How to Optimize Revenues Using Dynamic Pricing?

Introduction

Uber/Ola peak hour costs are increased than common fares. In IRCTC, Rajdhani costs enhance are reserving charge will increase, and in Amazon, costs for the precise product change a number of instances. Who decides when to alter these costs or to what extent? Who decides the best value on the proper time? The solutions to those questions fall underneath the realm of Dynamic Pricing. This text offers freshmen with assets and theoretical understanding to construct a primary Dynamic pricing algorithm.

Studying Goals

Perceive the fundamentals of pricing and completely different strategies of pricing
Delve into dynamic pricing, advantages-disadvantages, strategies, use circumstances, and many others.
Fundamentals of income administration.
Implement a Easy Dynamic Pricing Algorithm utilizing Python to maximise income.

This text was printed as part of the Knowledge Science Blogathon.

What’s ‘Worth’?

In August 2023, the value of onions was Rs120 per kg. What led to it? Crunch in provide because of exterior environmental elements and a gradual demand. The market, the customer, the vendor, demand, and provide decided the value. The identical goes for many merchandise we purchase and promote right now: film tickets, bus tickets, e-commerce, gas, and many others.

Within the principle of value, demand and provide dictate the costs at which items and providers will commerce. When customers’ funds for items and providers align with the marginal value of manufacturing, we obtain the optimum market value, additionally known as the purpose of equilibrium between demand and provide. Setting the best value on the proper time is quintessential for enterprise development. Pricing managers thus deal with getting near the “Proper Worth,” which might be achieved by way of information and analytics.

Elements Influencing Pricing

Organizational elements: Product inventory out there, funds constraints.
Advertising combine: Stage of its product life cycle, Product, Worth, Place, and Promotion.
Product value: Value of manufacturing and uncooked supplies.
Demand for the product: Demand for the services or products.
Competitors out there: Competitor pricing to a big extent, determines inner pricing.

What’s Dynamic Pricing?

Dynamic pricing makes use of current developments, real-time buyer conduct, provide and demand, and competitors pricing to evaluate the value of products offered. It permits items to be offered at completely different value factors, such that buyer satisfaction is met and companies can thrive.

Dynamic pricing is adopted when demand is elastic. Dynamic pricing can’t be adopted when demand is inelastic or completely inelastic. When prospects are extremely delicate to modifications in value, there’s a excessive Worth Elasticity of Demand, and this property is exploited by way of DP.

For instance – In Bangalore, when just one autorickshaw is offered at a selected time and explicit location on a wet day, a buyer keen to pay increased (twice or thrice the going charge – elastic value) will get it, whereas one other buyer who isn’t able to budge must take the BMTC bus the place costs stay fixed(inelastic).

What are the Targets of Dynamic Pricing?

Elevated earnings, income, flexibility, market share, and buyer satisfaction.
Cut back outdated stock, main to higher utilization of warehouse house and assets.
Balancing provide and demand.

Therefore, the success of dynamic pricing is the maximization of Income/Earnings/Capability/Market share/Buyer satisfaction. Instance – If in 2021, with out dynamic pricing, 1M models had been offered, and the natural trajectory for 2022 is 1.5M models. Deploying dynamic pricing, models offered ought to enhance to 2M with out dropping out on buyer NPS or different pricing indexes.

Merely put, the YOY enhance in Income and Items is the final word success metric of any dynamic pricing algorithm. 

For an AB experiment on dynamic pricing, the success/output metrics that may be thought-about are:

Common order worth (AOV)
Conversion charge (CR)
Income per customer (RPV)
Gross margin share (GMP)

Elements Influencing Dynamic Pricing

Provide: When provide is decrease, costs are increased.
Demand: When demand is increased, costs are increased.
Stock ranges: Costs are dropped if a list is outdated and out of style. Instance – clearance sale.
Buyer choice: Ola mini, prime, prime plus, and many others., have completely different pricing tiers.
Seasonality and festivals: Airtickets throughout festive skyrocket, and companies revenue from excessive buyer demand.
Location: Touristy places have increased costs.
Time of day: Midnight costs are increased than noon costs
Competitor pricing

Forms of Dynamic Pricing

Segmented Pricing: Scholar low cost on Amazon Prime, senior citizen low cost on trains.
Time-based Pricing: Accommodations and flights in India are increased in October/November(festive season) than in August/September.
Peak Pricing: Surge value on Uber/Ola
Pricing is predicated on rivals.
Worth Elasticity: The extra elastic the product, the higher fitted to dynamic pricing. All FMCG merchandise are priced this manner in Dmart/Reliance shops, and many others.

