30K calls / day • 1500 call center operators iSencia Belgium NV 3
  • 4.
    Setting the scene The network is the service. Geographically distributed network, multitude of street cabinets, devices, … Many variations of complex products and line technologies Rapidly evolving products and services Long half-life of old products and devices iSencia Belgium NV 4
  • 5.
    What was inthe past iSencia Belgium NV 5
  • 6.
    Old-style repair supportprocess Rem.intervention Customer calls 0800-... Gets an operator (after a while) Operator tries to understand what the customer's problem is And starts consulting all kinds of BSS/OSS support systems, launching network measurements, retrieving equipment states... Manual analysis of the information On-the-spot diagnosis and resolution decision iSencia Belgium NV 6 Call Manual Collect data Manual Analyse and diagnose Field intervention Solution
  • 7.
    This doesn't helpeither Large organization with many separate teams Each historically had own tools and methods Customers get sent around between teams Operators and technicians are often subcontracted => Volumes i.o. customer satisfaction BSS/OSS systems owned by many application teams Mix of legacy systems, new technologies, packages, in-house dvp ... iSencia Belgium NV 7
  • 8.
    Issues Customer oftenfails to correctly describe problem Manual data retrieval and line measurements are slow Customer and operator are blocked waiting for results Approach and results depend on individual skills Long average call handling times Risk for errors & inconsistent support decisions Technician's tools inconsistent with operator decision Low first-time-right ratio Low customer satisfaction Huge learning curve and high recurrent training costs iSencia Belgium NV 8
  • 9.
    Improving the supportprocesses iSencia Belgium NV 9
  • 10.
    Where do wewant to go? iSencia Belgium NV 10 Customer Experience • Consistent support answers and decisions • Improved FTR Simplification • Integrated view on customer, services, network • Automate steps where possible • Support the interactions with the customer Efficiency • Always access to up-to-date best practices & formalized knowledge • Shorter handling times • Less rework, training, knowledge transfer
  • 11.
    How to getthere DARE : Diagnosis And Repair Engine Single system for automated and interactive support processes • Call Center Operators for retail and professional customers iSencia Belgium NV 11 • Field Technicians • OLO Operators and Technicians for regulated and commercial products • Customer Self Repair • Proactive repair, automated line profiling, ... Consolidated, up to date business logic with 360°customer view • Frequent business logic updates • Data collection and remote repair in the network, even in home equipments • Integrated data views and interactive guidance
  • 12.
    How to feedit Call center coaches and field intervention experts extract support&repair cases and best practices iSencia Belgium NV 12 Organize support tasks : • Automated data collection • Interactive forms • Analysis modules • Diagnostic decisions • Advises for (automated) repair actions
  • 13.
    How to feedit Define and build a library of : • Data connectors to OSS/BSS systems iSencia Belgium NV 13 • Analysis modules • Reusable sub-processes Define domain-specific support processes : • reusing connectors and modules • specializing diagnostic decisions
  • 14.
    Example Refactoring theend-2-end process for repair support iSencia Belgium NV 14 X DARE IVR 1 2 3 4 4
  • 15.
    Pretroubleshooting & troubleshooting While customer on queue : • automated pre-troubleshooting • data collection • rules-based problem analysis • root-cause diagnosis • identify and prioritize troubleshooting processes • advise on possible repair actions iSencia Belgium NV 15
  • 16.
    Pretroubleshooting & troubleshooting When the operator gets the customer call : • diagnostic report available • detailed underlying data • guided interactive troubleshooting iSencia Belgium NV 16
  • 17.
  • 18.
  • 19.
    Decision examples iSenciaBelgium NV 19 IVR Filtering and routing decisions based on • Global outages, planned works in the area ? • Customer info : Bad payer? SLA? … • Customer's products • Language
  • 20.
    Decision examples Pre-troubleshootingprocess • Check customer's ordered products • Measure the physical line characteristics • Read network configuration, device status • Reach to customer modem and other CPEs • Check service configurations FIA, IDTV, VOIP, … • Compare configuration in the network and service management systems with expected values (e.g. activation of TV packages, VOIP features, ...) iSencia Belgium NV 20
  • 21.
    Decision examples Troubleshootingprocess • Is the PTS diagnosis what the customer is calling for? • Can the issue be solved by the operator? iSencia Belgium NV 21 • Must a ticket be created? • Is a field intervention required?
  • 22.
    And now abit more technical... iSencia Belgium NV 22
  • 23.
    DARE system overview IVR Service Gateway BIQ E2E testing SALY UTS/eTS Callback manager Pretrouble shooting ABR Technical situation iSencia Belgium NV 23 Backends NPS TBF Procodac Hlog & XDSL parameters SAFE expected attenuation ROSY (SALY GUI) DSLAM 4TEL BAS ABD FH NFH PSTE Customer facing serv/Assigned prod LAM Local Address Management SGW Modem Interaction Homer Skynet accounts SOCCABIS Skynet login/password DARE DARE Rules Engine DARE Process Engine DARE Manager Syst. Admin 2 STB HGW IMS NGN UMS … Product catalog PILA LLD OFA WFM Operator Cockpit CLI/CLE 1 Troubleshooting Console Advisor DATAMINER STB Info
  • 24.
    DARE modules Builton iSencia's Passerelle EDM Orange boxes : extensions Process editor iSencia Belgium NV 24 Passerelle process engine Jboss Drools rules engine Asset repository Application modules Backend adapters Task form entry raw data views Role-based security Process scheduler Application Preferences Execution traces Support modules Web Services (SOAP & REST) Operator Console Admin Manager Interface layer Backend simulator Diagnostic processes Analysis rules End-user Console Task form entry Web Services (SOAP & REST)
  • 25.
