Detailed Description
The scheme provided by the specification is described below with reference to the accompanying drawings.
The key factor positioning method for customer service level fluctuation provided by one or more embodiments of the present description can be applied to the scenario shown in fig. 1. In fig. 1, a user may request a service from a customer call center through a corresponding terminal. The request channel may include, but is not limited to, a short message, a telephone, or an APP. When a user requests service, the customer call center allocates a corresponding person or machine (generally called service cell two) to provide service for the user. The data obtained by serving cell two in the process of serving the user may be referred to as traffic data.
It should be noted that, in order to ensure the quality of the service provided by the customer call center, the service level (customer service level for short) of the customer call center is usually periodically evaluated. In one implementation, the customer service level may be assessed by one or more of the following: average Talk Time (ATT), Average post-Call operating Time (aftercall Work/aftercall Wrap-up, ACW), absolute satisfaction, research satisfaction, First-Call resolution (FCR), slew rate, check rate, throw rate, and success rate. Wherein, ATT refers to the average duration of the online conversation between waiter two and the user. ACW refers to the average length of time a task or job needs to be completed by service waiter two immediately after the call is completed with the user. The absolute satisfaction degree refers to the occupation ratio of clicking a satisfaction key by a user after the telephone is ended. Survey satisfaction refers to the percentage of users who are satisfied with the service in the questionnaire. The FCR is the percentage of the calls which are not received by the user within 24 hours after the telephone service is finished. The transfer rate refers to the ratio of the service Xiao II to the service of the other service Xiao II which is handed over by the user after the service Xiao II is served. The check rate refers to the proportion of the service category that the service Xiao II meets a plurality of service requirements of the user in the current service and checks a plurality of service categories. The ten thousand delivery rate refers to the number of complaints generated in each ten thousand service. The universal rate is the number of complaints that are generated in each universal service.
It is understood that any of the above criteria, when subject to fluctuations, will result in fluctuations in the level of customer service. However, what factors cause the index to fluctuate? First consider a reason that may be a business aspect, such as a service scenario, etc. Secondly, since the service provided by the service may be a person, the human variability will usually affect the above index. Therefore, the scheme provided by the specification can analyze the fluctuation of the index through two aspects of personnel and business. For example, factors characterizing business attributes (e.g., service scenarios, etc.) and personnel attributes (e.g., personnel distribution, etc.) of each index may be collected separately. The collected factors are respectively used as influence factors of multiple dimensions of each index. In addition, in order to more finely locate the reason for the fluctuation of the customer service level, the present specification may split the influence factor of each dimension into a plurality of factors based on the corresponding business logic. Wherein, a plurality of factors corresponding to the influence factor of each dimension can form a factor set.
The fluctuation of the customer service level may be a difference in polygons drawn in the radar map based on index values of each index of two periods (e.g., a current period and a previous period). The period here may be one day, one week or one month.
It should be noted that, the solution provided in the present specification is to locate the factor of each index fluctuation based on the factors in the above factor set. And then positioning the key factors of the customer service level fluctuation according to the factors of the index fluctuation.
FIG. 2 is a flowchart of a method for locating key factors of customer service level fluctuation according to an embodiment of the present disclosure. The method may be performed by the customer call center of fig. 1. As shown in fig. 2, the method may specifically include:
step 202, traffic data in the current period is acquired.
For example, the traffic data may be obtained from a background database at the customer call center. The period here may be one day, one week or one month. And the traffic data may include, but is not limited to: service decimal identification, service date, call duration, service category, and the like.
At step 204, at least one indicator for assessing a level of customer service is determined.
The indicators herein may include, but are not limited to: ATT, ACW, absolute satisfaction, research satisfaction, FCR, slew rate, check rate, ten thousand throw rate, ten thousand percent, and so on.
It should be noted that the index in this specification may have influence factors of multiple dimensions. Such as service scenarios and people distribution, etc. The impact factor for each dimension corresponds to a set of factors. The factors in the factor set may be obtained by refining the influence factors of the corresponding dimensions based on the corresponding business logic. The influence factors with three dimensions of a certain index are respectively expressed as: for example, the factor a, the factor B, and the factor C, the factor sets corresponding to the influence factors of each dimension can be respectively expressed as: { a1, a2, …, an }, { b1, b2, …, bm }, { c1, c2, …, ck }, wherein a1, a2 and the like represent factors in a factor set, and n, m and k represent the number of factors in each factor set respectively.
