Dynamic Pricing2018-10-14T13:58:49+00:00

Dynamic Pricing

Dynamic Pricing

Dynamic pricing has been on everyone’s lips again for several months now, especially with numerous board members and CEOs. On the one hand there is no way for companies to avoid dynamic pricing, on the other hand companies can quickly slip into illegality when applying dynamic pricing. This illegality should not be taken lightly, because in one case or another it can mean that CEOs are liable to take action. Especially in the areas of EU data protection and antitrust law. Nevertheless, no progressive company can do without it.

Definition of dynamic pricing

Dynamic pricing (= dynamic pricing) is the adjustment of prices to internal and external parameters. Dynamic pricing can refer both to products and to services.

Influential variables of dynamic pricing

Influencing factors dynamic pricing

The main influencing variables underlying dynamic pricing are:

  • Environment
  • Politics
  • Competition
  • Supply & Demand
  • (End-)Customers
  • Product utilization
  • Internal costs

Environment and politics in dynamic pricing

The areas of environment and politics are not considered in this article. Each of the two points in itself is highly complex.

Environment includes environmental catastrophes such as storms, floods, volcanic eruptions, famine, etc. But also economic changes such as unemployment, economic crises, trends (in the sense of profound social changes) are part of it. The permanently accelerated digitalization and its effect on whole cultures, societies, politics and economy complicate the weighting of the environment in the interplay of dynamic pricing.

In recent years, politics has been easier to calculate than the environment. But the global political context has become much more dynamic. The not insignificant lobbying of corporations and associations at state, federal and EU level has also become considerably less transparent. The weighting of politics within the framework of dynamic pricing must therefore now be carried out at the state level.

Note:
The weighting of the two points environment and politics within the dynamic pricing was already described by me in an essay at the beginning of 2017. Also the mathematical calculation by divergent statistical methods in interaction with unsteady differential equations of different order were part of the essay. Also included were the transfer examples of the mathematical expressions into the programming worlds for the development of self-learning algorithms as well as the construction of dynamic database models and their automated adaptation to the constantly changing results of machine learning. Further information on request.

Dynamic pricing – the competition

The information about the prices and the price changes of the competition, which can partly be determined e.g. with simple Screen Scraping or Data Scraping procedures, is important. However, some additional information is required.

Dynamic pricing observation of competition

With regard to competition, the following points apply to dynamic pricing:

  • Comparability of products (features, scope of functions, guarantee, warranty, right of return, etc.)
  • brand strength / brand bonus of the company
  • Customer structure
  • Product utilization of the competitor’s product
  • running costs of competing product
  • product development costs
  • Marketing costs Competition
  • Personnel

competitive product comparability within dynamic pricing

The comparability of competitive products is not always easy. A company whose sales and marketing is only nationally oriented and has an easily comparable product, e.g. a toothbrush, has little to fear in terms of complexity. Other companies, on the other hand, which operate internationally in marketing and sales and have to market numerous and complex products, have a harder time.

For example, tourism companies that place their travel products on all relevant sales and marketing channels as well as on all markets. They have to calculate their prices for each country, e.g. according to income groups or culture-specific, and enter into price wars with national and international competitors. There are thousands of travel products in tourism. A travel product often consists of different modules, e.g. flight, hotel or rental car. Each module, in turn, contains numerous price-influencing options. Departure/arrival times, departure point/destination, booking classes, holiday duration, room type (e.g. sea view, double room/single room, etc.), catering type or proximity to the beach are just a few examples of the diversity of each individual module. Therefore, competitive product comparisons in this industry are not trivial, even for large companies.

However, dynamic pricing requires the comparability of competitive products with one’s own products. In order to start dynamic pricing with reduced complexity, a Pareto analysis can be used to reduce the product range.