Income/Yield Administration

One can not speak about pricing and never focus on income administration. Optimizing pricing, stock, and distribution to foretell demand to maximise income.

The first intention of income administration is promoting the best product to the best buyer on the proper time for the best value and with an appropriate pack.
Segmentation, forecasting, optimization, and pricing are instruments utilized in income administration.
It really works greatest when merchandise/providers are value elastic.

Is Dynamic Pricing Authorized in India?

The authorized and moral features of AI and ML are much less mentioned in India, so let’s focus on them.

Dynamic pricing deceives a buyer into selecting a pricing that may not be in his/her greatest curiosity. Additionally, this might be discriminatory, so the query is – Is it authorized?
In India, part 3 of the Competitors Act 2002 prohibits value fixing.
The part prohibits any settlement between or “follow carried on, or determination taken by, any affiliation of enterprises or affiliation of individuals, together with cartels, engaged in an identical or related commerce of products or provision of providers,” which determines the market value.
If two events collude and set costs very related or practically related costs, then it’s unlawful. But when one social gathering does so with out the data of the opposite, then both social gathering just isn’t liable.
The best route is to have a significant Private Knowledge Safety Act(just like these within the EU) that safeguards Indian residents in opposition to predatory company practices.

Drawback Assertion

FlyAirportByAir is a taxi-chopper service in Bangalore that gives taxi service to Bangalore Airport. Because the demand is comparatively fluid and modifications based mostly on climate, weekends, and festivals, they need to introduce dynamic pricing to enhance the highest line. Write an optimum pricing perform that can maximize income given:

Prebooking begins 100 days earlier than
The whole seats per day is 100
Demand varies between 100 to 200 per day. Generate demand utilizing a easy Python code –>np.random.randint(100, 200)
To simplify pricing -> Worth = Demand – Tickets offered

Given the Days left to ebook, whole seats out there, and demand for the day, discover the best value for every day.

## World Variables
DAYS = 100
SEATS = 100

DEMAND_MIN = 100
DEMAND_MAX = 200

Forecasting demand is step one in fixing dynamic pricing. Demand varies with inner and exterior elements. Time-series forecasting or regression strategies can be utilized to foretell future demand.

demand_hist = [np.random.randint(DEMAND_MIN, DEMAND_MAX) for i in range(10000)]
plt.hist(demand_hist, bins = 100)
print(“imply”, np.imply(demand_hist) )
print(“STD”, np.std(demand_hist)

Demand is predicted utilizing the Random perform; the imply worth is 150 every day seats, and the STD is 28.9.

Instance

Let’s contemplate this instance: D0 is the date of the journey. As individuals solidify their touring plans near the date of the journey, demand tends to be greater than the preliminary days(D8).  Although the market demand for D0 is 8, solely 3 seats had been booked; my rivals take in the remaining.

On condition that demand is linear, the Python illustration of the identical:

def linear_demand(days_left, ticket_left, demand_level):

tickets_sold_per_day = int(ticket_left/days_left)
value = demand_level – tickets_sold_per_day ## ticket_left/days_left practically is 1.
return max(0,value)#import csv

Perform to calculate income:

def cumu_rev(days_left,
ticket_left,
pricing_function,
rev_to_date = 0,
demand_min = DEMAND_MIN,
demand_max = DEMAND_MAX):

if days_left > 0 and ticket_left >0 :

demand = np.random.randint(demand_min, demand_max+1)
p = pricing_function(days_left, ticket_left,demand )
q = demand – p # demand is linear Q is tickets offered
q = max(0,q)
q = min(ticket_left,q) ## can not promote greater than tickets out there

return q*p, p

Given this straightforward perform, let’s calculate the value and income for – At some point earlier than the journey, and the full tickets left are 3. (As a result of demand is randomly chosen, income and value would possibly fluctuate, random.seed(10) might be outlined to get fixed solutions on a regular basis)

income,p = cumu_rev(1, 3,linear_demand )
print(“Whole Income – “, income)
print(“Worth Per Seat – “, p)