    Asset repository •Process models, rules packages, images, translations … iSencia Belgium NV 25 • 2-digit versioning schema • Packaging/grouping of related groups of assets • Root-level : Project • Simple preparation, activation, roll-back of new project releases • Export/import of project packages between test envs & PROD
  • 26.
    Backend adapters iSenciaBelgium NV 26 • >150 adapters • Typically slow tasks (2s – 2min) • Asynch/synch bridge with internal buffering, resource pools (minimize internal resource blocking) • Configurable throttling settings to protect backends from overload • New/updated adapters can be uploaded without downtime
  • 27.
    Application preferences iSenciaBelgium NV 27 • Stored in DB • Web UI to import/export, CRUD, … • DARE modules & services can be reconfigured “live” • Used for analysis thresholds, refdata, backend connection settings etc
  • 28.
    Some usage scenarios iSencia Belgium NV 28
  • 29.
    Collaborative process definition Workflow editor Task form entry raw data views iSencia Belgium NV 29 Take advantage of : • Web-based access • Role-based security • Fully-featured graphical model editor • Asset versioning, personal workspace, model locking • Activate,update,rollback sets of related assets atomically • Test & simulation modes Passerelle process engine Drools rules engine Asset repository Application modules Backend adapters Role-based security Process scheduler Application Preferences Execution traces Support modules Web Services (SOAP & REST) Operator Console Admin Manager Interface layer Backend simulator Workflow processes Analysis rules End-user Console Task form entry
  • 30.
    (Don't) work atnight Workflow editor Task form entry raw data views iSencia Belgium NV 30 For automated/pro-active/batch runs • Schedule runs day-and-night • Not just remote in location, but also in time ! • Integrate automated self-diagnostics in your workflows • Automated notifications when needed • Role-based security for secure support and collaboration • Connect to your sysadmin monitoring tools via adapters, web-services, event logs etc Passerelle process engine Drools rules engine Asset repository Application modules Backend adapters Role-based security Process scheduler Application Preferences Execution traces Support modules Web Services (SOAP & REST) Operator Console Admin Manager Interface layer Backend simulator Workflow processes Analysis rules End-user Console Task form entry
  • 31.
    Execution monitoring &support Workflow editor Task form entry raw data views iSencia Belgium NV 31 The platform enables : • Browse active workflows • Integrated execution tracing, provenance info raw data views, graphical execution traces, Task form replay ... • Secure pause/stop/resume • Collaboration between business analysts and IT support • Testing options : mock testing, replay etc Passerelle process engine Drools rules engine Asset repository Application modules Backend adapters Role-based security Process scheduler Application Preferences Execution traces Support modules Web Services (SOAP & REST) Operator Console Admin Manager Interface layer Backend simulator Workflow processes Analysis rules End-user Console Task form entry
  • 32.
    Traceability approach •Each process execution gets unique ID, communicated to client apps • Each task's lifecylce is stored iSencia Belgium NV 32 • With backend data, analysis results etc • Internal monitoring, timing & error statistics
  • 33.
    Execution browser &raw data views iSencia Belgium NV 33
  • 34.
    Graphical execution trace iSencia Belgium NV 34
  • 35.
    Finally, log files! • All those application features are nice but... • In the end sys admins want log files! • Errors, timeouts etc get logged with specific error codes • Monitoring tools have their own rule-based decisions to react on error frequencies, assign tickets to application teams... iSencia Belgium NV 35
  • 36.
  • 37.
    Graphical tools: Salespitch vs facts • Graphical process models, rules-based logic … often presented as a way to let non-developers build business solutions • We've noticed that this works for contained scopes • But for larger contexts, with long-term maintenance, performance requirements etc …? iSencia Belgium NV 37
  • 38.
    Graphical tools: Salespitch vs facts What works best in our situation : • Close collaboration • Process experts formalize the definitions • Functional experts turned developers maintain : • Interactive processes with simpler decision rules • Threshold values for complex decision modules • Reference data iSencia Belgium NV 38 • IT team maintains : • Complex, high-throughput flows, modules • Builds adapters • & coaches the business developers
  • 39.
    Agility and Testing • Built for fast implementation & maintenance cycles. But errors have large impacts. • Gradually evolved to traditional release-bound test cycles for non-minor changes • Live changes still allowed for : iSencia Belgium NV 39 • Threshold changes • Ref data • Urgent rules fixes
  • 40.
    Data issues •Data distributed across many applications, databases... • For repair processes we need live data from network, devices, ongoing provisioning flows, … • Confronted with many intermediate layers & conventions • Backend systems all have different backgrounds, data structures • Data quality issues become visible in your projects (as in all integration solutions) • Where to find the authorative data source (batch replications, plain duplication, ...) iSencia Belgium NV 40
  • 41.
    Data issues •We store lots of data per diagnostic run (approx 150M data items / day) • Maintain 6 weeks of results in the operational DB • Older weekly partitions off-loaded to DWH • Decision analytics done there iSencia Belgium NV 41
  • 42.
    Organization issues •Meeting culture, budget allocations etc all slow down the real implementation work • Many different teams are impacted • They may see your integration work as a threat • End-users will see your work as threat (your decision automation project is paid for by cost-cutting programs) iSencia Belgium NV 42
  • 43.
    Thank you iSenciaBelgium NV 43