Other B2C industries, not exclusively the travel industry, also have to struggle with similar complexities in product comparability within dynamic pricing. Especially the globally producing B2C industries, which produce complex products consisting of thousands of parts internationally and hundreds of suppliers are involved in the product development process. The success potential of dynamic pricing in the travel industry is, despite the complexities, likely to be exploited more quickly than in other industries.

Note:
Internationally operating large companies usually work with dynamic multilingual content management systems (CMS) at the customer interface. Images, videos and texts are dynamically generated by bots and optimized by marketing experts. The optimized texts are then translated by bots into different languages so that the entire new content can be rolled out globally automatically (think global, act local). Culture-specific aspects, e.g. the avoidance of white color in Japan, are already taken into account in the programming. These complex content management systems often complicate the docking of a global system for dynamic pricing. Several test staging processes should be planned in advance.

Dynamic Pricing – Identify brand strength, brand significance and customer structure of the competition

The consideration of the brand strength of the competition is, besides the comparability of the competitor products, an essential influencing factor of dynamic pricing. The comparison of competitive brands with each other as well as with one’s own brand and the associated customer structures are extremely relevant. The marketing capabilities of the various competitors are decisive for the brand strengths.

In the online sector, the marketing capabilities of a competitor can easily be determined in general as well as product-specifically. The ability of companies to master the rules of online marketing can easily be determined qualitatively and quantitatively by analyzing SEO, SEM and social media activities.

Marketing activities and marketing capabilities on broadcast channels such as linear television, IP-TV and radio can be determined just as easily in an automated way.

The measurement and analysis of offline marketing involves a little more effort.

In the end, behind each competitor there is a general weighting as well as a marketing channel-based and product-oriented weighting of the marketing capability of a competitor. The competitive strength of a competitor can be determined on the basis of the market strength and the marketing capability weighting – in general as well as in relation to products and marketing channels.

The analysis of the marketing budget behind a product allows reliable conclusions to be drawn about a competitor’s priorities, marketing strategy and target groups.

Brand strength is responsible for the number of customers and their trust in the brand. The lower the strength of the brand, the higher the product marketing costs and the trust work at the point of sale (online & offline). The significance of the brand, in turn, depends on the type of target group. Often the image and the (apparently) underlying product quality play an important role.

For example, in the food trade Aldi has a great brand strength and the meaning for the customers is “with Aldi it is cheap”. At Tegut the brand strength is lower and the meaning for the customers is “at Tegut I get ecologically valuable food”. Only a few Tegut customers compare food prices with Aldi’s prices – not even for comparable products. The other way around, however, already.

Therefore, dynamic pricing requires a brand weighting of each competitor, because it provides information about the channel-based marketing capabilities, the strength of the brand and the customer structure of the competitor associated with the brand.

Dynamic Pricing – Competitor’s Product Costs

In general, companies know how their competitors work. With which systems and processes, with which suppliers and frequently also the personnel of the competition are known. One meets on congresses and fairs and frequently personnel changes remain in the same industry, so that also thereby information exchange takes place. However, this information is not used.

The fewest companies allow the product costs of their competitors to flow into dynamic pricing – even though a lot of information is available. Experience has shown that a senior sales employee can estimate the weighting of how quickly and cost-effectively the competition launches new products and how much revenue remains at the end of the process relatively precisely from the “gut,” so that extensive analyses can be omitted in the first step. We often calculated product costs and revenues per competitor’s product at great expense within the framework of dynamic pricing projects. In the end, we often found that the “spontaneous estimates” of sales were close to our results and weighting factors.

Product costs include the product supply costs as well as the running costs for product maintenance and the marketing costs. The product supply costs consist in particular of the three points market analysis, resource purchasing and (technical) product development. The costs of the product supply, regardless of whether it concerns physical or intangible goods, are the maintenance, care and continuous improvement of the products. The lion’s share of costs in most industries are the costs of marketing – multichannel offline and online.