Given this straightforward perform, let’s calculate the value and income for – At some point earlier than the journey, and the full variety of tickets left is 10. The value per ticket ought to be increased as a result of demand is extra( 3 to 10).

income,p = cumu_rev(1, 10,linear_demand )
print(“Whole Income – “, income)
print(“Worth Per Seat – “, p)#import csv

With a simple-linear pricing perform, it’s evident that as demand will increase, value additionally will increase. Let’s simulate this and attempt to optimize the pricing perform.

Stimulations Utilizing Pricing Features

Let’s stress take a look at this straightforward perform for 10,000 seat reserving simulations utilizing pricing capabilities 1. linear_demand, 2. linear_adj, and three. linear_opti_variable  and select the most effective pricing that provides the best income, which is the purpose of this train

1. linear_demand

Demand is prediction random.
Worth is the distinction between demand and tickets offered.
Therefore, if demand is increased, the value will even be increased.

def linear_demand(days_left, ticket_left, demand_level):

tickets_sold_per_day = int(ticket_left/days_left)
value = demand_level – tickets_sold_per_day ## ticket_left/days_left practically is 1.
return max(0,value)#import csv

2. linear_adj

Demand is randomly predicted.
Worth is linear however stepwise. An index opti is launched to optimize the earlier linear_demand perform right into a piecewise perform. That when demand is increased, extra tickets are booked, which in flip will enhance income.
OPTI is a set worth based mostly on demand.

def linear_adj(days_left, ticket_left, demand_level):
“””
As an example we anticipate quite a lot of site visitors/views and impressions.

If demand is excessive we cost increased at completely different charges
“””

if demand_level > 180:
opti = 3
value = demand_level – int( (ticket_left/days_left) + (opti*(demand_level/180)))
elif demand_level > 150:
opti = 2
value = demand_level – int( (ticket_left/days_left) + (opti*(demand_level/180)))

elif demand_level > 100:
opti = 1
value = demand_level – int( (ticket_left/days_left) + (opti*(demand_level/180)))

elif demand_level > 0:
opti = 0
value = demand_level – int( (ticket_left/days_left) + (opti*(demand_level/180)))

return max(0,value)#import csv

3. linear_opti_variable

Much like 2, an OPTI index is used, however this index just isn’t fixed, and like Kmeans, the optimum worth of OPTI must be chosen based mostly on the elbow curve.

def linear_opti_variable(days_left, ticket_left, demand_level, opti = 1):

value = demand_level – int( (ticket_left/days_left) + (opti*(demand_level/150)))

# value = demand_level – int (ticket_left/days_left)

## if opti = 0 then the second time period turns into 0
## As opti elevated second time period elevated.
## 150 as a result of on common the demand is 150, (100+150)/2
## IF demand is increased than 150, then value will cut back
## IF demand is decrease than 150 then value will enhance.

return max(0,value)

Recursive income perform to calculate cumulative income for all 10,000 simulations:

def cumu_rev(days_left,
ticket_left,
pricing_function,
rev_to_date = 0,
demand_min = DEMAND_MIN,
demand_max = DEMAND_MAX):

if days_left > 0 and ticket_left >0 :
#random.seed(10)
demand = np.random.randint(demand_min, demand_max+1)
p = pricing_function(days_left, ticket_left,demand )
q = demand – p # demand is linear Q is tickets offered

q = max(0,q)
q = min(ticket_left,q) ## can not promote greater than tickets out there

return cumu_rev(days_left = days_left-1,
ticket_left =ticket_left-q,
pricing_function = pricing_function,
rev_to_date = rev_to_date+p*q)
else:
return rev_to_date

1. Output utilizing linear_demand:

simulation = [cumu_rev(DAYS, SEATS,linear_demand ) for i in range(10000)]
plt.hist(simulation, bins = 100)
print(“imply”, np.imply(simulation) )
print(“STD”, np.std(simulation) )
plt.title(“Income For 10K Ticket Reserving”)

The typical income based mostly on linear_demand perform is Rs14,908. That is evident from the histogram.