It should be understood that the influence factors of different indexes and corresponding dimensions may be the same or different. In addition, the same influence factors of different indexes may be the same or different in the corresponding factor sets, and this specification does not limit this.
And step 206, determining the contribution degree of each factor in the factor sets corresponding to the index to the fluctuation of the customer service level according to the telephone traffic data.
In one implementation, the contribution of a factor may be determined as follows: and according to the telephone traffic data of the current period, counting factor current period index values, factor current period magnitude ratio and current period total index values. And acquiring the index value of the last period of the factor, the magnitude ratio of the last period of the factor and the total index value of the last period. And determining the contribution degree of the factor to the fluctuation of the customer service level according to the factor index value, the factor magnitude ratio and the total index value of the two periods.
In one example, determining the contribution degree of the factor to the customer service level fluctuation according to the factor index value, the factor magnitude ratio and the total index value of the two cycles comprises the following steps:
determining the contribution degree of a certain factor to the fluctuation of the customer service level according to the following formula: the contribution degree is (factor current period index value-factor current period magnitude ratio-factor previous period index value-factor previous period magnitude ratio)/(current period total index value-previous period total index value).
It should be noted that the calculation method of the factor current period index value, the factor current period magnitude ratio, and the current period total index value is described later. The definition of the factor previous period index value is similar to that of the factor current period index value, the definition of the factor previous period magnitude ratio is similar to that of the factor current period magnitude ratio, and the definition of the previous period total index value is similar to that of the current period total index value. It can be understood that the related data of the previous period may be recorded in the corresponding storage unit after the statistics of the previous period are good, and then directly read from the storage unit.
And step 208, selecting factors with larger influence from the factors in the factor sets according to the contribution degree.
In one implementation, the factors in each factor set may be sorted in order of increasing contribution. And then, selecting the factors in the top order from the factor sets as the factors with larger influence.
In another implementation, the factor with larger influence can be selected through the following steps:
and a, selecting influence factors from each factor set according to the contribution degree, thereby obtaining a plurality of influence factors.
The influencing factor here may refer to the factor with the largest contribution degree in each factor set. For example, it can be expressed as: ai. bj and cl, etc., where ai represents the set of factors: the most contributing factor in { a1, a2, …, an }; bj represents the set of factors: the most contributing factor of { b1, b2, …, bm }; cl represents the set of factors: { c1, c2, …, ck }, the most contributing factor.
And b, calculating the similarity of the multiple influencing factors pairwise.
For example, the similarity between ai and bj, ai and cl, and bj and cl may be calculated, respectively.
In one example, the similarity may be calculated by calculating a Jaccard similarity coefficient between the influencing factors. Taking the calculation of the similarity between ai and bj as an example, the calculation formula of the similarity can be expressed as: the similarity is min (the contribution of the ai, bj intersection to the overall fluctuation/the contribution of ai to the overall fluctuation, and the contribution of the ai, bj intersection to the overall fluctuation/the contribution of bj to the overall fluctuation).
It should be noted that the contribution degree of ai to the overall fluctuation is the contribution degree of ai to the service level, and the contribution degree of bj to the overall fluctuation is the contribution degree of bj to the service level, and the two contribution degrees are calculated in step 206. For the contribution of the intersection of ai and bj to the overall fluctuation, if ai is a middle section and a little two, and bj is a stolen scene, the contribution of the middle section and the little two to the fluctuation of the customer service level in the stolen scene can be understood. The specific calculation method can refer to the contribution calculation formula in step 206, and details are not repeated herein.
And c, determining factors to be refined from the multiple influencing factors according to the similarity.
Specifically, for the plurality of influence factors, it may be determined pairwise whether the similarity is greater than a threshold. And if the similarity between certain two influence factors is not larger than the threshold value, determining the two influence factors as the factors to be refined. And if the similarity between certain two influence factors is greater than a threshold value, selecting the influence factor with larger contribution degree from the two influence factors, and determining the influence factor with larger contribution degree as the factor to be refined.