The decisive factor is efficiency (cost/performance). In the age of digitization, it is also called the “degree of automation” or “maturity of digitization”. At the end of the day, the aim is to achieve corporate goals at the lowest possible cost. High costs are usually caused by the employees. For this reason, strategy consultants speak of “heads per product” or “heads per target” when determining the digital maturity level. Investors dream of highly efficient companies without employees, without real estate and best of all generally without hardware. Companies that consist only of software and in which even the servers are omitted, e.g. due to the use of blockchain procedures; the calculations currently still made on the server side are thus decentralized and distributed to the customers (similar to home banking).

Dynamic Pricing – Product Utilization and Marketing Costs in Competition

In general, companies know which competitive products are performing well or badly. As mentioned above, the digital marketing activities of competitors can also be analyzed relatively easily and automatically. Products in strong demand are generally marketed less aggressively. Especially if the product provisioning capacity, which is generally known, is limited or finite.

This can be used to determine which competitor products are/were worth how much marketing budget to the competitor at which points in time. If the price falls steadily with slight fluctuations in marketing activity, the purchase price or the product development costs of the competitor can be determined at the end of the respective product marketing with low tolerance. It is also possible to determine how much marketing budget is invested for the marketing of “shopkeepers” and what time span the competitor plans for the marketing of his products.

As a result of this information, intelligent strategies for the internal automated dynamic budgeting of one’s own product-related marketing costs can be mapped electronically. This requires not only real-time prices, but also past price differences of competitors within different time periods.

Competition personnel costs as part of dynamic pricing

In many industries, personnel costs are a significant factor in relation to the quality of staff training. In general, we screen the job advertisements on the relevant job portals and on the company websites over many years. We also look at the Xing and LinkedIn profiles of candidates who are highly likely to have received the job advertised. We compare the job advertisements and the candidate profiles with a self-developed skills matrix.

This is done on a sectoral basis, e.g. marketing, product development, IT, innovation management, finance, etc. Based on regional and national sector salary indices, we thus obtain a reliable skills personnel cost index for the specialist areas of companies. In this way, the personnel costs and the associated company skills are weighted within the framework of dynamic pricing.

Conclusion Competition analysis for dynamic pricing

European companies unfortunately have too weak a competition monitoring system integrated into their business processes because it is not consistent. In addition, largely established competitors are used for competition monitoring. Far too little focus is placed on “non-industry” companies.
For example, HostEurope, Strato, 1&1, One, etc. have constantly observed each other and have entered the price war against each other without innovation. In the meantime, a B2C online mail order company named Amazon has suddenly developed and successfully marketed innovative B2B cloud offers.

Integration of customers and consumers into dynamic pricing

Dynamic pricing at the customer interface is significantly more than displaying more expensive prices to Apple users.

The key question is:

“For which consumer target groups can which products be offered at which prices and in which period on which channel?”</block rate>”</block rate>”</block rate>”</block rate>”</block rate>”</block rate>>”</block rate>>”</block rate>>”</block rate>>”</block rate>>”</block rate>>”</block rate>>

Customer-based Dynamic Pricing

The dynamic pricing system must work with target groups based on historical data as well as real-time data.

Note:
The prerequisite for dynamic pricing is dynamic target group matching (automated user target group matching). Dynamic target group matching is in turn preceded by target group identification and definition. Target groups are a complex topic. Especially when target groups are partially addressed in real time. The real-time demand recognition and offer playout is relevant, for example, for target groups who make effect or impulse purchases. Further information on request.

For the target groups determined on the basis of heuristic data, CRM, merchandise management or booking systems are generally used. These are supplemented by target groups determined from real-time data. For example, by profiles of website visitors (website visitor profiling), social media users or mobile users. In some cases, e-mail addresses are also sufficient for profiling (E-Mail address provides information about users of the e-mail address).

However, offline outlets should also benefit from target group analysis results from real-time and heuristic data. Why should dynamic pricing be used only and exclusively for online channels? Merging offline and online channels is difficult. However, this is still a requirement in many industries. For example in consulting-intensive or personality-sensitive industries.

