The Bayes’ theorem describes the probability of an event, based on prior knowledge of conditions that might be related to the event.
Posterior probability:
P(A│B) = P(A⋂B)/P(B)
P(B│A) = P(A⋂B)/P(A); i.e., P(A⋂B) = P(B│A)×P(A)
P(A│B) = P(B│A)×P(A)/P(B)
Example 1: A new body wash brand Daisy has been launched. The brand is targeting the users of Dove, another body wash brand. From research studies, it is known that 5% of the consumers in this market are Dove buyers, and that a high proportion, 40%, of them shop at Watsons, a personal care retailer. It is also known that 25% of shoppers of personal care products shop at Watsons.
The brand manager is considering conducting a sampling programme for Daisy at Watson’s. However, she feels it will only be viable to do so if at least 10% of Watson’s shoppers are Dove buyers. Should she go ahead with the programme?
The answer is pretty clear from Exhibit 32.6, which sums up the information. However, using Bayes theorem:
$$P(D): Proportion \,of \,Dove \,buyers = 0.05$$ $$P(W): Proportion \,of \,Watsons \,shoppers = 0.25$$ $$P(W|D): Dove \,consumers \,who \,shop \,at \,Watsons = 0.40$$ $$P(D|W): Watsons \,shoppers \,who \,buy \,Dove$$ $$P(D│W)=\frac{P(W│D)×P(D)}{P(W)} =\frac{0.40×0.05}{0.25}=0.08=8 \text{%} $$Since the proportion of Dove buyers (8%) is less than 10%, the Daisy brand manager will not find it viable to conduct the proposed sampling exercise.
Example 2: A supermarket chain, Shop n Save (SnS), is facing stiff competition from market leader FairPrice (FP). The chain currently has a strong 32% base of regular shoppers, but is concerned that a high proportion of these regular shoppers are also shopping at FairPrice.
Based on a published research on FairPrice, the SnS team know that 85% of supermarket shoppers shop regularly at FP, and 24% of these shoppers also shop regularly at SnS.
Based on this information, can the SnS team compute what proportion of their regular shoppers also shop at FP?
$$P(S): \text{Proportion who shop regularly at SnS} = 0.32$$ $$P(F): \text{Proportion who shop regularly at FP} = 0.85$$ $$P(S|F): \text{Proportion of FP regular shoppers who shop regularly at SnS} = 0.24$$ $$P(F|S): \text{Proportion of SnS regular shoppers who shop regularly at FP}$$ $$P(F│S)=\frac{P(S│F)×P(F)}{P(S)}=\frac{0.24×0.85}{0.32}=64\text{%}$$As depicted in Exhibit 32.7, often we begin with initial or prior probabilities for certain events. Then we obtain additional information from, for instance, a sample of the universe. Given this information we compute the revised or posterior probability.
Consider n mutually exclusive and collectively exhaustive events, A1, A2 … An. Given that event B has occurred, we can use Bayes theorem to find the posterior probability:
$$P(A_i│B)=\frac{P(A_i⋂B)}{P(B)}=\frac{P(B│A_i)×P(A_i)}{P(B)}=\frac{P(B│A_i )×P(A_i)}{P(B│A_1)P(A_1)+P(B│A_2)P(A_2)… P(B|A_n)P(A_n)}$$ $$P(A_i│B)=\frac{P(B│A_i)×P(A_i)}{P(B│A_1)P(A_1)+P(B│A_2)P(A_2)… P(B|A_n)P(A_n)}$$Example 3: If the lung cancer test results for a patient are positive (T+), what is the probability, P(C+|T+), the patient has lung cancer (C+)?
Given:
NOTE: Laboratory tests reveal sensitivity and specificity. However, what the doctor needs to know is the likelihood that his patient is suffering from cancer, given that the test results are positive, i.e. the posterior probability, P(C+|T+).
Question: P(C+|T+): If the cancer test results for a patient are positive (T+), what is the probability the patient has cancer (C+)?
$$P(C+│T+)=\frac{P(T+ ∩ \,C+)}{P(T+)} = 0.0095/0.084 = 0.11$$There is 11% probability that patient has lung cancer, given the test turns out to be positive.
By applying the Bayes’ Theorem, we are able to transform the probabilities from lab test or research study, into probabilities that are useful.
