The usual metrics for the service level agreements are
This is a percentage of the number of calls that have been abandoned or have not been received over the number of calls that have been received. This is used for inbound services.
Abandonment rate = no of calls not taken / total number of calls that had come
The calls that have not been taken are defined in terms of the maximum wait time or the speed of answer. Each client specifies a time limit for the abandonment. If any customer waits for longer than this time and then disconnects it is called as an abandoned call. If the agent receives the call within the prescribed time limit the call is considered as taken. In case the customer is willing to wait for more than the wait time and the agent addresses the call it would still be considered as a valid call taken. Calls that are disconnected within the wait time are not considered as abandoned calls.
For e.g.: If the SLA says that the abandonment arte should not be more than 5%, it means that for every 100 calls that are received from the client not more than 5 calls should be abandoned.
The speed of answer is the time within which the call has to pick up. Any customer who holds on till the wait time ends and then disconnects is equivalent to call abandonment.
The wait time is calculated after the call is routed to the specific call center. The time spent in the queue waiting for the call to be taken up by the final call center person is the wait time. The time to be decided depends on the impatience index of the customers in their respective counties and also on the type of the service. Typically it is seen that the wait time for US is the smallest because of the assumption that the impatience index of people in US is very high.
The type of the service would mean the sector or the kind of service.
Banking, financial or technical support services have higher wait time than customer care support services This is because a person who has a technical problem would be much more patient than one who is waiting to ask some thing about a new launch in the customer care support.
The abandonment rate is usually determined based on the speed of the call.
For e.g.: Some companies have a speed of answer of 2 minutes. This means that the calls have to be taken up within 2 minutes.
The importance of this measure is in terms of quality. If the calls take a long time to be picked up it gives a wrong signal about the reputation of the client in front its customers.
The average handling time is the time taken to process a call on an average.
This time is exclusive of the wait time. It is calculated from the time the agent picks up the call.
Thus metric can be given b the client or the by the service provider to the agents.
The clients give the AHT in case they pay the service provider in terms of the call processing hours.
In most of the cases the average handling time is a metric that is internal to the service provider. Since in most of the companies the call charges are borne by the service provider they insist on the call handling time not exceeding a limit.
At the same time there would also be a lower limit below which the call handling time should not fall. This is very important also from the point of view of the service provider. In most cases the payments are made based on the number of calls that are processed. Hence in such conditions, when an agent takes a very long time to process a call, he would be missing out on call, which he could have otherwise processed.
Hence the service providers always keep a tab on the average handling time of each of the agent as well as of the whole call center. When the AHT of the whole center increase the call monitoring team tries to find out the
The average handling time depends to a large extent on the kind of service being provided. IN a technical support system the AHT would be around 10-20 minutes whereas in customer care operation it would be as low as 4-5 minutes.
This is usually a percentage of calls that would be taken by an agent as against the number of calls received.
Service level = No of calls taken by the agent/ Total number of calls received
Service level = 1 – Abandonment rate.
For e.g.: If the service level for a company is 80% it means that for every 100 calls received at least 80 calls would be taken.
Typically this rate is high in customer care services where the number of calls processed is high. This service level is usually given by the client and is mentioned as a part of the SLA
This metric is used in an outbound service. The number of calls made per hour would be total number of calls made by the center in an hour. The client would specify this number based on their business needs.
The center level the calls are split between the various agents at the call center level. Hence in a an outbound service an agent typically is aware of the number of calls that he ahs to make beforehand
This metric is also used only in the outbound services. This is applicable especially to a sales outbound service. The conversion rate would be the percentage of calls that have resulted in sales as opposed to the total number of calls made. The sales made are immediately noted down.
Conversion rate =
No of calls that were converted to sales/ total number of calls made
In outbound center where sales are not the main prerogative the metric is done in terms of contacts obtained per hour. The contacts would include the entire contact details of the customer called.
There are certain other metrics for SLA’s which are usually secondary to these above metrics
- FCR: First call resolution
This is applicable to technical support calls or services where some kind of action is needed from the other end. Many a time when the problem is not resolved in the first time the customer calls in again to
FCR gives the ratio of the number of calls for which the customer issue was totally resolved as a ratio of the entire alls received.