The principle of comparing the similarity with the threshold may be as follows: if the similarity of the two influencing factors is not larger than the threshold value, the overlapping degree of the two influencing factors is smaller. And because the two influencing factors can be the reason for influencing the fluctuation of the current index, the two influencing factors are both necessary to be further refined, and both the two influencing factors are selected as the factors to be refined. Similarly, if the similarity of the two influence factors is greater than the threshold, it indicates that the overlap of the two influence factors is relatively large, and the influence of the two influence factors on the current index is also almost the same. Therefore, only the influence factors with larger contribution degree are selected for refining, so that the number of the factors to be refined is reduced, and the positioning efficiency is improved.
Now, with reference to the actual scenario, the following steps a to c are exemplified: assume that one set of factors is: { front section small two, middle section small two, rear section small two }, and the contribution degree of the middle section small two is the largest; another set of factors is: { stolen scene, cheated scene }, and the contribution degree of the stolen scene is the largest. Then the similarity between the middle-sized bar and the stolen scene can be calculated. If the similarity is more than 50%, judging whether the contribution degree of the middle section of the small second is larger, and if so, refining the middle section of the small second; and if not, refining the stolen scene. And if the similarity is not more than 50%, refining the middle-section Xiao-II scene and the stolen scene.
And d, refining the factors to be refined so as to obtain refined factors.
Here, the factor to be refined may be refined by specifying a factor in the factor set, or the factor to be refined may be refined by selecting the target factor set and then refining the factor to be refined by a factor in the target factor set.
When the target factor set is selected, the selection process of the target factor set may be: and determining the information condition obtained after the factors to be refined are refined by the factors in other factor sets. And selecting a target factor set from other factor sets according to the obtained information condition. And refining the factors to be refined by using the factors in the target factor set so as to obtain the refined factors.
The other factor sets herein refer to the factor sets except the set to which the factors to be refined belong in all the factor sets corresponding to the current index. Taking all the factor sets corresponding to the current index as: { a1, a2, …, an }, { b1, b2, …, bm }, { c1, c2, …, ck }, wherein the factors to be refined are: ai for example, the other set of factors may be { b1, b2, …, bm } and { c1, c2, …, ck }.
In one example, the above information situation may be determined by the following formula.
Information increment ∑ -p (xi) ln (p (xi))
Wherein xi is a refined factor, and when the used factor set: each of { b1, b2, …, bm } treats refinement factors: when ai is refined, xi is respectively as follows: aib1, aib2, … and aibm, wherein the number of x is m. Similarly, when the factors are set: each factor in { c1, c2, …, ck } treats refinement factors: when ai is refined, xi is respectively as follows: aic1, aic2, … and aick, wherein the number of x is k. Assume a set of factors: each of { b1, b2, …, bm } treats refinement factors: and after ai is refined, the obtained information increment is larger. Then { b1, b2, …, bm } may be chosen as the set of target factors, and the refined factors are: aib1, aib2, …, aibm.
And e, calculating the contribution degree of the refined factors to the fluctuation of the customer service level.
The factors after the refinement are as follows: aib1, for example, its contribution to the fluctuation of customer service level can be understood as the contribution of the intersection of ai and b1 to the overall fluctuation. The method for calculating the contribution of each refined factor may refer to the contribution calculation formula in step 206, and will not be described herein again.
And f, selecting factors with larger influence from the factors after the thinning according to the contribution degree of the factors after the thinning.
For example, the refined factors may be sorted in order of contribution from large to small. And selecting the refined factors which are ranked in the front and have the sum of the contribution degrees larger than a threshold value as the factors with larger influence.
As in the previous example, assume that the sorted refined factors are: aib1, aib2, aib3, …, aibm, and assuming that the sum of the contributions of aib1, aib2 and aib3 is greater than 80%, aib1, aib2 and aib3 are selected as the larger influencing factors.