At the end, each target group is dynamically assigned product price difference factors and recognition probability weightings. Target groups relevant for dynamic pricing, nourished by real-time and historical data, are:

  • Recently played offers
  • Last purchases/bookings
  • Character properties
  • Interests
  • Socioeconomics
  • Generations
  • Actively used marketing channels & the periods of activity

If no patterns are available for past data, e.g. “flies to Boston every May” or “buys 1000 grams of the same type of tea every 4 to 5 weeks”, the most current customer data, e.g. current search profile from the website, is weighted higher for dynamic pricing than customer information from the past. However, the weighting is based not only on the data topicality, but also on the data quality or the reliability of the data source as well as the significance of the data source for the customer; e.g. an e-mail newsletter registration of a prospective customer is far less relevant for sales than a contact form sent by the prospective customer, for which the prospective customer had to invest more time than for the specification of his e-mail address.

In dynamic pricing, the target groups are permanently in the real-time interplay of revealing weightings. The complexity of “user-target group allocation in dynamic pricing” is sometimes also due to the fact that a person is in several target groups at the same time and his “target group membership weightings” are subject to permanent changes.

Value of demand forecasts for dynamic pricing

“Betting on future needs developments” is a risky game. The current volatility of world events is increasing the risk factor. Sociological changes and far-reaching social trends also play an important role in prognoses because they determine prices.

However, in some industries it is still possible to successfully develop forecasts for dynamic pricing even with low resource input.

Energy brokers, for example, can abstract data from CO2 emissions trading with data on raw material purchasing and extraction. In the tourism industry, the demand for package tours per country can be partly predicted on the basis of pure flight bookings. Initial studies have shown that for certain O&D’s and airlines in defined markets, the number of flights is 6 to 8 weeks ahead of the package tours.

Often numerous other parameters are used for forecasts. Exchange rate developments, gross domestic products, import/export ratios, etc.

The weighting of the forecasts within dynamic pricing is constantly adjusted to the permanent improvement of the probabilities of occurrence of the forecasts. In most industries, product demand forecasts for most products are initially rated very low. Over time, the forecasts generally become more reliable and thus increasingly weighted higher. There are companies in which demand forecasts now account for 30% of the total weighting within the dynamic pricing system and have thus become a strong influencing factor of dynamic pricing.

supply and demand

Information on supply and demand should be available to Dynamic Pricing as a company-wide supply-demand KPI and as a product-specific KPI for each product. In addition, the supply-demand values of the entire markets in which the products are sold as well as the estimated values of the competitors should also be available. Thus, the dynamic pricing system can permanently calculate real-time potential analyses. This is done on the basis of competition and market analyses as well as the company’s own internal supply and demand situation and the forecast demand.

Internal Finances within Dynamic Pricing

These same types of information that you should be aware of from the competition are also relevant for determining your internal costs per product. However, it goes without saying that your data should be more precise and more comprehensive when it comes to taking your cost structure into account in dynamic pricing.

Prerequisites for dynamic pricing are a distribution of all costs (fixed costs, variable costs and mixed costs) and revenues to all products (in monetary units). Each product should receive a realistic cost and revenue value, related to the entire product process, – defined according to Kaizen; from market research and product development to product provision, maintenance and marketing to the removal of the product from all systems.

Thereafter, each product shall be assigned an expense value (in monetary and hourly units) and a capitalized earnings value (in monetary units).

The following questions can thus be answered:

  • What is the revenue in relation to the cost per product?
  • How high is the expenditure (in money and hours) in relation to the yield per respective product?
  • How high is my profit or loss per product?

Only when this information is available as data in real time can the dynamic pricing system work completely. Now 2 essential parameters are set in the financial area. On the one hand, the minimum sales prices per product and the turnover and profit expectations for the company in general.