FCR = Calls in which the problem was solved in the first call/ total calls taken
A higher rate indicates a good service level. But there is a trade off between the AHT and FCR. Usually agents whose CR is high have a very hg AHT and the vice versa. Hence it’s the service provider’s prerogative to balance the two. This is because the AHT would affect its costs while the FCR would directly impact the service quality.
Dispatch rate
This is also a metric that is typically used in technical support service. The agent usually has to dispatch certain support parts to the customer
Call quality
The call quality depends on the opening and closing of the call problem recovery etc. The other measures used in outbound services re revenue per call, margin per call etc.
Hence it can be seen that the metrics depend on the following:
- Type of the service provider
A third party sourcer would have SLA’s given by the client whereas a captive unit would have the SLA’s provided by the company. Hence in the latter case the purpose would be customer service as well as cost reduction.
Technical support service would concentrate on metrics like AHT, FCR along with the other measures, whereas in customer care, the speed of answer would be an important metric. Outbound services o the other hand have a different set of metrics like sales per hour, contacts per hour etc. Email services would be based on the number of emails sent.
- Type of the business involved
Financial services would concentrate on different things as compared to an Internet service provider. Hence the type of the business also affects the SLA’s that would be given.
- Negotiating power of either parties
A very important factor affecting the SLA’s would be the negotiating power of the client and the service provider. Established client companies usually can demand very stringent service levels to be met. Hence when the service provider is not big enough the SLA’s are too a large extent governed by the clients. On the other hand if the service provider were a well reputed on they would have the power to negotiate the various SLA’s.
- Seasonality of the demand
Many a time the SLA’s are changed during times of high demand or peak load.
Hey are made more stringent during periods of high demand or peak load.
This is done to ensure that the customer servicing is at a high level.
Penalty given in case the SLA’s are not met
The SLA’s are measured over a certain period of time, which could be hour, daily, or weekly. Typically in most companies the SLA’s are measured on a daily basis.
In case of third part outsourcing the client keeps track of the days during which the service levels were not met. The penalty is taken as a percentage of the invoice.
For e.g.:
If a company has a service level of 95% and in a month it doesn’t meet this level on 2 days, and if the payment were made on a call basis, the service provider would not be payed for the 2 days during which the requirement was not met.
Capacity management
On a periodic basis, service providers are given an estimate of the calls that would be routed to them. The forecasting is done on a historic basis. The same situation is compared to the previous year’s scenario. This information could be provided on a monthly or a on a quarterly or on a weekly or on a fortnightly basis.
The service providers then staff their centers based on the demand forecasting. Usually they keep a safety stock of about 10 percent in order to catch up for spikes in the demand
and absenteeism.
The estimate of the staffing is done using the
The classical M/M/N queuing model, also called the Erlanger-C model, is obtained by further assuming Poisson arrivals, exponentially distributed service times, and no abandonment. It is the model most often used in call-center analysis, but it has one glaring defect: Call abandonment is not a negligible or minor aspect of call-center operations.
The Square Root Rule for Safety Staffing
We consider the M/M/N model with abandonment added.
Let R = λ/μ denote the (average) offered load, where λ is the average call arrival rate and 1/μ_ is the mean call duration. (R is measured in units of service duration per unit of time.) The principle is as follows: For moderate to large values of R the appropriate staffing level required =
N = R +β√R
Where β is a positive constant that depends on the desired level of service? Of course, in practice the value of N derived from this formula must be rounded to an integer. The second term on the right side of the equation above may be described as the excess capacity needed, beyond nominal requirements (the first term), to achieve the target service level in the face of stochastic variability. The equation above shows that the required excess capacity grows less than proportionately with the load of calls to be handled. This phenomenon is aptly described as statistical economies of scale.
Client – Vendor selection
The clients never source their entire call volumes to just one company. Similarly from the service provider’s end, they do not service 100% of the call volumes of the client. The reasons for the same are discussed further. But at the same time there is economies of scale involved in this are high. For a larger volume of calls the safety stock requirement would not be directly proportionally larger.
Let us consider a hypothetical situation:
There are 2 clients A and B. Client A services all its calls through vendor V.
Client B on the other hand services the calls equally through 4 vendors’ v1, v2, v3 and v4.
(All the four vendors are assumed to have the same capacity and same skill level of the employees).