Now, with reference to the actual scene, the following steps d to f are exemplified: the factors to be refined are assumed as follows: the middle section is two, and the target factor set is: { stolen scene, cheated scene }, a scene with a large contribution degree to customer service level fluctuation in each service scene of the middle-section-two can be extracted as a large influence factor.
It is understood that the above steps 206 to 208 are repeated to obtain the factors having larger influence corresponding to each index.
And step 210, summarizing and calculating the factors with large influence corresponding to each index to determine key factors of customer service level fluctuation in the current period.
Specifically, the total number of times that the different major factors respectively occur may be counted. And selecting the factors with the largest total times and larger influence as key factors of customer service level fluctuation in the current period. For example, suppose that for the index ATT, the selected factors with larger influence are: a1b2, a2b 3; for the index of absolute satisfaction, the selected factors with larger influence are as follows: a1b 2. There are therefore two major factors in total: a1b2 and a2b3, the total occurrence times of the two are respectively: 2 times and 1 time. Therefore, a1b2 is determined as a key factor in the fluctuation of customer service level in the current cycle.
In summary, in the embodiment of the present specification, a single factor is used to control and screen a factor with a higher priority, and when performing the factor refinement analysis, the information entropy is used to prune the factor lower level disassembly, and the Jaccard similarity coefficient evolution algorithm is used to prune the same level factor. The method can simultaneously find the reason of the biggest influence of personnel and business factors on the fluctuation, and key factors can not be omitted or misjudged.
The calculation method of the factor current period index value, the factor current period magnitude ratio, and the current period total index value in step 206 will be described below.
Before the above calculation process is executed, index values of a certain index may be calculated from multiple dimensions according to traffic data of a current period, so as to obtain index values of multiple dimensions. And carrying out exception processing on the obtained index value according to a Grubbs (Grubbs) test method and the statistical standard deviation. And determining corresponding factors from the factor set corresponding to the influence factors of the dimension for the index value of each dimension after exception processing. And then, according to the corresponding relation between the index value of each dimension and the factors, determining the factor current period index value, the factor current period magnitude ratio and the current period total index value of all the factors in the factor set corresponding to the influence factor of the dimension.
Determining an index value of an index from the dimension of the personnel distribution, or taking the influence factor of the current dimension as the personnel distribution, wherein the corresponding factor set is as follows: for { front section, middle section, and back section, small two } as an example, the process of determining the correspondence between the index value of the dimension and the factor may be as shown in fig. 3. In fig. 3, the following steps may be included:
step 302, counting the index value of a certain index serving two in each call service of the current period.
Taking the index as ATT for example, assuming that 100 call traffic is collected in advance, the ATT value of the corresponding service level two in each call traffic may be counted here.
And step 304, calculating the average value and the standard deviation of the corresponding index values for different service factors.
As in the previous example, the fraction of traffic in a 100-call session may be equal in service. And for the same service level two, determining a corresponding average value and a standard deviation according to the ATT values of the service level two in the multi-channel traffic.
Step 306, for the same service level two, determine whether the corresponding index values are outside the three standard deviation ranges before and after the mean value. If so, go to step 308; otherwise, step 310 is performed.
And 308, eliminating index values outside the three standard deviation ranges before and after the mean value.
And 310, establishing a corresponding relation between each index value and each factor in the factor set.
If the index value is smaller than the difference between the mean value and the standard deviation of the corresponding service decimal fraction, marking the index value as the front section decimal fraction; if not, judging whether the index value is larger than the sum of the mean value and the standard deviation of the corresponding service measure two, and if so, marking the index value as a front section measure two; if not, the index value is marked as the middle section and the small section.
After each index value is marked as described above, the correspondence between each index value and the factor can be obtained.
It is assumed that the correspondence between each index value and the factor is as shown in table 1.
TABLE 1
Service waiter
|
Index value
|
Factors of the fact
|
Zhang three
|
10s
|
Front section of the second
|
Zhang three
|
50s
|
Middle section of the second
|
Li four
|
15s
|
Front section of the second
|
Wangwu tea
|
100s
|
Back section xiao di |
For the example of the corresponding relationship in table 1, assume that the factors are: the front segment is smaller by two, then the factor current period index value is: (10s +15s)/2 ═ 12.5 s. The total index value of the current period is as follows: (10s +50s +15s +100s)/4 ═ 43.75 s. Factor the current cycle magnitude is 50% 2/4. That is, the ratio of the magnitude of the current period of the factor may be the ratio of the number of times of the factor service to the total number of times of the factor service in the factor set to which the factor belongs.