In this example if you underwent the cancer test, and the result was positive, you would be terrified to know that 95 percent of patients suffering from cancer get the same positive result. On the other hand, the realization that the probability you have cancer is 11% is less terrifying. Though still not good news, it is much more useful and relevant information.
By rules of conditional probability:
$$P(C+│\,T+)=\frac{P(T+ ∩\, C+)}{P(T+)} = \frac{P(T+│\,C+)×P(C+)}{P(T+)}$$ $$Posterior = Likelihood × \frac{Prior}{P(T+)}$$ $$\mathbf{Posterior ∝ Likelihood × Prior}$$Likelihood is the data – in this case the likelihood the test is positive given that the patient has cancer.
Prior represents additional knowledge – 1% of the target population has cancer. Prior could come from anywhere. It could be a hunch, data taken from another study, anything we feel could have a bearing on the calculation.
Posterior is the information we are interested in. The probability that patient has lung cancer, given the test turns out to be positive.
$$P(C+│\,T+)=\frac{P(T+ ∩\, C+)}{P(T+)} = \frac{P(T+│\,C+)×P(C+)}{(P(T+)}$$ $$P(C−│\,T+)=\frac{P(T+ ∩\, C−)}{P(T+)} = \frac{P(T+│\,C−)×P(C−)}{(P(T+)}$$Taking ratios:
$$\frac{P(C+│\,T+)}{P(C−│\,T+)}=\frac{P(T+│\,C+)}{P(T+│\,C−)}×\frac{P(C+)}{P(C−)}$$ $$\mathbf{Posterior \,odds = likelihood \,ratio × prior \,odds}$$Example 4: A leading pharmaceutical firm segmented its customers, based on their priorities and preference. The segmentation process may be summed up into these 3 steps:
The company identified 3 segments:
The firm is particularly interested in quality seeking customers, and wants to tailor services for this segment.
To identify these customers, a 5-minute, screening questionnaire was developed.
Based on statistics from the quantitative study/cluster analysis, the probability that the screening test would correctly identify a quality seeker, P(T+/ Q+), is 90%. But 8% of customers who were not quality seekers would also test positive, P(T+/Q-) = 8%. The cluster analysis also revealed that 20% of the firm’s customer are likely to be quality seekers, P(Q+) = 20%.
Based on this information, we can tell what proportion of customers who are screened as quality seekers, are actually quality seekers:
$$P(T+)= P(T+ ∩\, Q+)+P(T+ ∩\, Q-) $$ $$P(T+ ∩\, Q+) = P(T+│\,Q+) × P(Q+) = 0.90 × 0.20 = 0.18$$ $$P(T+ ∩\, Q-) = P(T+│\,Q-) × P(Q-) = 0.08 × 0.80 = 0.064$$ $$P(T+)=0.18+ 0.064=0.244$$ $$P(Q+│\,T+)=\frac{P(T+ ∩\, Q+)}{P(T+)} = \frac{P(T+│\,Q+)×P(Q+)}{P(T+)} = \frac{0.90 ×0.20}{0.244}=73.8 \% $$73.8% of customers that are screened positive, will actually be a quality seeker.
Note: To find content on MarketingMind type the acronym ‘MM’ followed by your query into the search bar. For example, if you enter ‘mm consumer analytics’ into Chrome’s search bar, relevant pages from MarketingMind will appear in Google’s result pages.
Marketing has changed. More so in practical terms, and marketing education is lagging.
The fundamental change lies in the application of analytics and research. Every aspect of the marketing mix can be sensed, tracked and measured.
That does not mean that marketers need to become expert statisticians. We don't need to learn to develop marketing mix models or create perceptual maps. But we should be able to understand and interpret them.
MarketingMind helps. But the real challenge lies in developing expertise in the interpretation and the application of market intelligence.
The Destiny market simulator was developed in response to this challenge. Traversing business years within days, it imparts a concentrated dose of analytics-based strategic marketing experiences.
Like fighter pilots, marketers too can be trained with combat simulators that authentically reflect market realities.
But be careful. There are plenty of toys that masquerade as simulators.
Destiny is unique. It is an authentic FMCG (CPG) market simulator that accurately imitates the way consumers shop, and replicates the reports and information that marketers use at leading consumer marketing firms.
While in a classroom setting you are pitted against others, as an independent learner, you get to play against the computer. Either way you learn to implement effective marketing strategies, develop an understanding of what drives store choice and brand choice, and become proficient in the use of market knowledge and financial data for day-to-day business decisions.