Case 1:
Client A:
Arrival rate of the calls →λ
Average handling time →1/μ
Hence the staffing required → N1 = (λ/μ + β√λ/μ)
Case 2:
Client B:
Arrival rate of the calls →λ/4
Average handling time →1/μ
Hence the staffing required → n1 = (λ/4μ + β√λ/4μ)
The sum of the total agents required =
N2= n1+n2+n3+n4 = 4(λ/4μ + β√λ/4μ)
→λ+2β√λ/μ
Hence in the second case N2 is larger than N1 by β√λ/μ. Hence the costs incurred for the agents are higher in case 2.
A vendor for the same reason would prefer not taking in a small chunk of the service.
Service provider’s perspective
Why would they prefer sourcing less percentage of calls?
Most of the companies do not source 100% call volumes of their clients. This is especially very true with small service providers or newly established call centers. They feel that this way they would be mitigating the risk of the client’s business not doing well. In case the service providers service a large chunk of the call volumes of the client, they invest a lot in the manpower. The costs involved for each agent is very high as they have to go through extensive training and are typically given a lot of incentives due to the nature of the industry. Hence in case their client business does not do well they lose a lot on the
Client’s perspective
If the payment to the service provider is based on number of calls then the client would not be bothered about giving the entire volume to one vendor. Giving the entire volume to a single vendor would once again make the client extremely dependent on the vendor and would increase the risks involved. I some cases a client tries out two or here companies to figure out the best company and then gives a higher percentage of calls to that vendor.
Payments
The payments are made by the clients on different basis. For inbound the payment is done usually on a per call basis. For outbound the payment could be sales per hour or calls per hour etc.
Most of the payments are transaction based and are not fixed in nature. In a dynamic environment like a call center where in the demand variability is high, transaction based system makes more sense. The customer can derive value from the call center through the number of calls. The service provider tends to be performance oriented when the measure is on a per call basis. The service provider can benchmark the number of calls as against their performance measures. This in turn would help them in the remuneration of the employees.
One of the companies had a system wherein the client paid the service provider on a per agent basis. These were in those cases where in special agents were recruited for the particular project. At this point of time the agents were not trained. But once the raining got over and the agents were skilled enough the system was hanged to transaction based system. This way the service provider will not be at the losing end n the initial stage where the agents were not skilled enough.
Contractual process
In any BPO industry there are contracts between the BP industry and the outsourcing firm. These contracts might be obtained through bidding or normally. The size of a call center is determined by the peak demand and not by the average demand. Hence when the companies bid there is a kind of problem that exists due to the inability to predict the actually demand. As a part of the project we will try and figure out how contracts are signed up in the BPO companies.
If the client is an established one, they outsource the process of choosing the vendors to a consulting company. They come in the stages of negotiations and final bidding process .
If the service provider is an established the clients go up to them, offer their presentations and negotiate the deals. Hence here the negotiating power of either player is an important factor in determining the contract terms and conditions.
Performance metrics
The performance metrics varies from company to company. They vary based on the type and the policies of the company.
They maybe broadly classified under the following heads:
Business outcome Metrics: This measures the quality of service from a results perspective.
Service metrics: These measures the health of a process using the information provided both by the client and the BPO service provider.
Transformation Metrics: This is used to measure the progress or a planned change.
These above metrics are measured using the following parameters
- General: Cost of the project and client satisfaction
- People: Turnover, experience and performance reviews of people working on the project
- Capacity: Maximum value that can be attained by the BPO provider without disrupting other services.
- Reliability: Number of gas in service and recovery time by the BPO.
- Efficiency: Increase in output for the same resources.
- Maintenance Services: Time to respond by phone and the time taken to resolve a problem by the BPO.
- Repetitive process: The error rates and the amount of re-work.
- Call answering: Wait time, the percentage of inquiries successfully resolved on the first call and the satisfaction level of the client
- Deliverables: Timelines and conformance with the specifications.
Attrition and HR staffing
One of the serious concerns of the BPO industry is the rate of attrition especially in the inbound and the outbound center. The attrition rate becomes even more difficult in a real time operation like the call center. Even if one agent is not present during his time, the load on the other agents becomes high. Hence it’s very important of the companies to maintain the optimum required capacity levels at all times.
