Conclusion: To improve call centre resources scheduling, some organisations have implemented software agents to either improve users’ experience and/or reach the right expert at the right time. However, self-service success depends on the quality of information available to the software agent and its analytical ability to provide reliable recommendations. Any deficiency in these resources will leave the software agent with no alternative but to call the live agents, thereby making the investment in agent technology questionable. Organisations should assess the software agent maturity and determine which level should be reached to fulfil the business imperatives. This note provides a self-assessing approach to address software agent shortcomings.

Observations: As part of digital transformation, various business sectors (e.g. Banks) have implemented software agents to enrich customers’ experience and provide a modern way to work with the organisation. For example, instead of calling or emailing the call centre, users can enter questions in a familiar chat window. As a result, the virtual agent searches the organisation’s various knowledge databases for answers. If an answer is not found, the chat session is routed to a live service agent who can assist while reviewing the transcript and looking for matching issues. Organisations are hoping to realise the following benefits from the software agent technology:

  • Call centre cost is reduced by receiving fewer routine questions.
  • Users can solve issues faster with modern, self-service tools.
  • Software agent contacts the right people based on the current situation.
  • Live service agents can address more sessions simultaneously.
  • Software agents cover after-hour support, whereby support staff follow-up the issues the next working day if need be.
  • Users and agents chat in their own languages with instant translations.

However, a software agent’s successful implementation requires advanced Artificial Intelligence analytical capability to understand clients’ motives and their preferences prior to providing recommendations. This requires the software agent to access the relevant databases to recommend alternative courses of action whenever needed. As a result, the call centre staff’s inability to resolve issues at the first point of contact will be minimised.

To respond to these challenges, IBRS has developed a software agent self-assessment maturity model. The maturity levels are defined as follows:

  • Level 1 – Basic: The software agent does not offer search facilities in addition to those of Google search engines. It is characterised by ill-defined processes, whereby the dependency on live agents remains high and the use of software agents becomes increasingly unpopular.
  • Level 2 – Menu-driven: The software agent is built on the foundation of pre-defined menus. Any inquiry that cannot be associated with a pre-defined solution is immediately routed to a live agent.
  • Level 3 – General-purpose oriented: At this level, the service processes are well defined. As a result, the software agent can address the general questions even if they are unsupported by predefined solutions. It can also analyse unstructured data according to predefined rules and determine the best answers to the caller’s inquiry.
  • Level 4 – Advanced: At this level, the software agent can provide multiple solutions to the same inquiry, highlight the advantages and disadvantages of each option and recommend the most reasonable one.
  • Level 5 – Optimised: At this level, the software agent has achieved its objectives in terms of improving call centres effectiveness, making it easier and quicker to address the caller’s inquiries and increasing the customers’ trust in the organisation that they are dealing with.

Software agents should be evaluated in terms of the following qualifiers:

  • Input: The ability to communicate with the software agent by text or voice, and from any device at any time.
  • Language: Can interact with the software agent in multiple languages as needed.
  • Databases connection: The software agent is connected to multiple databases that can be used to provide practical and relevant recommendations.
  • Principle-based: Service policies and business rules are embedded within the software agent to address general-purpose inquiries without using predefined solutions.
  • Ability to learn: The software agent can memorise previous findings for future reuse.
  • Client relationship: The software agent has the option to retain the caller’s contact details for proactive future updates if need be.
  • Customer satisfaction: The software agent verifies the caller’s satisfaction with the provided recommendations.
  • Live agent handover: The software agent can handover the inquiry to a live agent if no answer is found.
  • Analytics: The ability of the software agent to analyse unstructured data in order to reach a deeper understanding of the inquiry.

IBRS maturity model is illustrated as follows:

Maturity/Qualifier

Level 1
Basic

Level 2
Menu Driven

Level 3
General Purpose Oriented

Level 4
Advanced

Level 5
Optimised

Input

Text is only supported.

Text and voice are supported.

Text and voice are supported from any device and at any time.

Text and voice are supported from any device and at any time.

Text and voice are supported from any device and at any time.

Language

Multiple languages are unsupported.

Multiple languages are unsupported.

Multiple languages are supported.

Multiple languages are supported.

Multiple languages are supported.

Databases connection

The software agent is not connected to external databases. All knowledge is coded in the software agent itself or totally reliant on Google search facilities.

The software agent is not connected to external databases. All knowledge is coded in the software agent itself.

The software agent is connected to specialised databases relevant to the industry at hand.

The software agent is connected to specialised databases relevant to the industry at hand, as well as to other databases of general interest such as legislation.

The software agent is connected to specialised databases relevant to the industry at hand, as well as to other databases of general interest, plus the ability to provide multiple options to address the inquiry at hand.

Principle-based

None.

Only operates from a predefined menu.

In-addition to available predefined menus, it can address general inquiries by using the organisation’s policies and general practices.

Can answer hypothetical questions if need be.

Can answer hypothetical questions and analyse the probability of making it happen.

Ability to learn

None.

None.

Manual data gathering for basic reporting and trend analysis.

Store findings for future use.

Store findings for future use and keep the information up-to-date where applicable.

Client relationship

No further contact is made with the client after the call is closed.

No further contact is made with the client after the call is closed.

No further contact is made with the client after the call is closed.

Follow-up contact is made with the customer if new information is detected and worth communicating.

Follow-up contact is made with the customer if new information is detected and worth communicating.

Customer satisfaction

Customer satisfaction is not requested.

Customer satisfaction is not requested.

Customer satisfaction is requested.

Customer satisfaction is requested and found positive.

Customer satisfaction is requested and changes implemented if need be.

Live agent handover

No handover is made.

No handover is made.

Handover to a live agent is made.

Handover to a live agent is made.

Handover to a live agent is made.

Analytics

No analytical capability.

No analytical capability.

No analytical capability.

Analytical capability in place.

The analytical capability in place is appreciated by the customers.

Business impact

Software agents are unused.

Software agents are rarely used. Preference is to deal with live agents.

Software agents are frequently used.

Software agents are trusted.

Software agents have achieved their objectives.

Next Steps:

  • Define the software agent motives in terms of contact centre performance improvement and cost reduction.
  • Assess the capabilities of the software agents in place in accordance with the maturity qualifiers defined in the above table.
  • Determine which maturity level should be sought.
  • Measure the current performance to achieve the desired maturity level.
  • Focus on the improvement of quick wins that should be undertaken to reach the desired maturity level.


Wissam Raffoul

About The Advisor

Wissam Raffoul

Dr. Wissam Raffoul was an IBRS advisor between 2013 - 2021 who specialised in transforming IT groups into service organisations, with particular expertise in IT Service Management (ITSM), process optimisation, outsourcing and Cloud strategies, enterprise systems management solutions and business-centric IT strategies. Prior to joining IBRS in August 2013, he was General Manager strategic consulting in Dimension Data advising clients on applying technology to improve business performance. Prior to joining Dimension Data, he was a Vice President in Gartner/META Group and issued various research publications covering service delivery processes, centre-of-excellence models, managing outsourcing vendors, benchmarks, maturity models, IT procurement evolution and supply/demand models. In previous positions, he headed HP ITSM consulting Practice in Australia. He also acted as an infrastructure manager, reporting to the CIO at a number of large organisations in government and in the financial and petrochemical industries.