In summary, by the key factor positioning method for customer service level fluctuation provided by one or more embodiments of the present specification, factors influencing customer service level fluctuation can be found from multiple dimensions such as personnel distribution and service scenes, and key factors cannot be missed or misjudged. In addition, the scheme of the specification introduces influence factors of multiple dimensions, so that the practicability and the effectiveness are better. Finally, because the algorithm model provided by the specification is relatively low in coupling, a person skilled in the art can introduce more influence factors at any time, and therefore the expansibility of the algorithm model is better.
Corresponding to the method for locating key factors of customer service level fluctuation, an embodiment of the present disclosure further provides a device for locating key factors of customer service level fluctuation, as shown in fig. 4, the device may include:
an obtaining unit 402, configured to obtain traffic data in a current period.
A determining unit 404 for determining at least one indicator for evaluating the customer service level, the indicator having a plurality of dimensions of influence factors, each dimension of influence factors corresponding to a set of factors.
The indicators herein may include one or more of: average call duration ATT, average post-call working duration ACW, absolute satisfaction, investigation satisfaction, first-question resolution FCR, roll-over rate, check rate, ten thousand throw rate, ten thousand success rate and the like.
The determining unit 404 is further configured to determine, for each index, a contribution degree of each factor in the plurality of factor sets corresponding to the index to the customer service level fluctuation according to the traffic data.
The determining unit 404 may specifically be configured to:
and for each factor in the factor set, counting the index value of the current period of the factor, the magnitude ratio of the current period of the factor and the total index value of the current period according to the traffic data of the current period.
And acquiring the index value of the last period of the factor, the magnitude ratio of the last period of the factor and the total index value of the last period.
And determining the contribution degree of the factors to the fluctuation of the customer service level according to the factor index values, the factor magnitude ratio and the total index value of the two periods.
The determining unit 404 may further specifically be configured to:
and according to the traffic data of the current period, respectively calculating the index values of the index from multiple dimensions, thereby obtaining the index values of the multiple dimensions.
And performing exception processing on the index value according to a Grabbs test method and the statistical standard deviation.
And determining corresponding factors from the factor set corresponding to the influence factors of the dimension for the index value of each dimension after exception processing.
And determining the factor current period index value, the factor current period magnitude ratio and the current period total index value of all factors in the factor set corresponding to the influence factor of the dimension according to the corresponding relation between the index value of each dimension and the factors.
The selecting unit 406 is configured to select a factor having a larger influence from the factors in the factor sets according to the contribution degree determined by the determining unit 404.
The selecting unit 406 may specifically be configured to:
and selecting the influence factors from each factor set according to the contribution degree, thereby obtaining a plurality of influence factors.
And calculating the similarity of the multiple influencing factors pairwise.
And determining factors to be refined from the multiple influence factors according to the similarity.
And refining the factors to be refined so as to obtain refined factors.
And calculating the contribution degree of the refined factors to the customer service level fluctuation.
And selecting factors with larger influence from the factors after the thinning according to the contribution degree of the factors after the thinning.
The selecting unit 406 may further specifically be configured to:
and determining the information condition obtained after the factors to be refined are refined by the factors in other factor sets.
And selecting a target factor set from other factor sets according to the obtained information condition.
And refining the factors to be refined by using the factors in the target factor set so as to obtain the refined factors.
The selecting unit 406 may further specifically be configured to:
and judging whether the similarity is greater than a threshold value for a plurality of influence factors pairwise.
And if the similarity between certain two influence factors is not larger than the threshold value, determining the two influence factors as the factors to be refined.
And if the similarity between certain two influence factors is greater than a threshold value, selecting the influence factor with larger contribution degree from the two influence factors, and determining the influence factor with larger contribution degree as the factor to be refined.
The selecting unit 406 may further specifically be configured to:
and sorting the factors after the thinning according to the sequence of the contribution degrees from large to small.