The attrition rate is high pure for the reason the nature of the work is very demanding and pressure driven. Most of the employees who join this company are passed out students who join the company for the sake of earning quick money. Some of the employees leave the call center once the training period is over. During the period of April and May the fallouts are higher because many students apply for higher studies and get selected.
Company Study
Company A
Introduction
Their main job is consulting but they are also involved in making software solutions.
They have a BPO and a call center. The BPO outfit does only the documentation part. They send mails and replies to different kinds of queries that they get. The call center handles the calls received for the client.
Call center
Their clients are different kinds of service providers. These provide services to many national and international companies with a global presence to provide a seamless and integrated service. In the case of ISPs, the service provided is computer troubleshooting, people call when they have problems with their Internet connection. The operation runs across India, Philippines, and Belgium.
There are two groups within the call center – DSL and Dial up .The former group deals with the calls that come for trouble shooting in the leased line section whereas calls of the customers who have problem with their dial up connection
Each client has a dedicated set of human resources who work on their project. At the same time, multiple projects were going on at the same time, which might have the taskforce rotated on the projects which are deemed similar. The personnel aren’t rotated on assignments, which may be diverse in nature
Processing the call
The call is processed at the site, which is Tier 1, and if the query is not satisfied then the call is escalated to Tier 2 , which is the service provider itself.
There is a documentation to be done for each call. First the billing telephone number is asked for, then it is transferred to the billing department f the account is suspended
The customer when he calls is answered by an IVR, where he gives all details. The call is processed or transferred to sales for disconnect if the customer asks for it.
Minimum standard failure
There are five such standards
- read the opening script of pleasantries
- identify the name of the caller
- identify the caller and take a call back number
- apologize for any errors
- restart the issue
On the basis of these performance scorecards, the agents are evaluated and the quality is assessed by the quality department or by the customers through a survey sent out to them which is called POSE, where they are rated from 1 to 5 where 4 and 5 amounts to 100 points and 1 to 0 points.
Also the first call resolution (FCR) is an indicator of whether the same problem is solved within the first attempt and no recalls have to be made again on them in 72 hours.
Flowchart of A’s capacity utilization process
Demand forecasting
Call processing
Typical call addressed by the agent
Issue resolution flow
Company E
Introduction
The company is a captive unit for its US parent company which operates financial services for corporate and individuals for a host of product lines.
SLA’s used
- To fulfill tax returns with zero defects
- Turnaround time is the budget allocate to any project which ranges from 4 to 15 hours
- Timelines are the deadlines against which the individuals have to perform
- Timely quality delivery is an essential work aspect
Working of the projects
Product lines, which the company operates: corporate, people, partnership, financial services, individual tax returns, and expatriate tax returns
The working of a project works with a core group with the requisite skills for the project at hand and there is a crossover group, which is picked up to meet deadlines. The effort for any project is to have major people with core skills in it employed in the project.
Capacity utilization
Capacity managers are the schedulers which work with skill set and available work hours of the employees as inventory and make schedules to maximize the capacity utilization for each skill level. They also allocate people to projects not only on basis of their timelines and skills but also recommendation soft their superiors.
Their tax filing system is complicated by the existence of over 5000 taxing jurisdiction sin us at the state, federal and county level.
Training programs
Hence the smooth functioning of the company is possible only through a comprehensive training and retraining system in place for new recruits and old employees. The measures of doing that range from training programs on induction , upgraded courses on a continual basis ,lunch-n-learn systems and click and learn systems through an elaborate IT system in place for the same.
The training focuses on both hard and soft skills with the first year having 400-600hours of training.
Employee remuneration and feedback
Feedback survey given to users in US ranks the people on their technical skills and their people skills which has a bearing on their remuneration based on a rating going positively progressively from 1 to 3.teh grade of the return is set with people’s salaries having base component and an incentive scheme,. For people working beyond 40 hours, a salary of one and a half time she pay is added.
Contracts
The contracts for the company are made in the US and delivery center is in India.
Company D
Introduction
The call centers in D service their own customers and do not offer these services to any outside client.
The call centers in D serve three main functions
- Technical support
- Customer support
- Sales support
- Collection
The calls could of two types – inbound or outbound. The calls from the collection department are of the outbound type whereas the other three are inbound in nature.