2. Output utilizing linear_adj:

simulation = [cumu_rev(DAYS, SEATS,linear_adj ) for i in range(10000)]
plt.hist(simulation, bins = 100)
print(“imply”, np.imply(simulation) )
print(“STD”, np.std(simulation) )
plt.title(“Income For 10K Ticket Reserving”)

The typical income based mostly on linear_adj perform is Rs16,146. That is evident from the histogram.

3. Output utilizing linear_opti_variable:

Step one right here is to decide on the OTPI worth which offers the best income:

opti_mean = []

for j in vary(20):
simulation = [cumu_rev(DAYS, SEATS,partial(linear_opti_variable, opti= j) ) for i in range(10000)]
opti_mean.append(np.imply(simulation))

plt.plot(opti_mean)
plt.title(“Optimum Worth For Income Maximization”)

def argmax(lst):
return lst.index(max(lst))
print(“The Greatest OPTI worth is -” ,listing(vary(20))[argmax(opti_mean)])

>> Output >> The Greatest OPTI worth is – 1

One of the best OPTI worth is 1 based mostly on the elbow curve. Now let’s discover income for OTPI = 1.

simulation = [cumu_rev(DAYS, SEATS,partial(linear_opti_variable, opti = list(range(20))[argmax(opti_mean)]) ) for i in vary(10000)]
plt.hist(simulation, bins = 100)
print(“imply”, np.imply(simulation) )
print(“STD”, np.std(simulation) )

The typical income based mostly on linear_adj perform is Rs15,838. That is evident from the histogram.

Analysis of Pricing Features

Based mostly on Maximizing Income, linear_adj is the most effective pricing perform. FlyAirportByAir can take a look at this perform and based mostly on the AB experiment, its strengths and weaknesses might be evaluated. Studying from this can be utilized to enhance efficiency over time.

Conclusion

Throughout industries like airways, railways, tourism, ticketing, and many others, DP has been deployed efficiently. When carried out rightly, dynamic pricing offers companies with flexibility and a possible development lever. With the best changes and elements, DP yields increased buyer satisfaction. This text offers a newbie’s information to the world of DP.

Key Takeaways:

Dynamic Pricing goals to optimize income, earnings, and buyer satisfaction.
The strategies utilized in dynamic pricing fluctuate from business to business.
One of the best strategies are chosen based mostly on AB outcomes, and iterations and enhancing the algorithm over time.
It may be utilized solely when demand elasticity exists.

Good luck! Right here’s my Linkedin profile if you wish to join with me or need to assist enhance the article. Be at liberty to ping me on Topmate/Mentro; you’ll be able to message me together with your question. I’ll be joyful to be linked. Try my different articles on information science and analytics right here.

Steadily Requested Questions

References

Kaggle Mini Programs: Airline Worth Optimization Microchallenge (https://youtu.be/irjpteecxdg?si=aUH2ifTekQutW-9n)
Coursera: Fundamentals of income administration. (https://coursera.org/be taught/fundamentals-of-revenue-management)
HBR Evaluation: 7 Classes on Dynamic Pricing (Courtesy of Bruce Springsteen) (https://hbr.org/2022/09/7-lessons-on-dynamic-pricing-courtesy-of-bruce-springsteen)
Dynamic Pricing Mannequin utilizing value multipliers for on-line bus ticketing platform. (https://www.krjournal.com/index.php/krj/article/view/38/357)
Dynamic Pricing Methods for Multiproduct Income Administration Issues (https://www0.gsb.columbia.edu/college/cmaglaras/papers/multi_rm.pdf)
Worth Optimisation: From Exploration to Productionising (https://www.youtube.com/watch?v=wPxDibqdg_w)

The media proven on this article just isn’t owned by Analytics Vidhya and is used on the Creator’s discretion.

Associated