And selecting the refined factors which are ranked in the front and have the sum of the contribution degrees larger than a threshold value as the factors with larger influence.
The summarizing unit 408 is configured to summarize the factors with large influence corresponding to the indexes selected by the selecting unit 406 to determine key factors of customer service level fluctuation in the current period.
The summarizing unit 408 may be specifically configured to:
and counting the total times of the occurrence of the different factors with larger influences.
And selecting the factors with the largest total times and larger influence as key factors of customer service level fluctuation in the current period.
The functions of each functional module of the device in the above embodiments of the present description may be implemented through each step of the above method embodiments, and therefore, a specific working process of the device provided in one embodiment of the present description is not repeated herein.
In the device for locating the key factor of customer service level fluctuation provided in one embodiment of the present specification, the obtaining unit 402 obtains traffic data in a current period. The determination unit 404 determines at least one index for evaluating the customer service level, the index having a plurality of dimensions of influence factors, each dimension of influence factors corresponding to a set of factors. For each index, determining section 404 determines, based on the traffic data, the degree of contribution of each factor in the plurality of factor sets corresponding to the index to the fluctuation in customer service level. The selecting unit 406 selects a factor having a larger influence from the factors in the factor sets according to the contribution degrees. The summarizing unit 408 performs summarizing calculation on the large-influence factors corresponding to the respective indexes to determine key factors of the customer service level fluctuation in the current period. Therefore, the accuracy of positioning the key factors of the customer service level fluctuation can be improved.
The key element locating device for customer service level fluctuation provided by one embodiment of the present specification can be a module or unit of the customer call center in fig. 1.
Corresponding to the method for locating the key factor of customer service level fluctuation, an embodiment of the present specification further provides a device for locating the key factor of customer service level fluctuation, as shown in fig. 5, where the device may include: memory 502, one or more processors 504, and one or more programs. Wherein the one or more programs are stored in the memory 502 and configured to be executed by the one or more processors 504, the programs when executed by the processors 504 implement the steps of:
and acquiring the traffic data in the current period.
At least one indicator for assessing a level of customer service is determined, the indicator having a plurality of dimensions of influence factors, each dimension of influence factors corresponding to a set of factors.
And for each index, determining the contribution degree of each factor in a plurality of factor sets corresponding to the index to the fluctuation of the customer service level according to the traffic data.
And selecting factors with larger influence from the factors in the factor sets according to the contribution degree.
And summarizing and calculating the factors with larger influence corresponding to each index to determine the key factors of the fluctuation of the customer service level in the current period.
The key factor positioning device for customer service level fluctuation provided by one embodiment of the specification can improve the accuracy of positioning the key factor for customer service level fluctuation.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the apparatus embodiment, since it is substantially similar to the method embodiment, the description is relatively simple, and for the relevant points, reference may be made to the partial description of the method embodiment.
The steps of a method or algorithm described in connection with the disclosure herein may be embodied in hardware or may be embodied in software instructions executed by a processor. The software instructions may consist of corresponding software modules that may be stored in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor such the processor can read information from, and write information to, the storage medium. Of course, the storage medium may also be integral to the processor. The processor and the storage medium may reside in an ASIC. Additionally, the ASIC may reside in a server. Of course, the processor and the storage medium may reside as discrete components in a server.
Those skilled in the art will recognize that, in one or more of the examples described above, the functions described in this invention may be implemented in hardware, software, firmware, or any combination thereof. When implemented in software, the functions may be stored on or transmitted over as one or more instructions or code on a computer-readable medium. Computer-readable media includes both computer storage media and communication media including any medium that facilitates transfer of a computer program from one place to another. A storage media may be any available media that can be accessed by a general purpose or special purpose computer.
The foregoing description has been directed to specific embodiments of this disclosure. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.
The above-mentioned embodiments, objects, technical solutions and advantages of the present specification are further described in detail, it should be understood that the above-mentioned embodiments are only specific embodiments of the present specification, and are not intended to limit the scope of the present specification, and any modifications, equivalent substitutions, improvements and the like made on the basis of the technical solutions of the present specification should be included in the scope of the present specification.