Customer care support
This function is spread over India and Philippines. In Indian there are two centers – Hyderabad and Bangalore. The total number of employees working for the customer care support in India would be around 200.
The customers could be from US, Europe, Canada or Middle East. Within a call center there are different groups that service the different region customers.
The function of the customer care service is to clear the doubts of the customer regarding the various credit rating plans. The company provides products to the customers based on credit. Each customer would be approved of the credit policy depending upon his credit rating. This rating is obtained related to the banking credit systems.
Organization hierarchy
The employees in the call center work in different groups. Different groups could service different customers and different functions. Each group has about 8-10 employees. They are headed by a team lead. A manager heads around 5-6 team leads.
The job of the team lead is to monitor the functioning of the group. The team lead also provides with the statistics of his group every week. The statistics would include information about the average time of the calls, number of calls handled by the group in that week and the various quality control dimension parameters.
There is a department, which continuously monitors the calls of the employees.
Call processing
The average time for any call would be around 280 seconds. This would include the average holding time and the average call time. The average waiting time for the customer whose call has been routed to this call center would vary from 20 to 40 minutes.
Each employee works for 8 hours each day with three breaks – two aux and one main lunch break. The average number of calls processed by any employee is around 80 per day.
The work timings for the groups vary depending upon the kind of customers they service. Around 5-6 groups service the US clients and about 4-5 services the Europe clients. The timing for the group handling the US customers is between 7 in the evening to 6 in the morning. This corresponds to the US bank timings of 9 to 6.
Demand Fluctuations
Within a day the demand is usually the highest in the mornings in the specific country. For e.g. for US clients the demand is maximum around 8 to 12 in the night.
Within a week the demand is the lowest in the weekend and the employees do not work during weekends.
In a month the demand shoots up whenever the due date is nearing. The due date is common for many customers especially during a scheme.
For e.g. if there was scheme A which had a last date of 30 of the month, most of the customers would avail of the offer towards the end i.e. around 28th to the 30th of the month
Hence from the next month their due date for the payment of the credit would be 28th of the month. Hence under such circumstances the number of queries shoots up when the due date is nearing.
Typically the demand is high towards the end of the month.
Quality deployment
The employees are evaluated based on the following quality parameters
- Introduction
- Communication skills
- Problem Solving
- Resolution of the problem
- Closing of the call
- Customer satisfaction.
Calls are audited in random fashion and the above-mentioned quality parameters are evaluated for each of the calls audited. The employees are scored on a basis of 1 to 5, 5 being poor performance and 1 being very high performance levels. Employees who score in the range of 1to 3 are given incentives whereas the employees scoring around 4-5 are sent to Care Bays for development.
Human resources
The attrition rate is very high in D. It is around 30%.
Company T
Introduction
SLA used
This varies according to the type of the program used
- Inbound program: average handling time, average speed of answer, call abandonment rate, service level and call quality
- Outbound programs: sales per hour, conversion rate, contacts per hour, abandonment rate and call quality
- SLAs vary by nature of service/product
- They may be half hourly in case of inbound programs
- They are hourly in case of outbound programs
- SLAs may be monthly depending on the metric involved
- The delays in the SLAs may be classified into
- Controllable: due to improper/incomplete planning
- Uncontrollable: due to third party/client reasons
- The penalty of not meeting SLA is a %age of the invoice
Seasonality
- The seasonality existing in the demand during the day, month, week and year can vary with the product/service
- Banking support could have volume variation due to salary credits-predictable
- ATM outage-unpredictable
- Telemarketing-depending on the nature of the product sold
HR and staffing
- Internal incentives depend upon the longevity and long term performance
- Monthly rewards, incentives and recognition depend on meeting the SLA and the internal efficiency metrics such as productivity
- The staffing variation:
- UK programs between 13:30 -1:30 (IST) i.e. 9 am to 9 pm (BST)
- US programs between 18:30 -9:30 (IST) i.e. 9 am(EST) to 9 pm (PST)
- For telemarketing in a B2C environment peak staffing is between 17:00 to 21:00
- For telemarketing in a B2B environment peak staffing is between 9:00 to 17:00
- Manpower variation is accounted for by proper planning for it on basis of historical attrition numbers and new requirements on the basis of new projects
- The training:
- Accent training
- American / British/Canadian culture
- Customer service skills
- Sales skills
- Telephone skills
- Keyboard skills
- Product training
- System training
- There is a safety stock kept but this is not dictated by the clients
- The clients typically have a set of dedicated resources for themselves
Contracts
- Drafted by operations/legal department under the guidance of the CEO
- Largely client jurisdiction is applicable for dispute resolution
- The biding process typically involves a RFP and varies from client to client
- The company adheres to the COPC industry benchmarks
Analysis of the operational issues
SLA
Type of service
Captive vs. Third party outfit
Scale of the service provider
Company E
(Large-scale company)
Capacity management: The Company has a robust system of capacity utilization, which includes personnel in core skill areas complemented by the crossover agents, which are used as the buffers to meet timelines for projects effectively.
Thus, a larger scale of operations enables the manpower allocation across projects more easy.
Company A
(Large-scale company)
Capacity management: The smoothening of schedules can be done via effective capacity planning more effectively in the case of enterprises on a larger scale. Also, similarly due to the scale of operations and the similarity of projects, which require minimal retraining costs, it is easy to enable the personnel to shift from the projects DSL to Dial.
Negotiating power: The scale of operation globally has a bearing on the contract setting. Due to better negotiating power, the SLA is negotiated so as to suit the service provider.
Share of project pies: the company owing to its economies of scale takes up projects with a much larger share of its size. This is feasible since the company can provide economy sizes of return to scale and management is easier for bigger and less number of projects. thus, it ensures better capacity utilization.
Company T
(Small-scale company)
Capacity management: The set of resources for each project in this case is dedicated. The company has a complicated set of interrelated SLAs as fallout of its comparatively smaller scale of operations to meet the industry standards.
Share of project pies: the company owing takes up projects with around 15-20% share of its size. This is feasible since the company has lesser scale of operation since,
- It would prefer to hedge its risks in projects over a much bigger range
- the experiential curve learning would be more broad in nature with a portfolio of projects
- the manpower for a bigger chunk, hence the negotiating power for the same might not be possible for the company to achieve a bigger share of the project.
Conclusions
The BPO outfits pertaining to company A.E,T and D were studied in depth. The data collection phase involving SLA implications, capacity management issues due to seasonality and outages and the contractual processes was undertaken keeping in mind the confidential nature of these companies. Through the analysis the key differences that arose in the functioning were explained by:
The nature of the call center program
- Outbound
- Inbound
- Back office
This had a bearing on the SLA structure and payment schemes of the company and could be sufficiently affected during the contractual process by the scale of the BPO outfit and whether it was captive or non captive in nature. Hence, the issues related to how these companies manage their service levels has a strong implication on its operational strategies.
Appendix 1
SLA’s
- What are he type of SLA’s used
- What is the interval of the SLA?
- What are the specifications with the client?
- Are there any delays in the SLA’s?
- What is the cost involved in the SLA’s?
- What kind of seasonality exists in the demand during the day, month, week and year?
- What are the penalties of not meeting the SLA’s?
HR and Staffing
- What are the internal incentive schemes given by the HR department?
- What are the incentive schemes given by the client to the organization?
- What is the staffing variation?
- How the staffing is maintained during the peak hours and during the lean period?
- How is the attrition maintained?
- IS it very high?
- What does an employee who joins newly have to go through? (in terms of training)
- Is there any safety stock for staffing?
- Does the attrition have any effect on the SLA’s?
Payments
- How are the payments made?
- Is the payments transaction based or agent based?
- What is the trend that is being followed? (transaction based or agent based)
- What would a service provider prefer?
Other
- What is the main objective ( to reduce attrition or to increase the calls)
- What is the usual variability in the demand?
- How is the utilization rate?
Performance metrics
- What are the performance metrics?
- How are they measured?
Clients
- Do the clients have a dedicated set of resources?
They have a dedicate set of resources.
- If there is no dedicated set of resources how are the calls serviced?
Contracts
- How are the contracts made?
- What is the bidding process like?
- What are the improvements made on the contracts?
- Does the client pose any kind of improvement schemes?
Front end
- What are the different types of calls that are made?
- IS there any time limit for the calls?
- What are the monitoring systems?