The dynamic pricing system now calculates the following from the relevant and already described influencing variables

  • Environment/Politics
  • Competition
  • Target groups
  • demand forecasts
  • Own supply-demand situation

one real-time price per individual demand.

The system tries to achieve the highest possible price. If, according to the dynamic pricing system, the target minimum price per product cannot be achieved, a decision is made as to whether the product should nevertheless be sold and, if so, how many sales (per period) should be permitted. This is generally done automatically, but can also be done manually during ongoing operations. At the same time, a report on the dynamic price development per product is generated and made available to the respective departments. In particular the top management as well as the finance, product management, sales and marketing departments.

Structure Dynamic Pricing

In the background, the system always calculates the overall financial situation. The dynamic pricing system attempts to meet the company’s revenue and sales expectations. For example, it calculates at which points or other product sales below the minimum sales price the negative gap is currently compensated.

Machine learning within dynamic pricing

Each single influencing factor of dynamic pricing is complex. Competitive and market observation across different countries as well as the permanent dynamic and automated real-time allocation of people to target groups or the comparability of one’s own products with those of the competition. But also the consideration of the constantly changing supply and demand situation in the present, as well as the consideration of demand forecasts, involves complexities. Every influencing factor of dynamic pricing should be improved over time.

The numerous parameters of each influence quantity and the constantly changing weightings make it difficult to improve the influence quantities by human hand. Even more difficult than the handling of an influencing variable is the complexity of the influencing variables in real-time interaction with each other. This requires intelligent dynamic pricing control software on the one hand and software that independently checks and improves the results of the control software on the other.

Artificial intelligence in dynamic pricing and suitable industries

The path from machine learning to artificial intelligence is not far away. The development of an AI for dynamic pricing (as a saleable product) will probably be developed by SAP in Europe if no “IT” company from the USA or China makes it ahead of time. However, even SAP will probably not have enough data to develop the dynamic pricing KI. Ultimately, SAP is only a service provider for companies and should not have access to its customers’ data.

But who could develop artificial intelligence for dynamic pricing? Which companies or industries would have the greatest added value?

To exploit the complexity of the target group mapping, a B2C orientation would make sense. Dynamic pricing should also include product development, so pure B2C (eCommerce) sales platforms are out of the question. Which industries represent the complete process? Product development, product purchase, product provision, product marketing (offline and online) and billing?

These would be, for example, the automotive, pharmaceutical/chemical, travel, energy/oil, banking, insurance or real estate industries. However, at the consumer or customer interface not only a high offer output, but also high product sales are relevant. The more data, the better. The automotive industry and the real estate construction industry are thus eliminated – because of the high price, too few cars or houses are sold in order to quickly achieve a high learning potential for AI. The pharmaceutical industry is also eliminated due to the distribution channels via pharmacies. There remains energy/oil, e.g. mineral oil companies and their selling ways over gas stations, the insurance and banking industry as well as the travel industry.

The oil companies with their petrol station networks are not optimal because of the state regulation on how often the petrol/oil sales price can be changed per day. Insurance companies, banks and the travel industry would remain. However, the product offering volumes and the number of products sold by insurance companies and banks are much lower than in the travel industry. In addition, the banking and insurance industry works very closely with intermediaries (e.g. Check24) and has no access to the data within the customer journey that is valuable for dynamic pricing. At first glance, the travel industry seems to be very suitable. In this industry, the development of dynamic pricing can be started from many sides or specialist areas and gradually merged.

Note:
Of course other industries or groups can also develop a dynamic pricing AI. For example, there may also be sufficient data available for automated dynamic pricing on past and current worldwide automobile sales information. Especially if the data from car rental is added (Car2Go, Drive Now, etc.). Also, it is not necessary to develop a complete dynamic pricing KI with all influencing variables. This year, we developed interfaces between production planning and car buyers for two automotive groups. There we came into contact with digital intelligence, which plans and monitors the orchestration of the worldwide production of all parts, whether produced internally or by suppliers. The automotive companies have decided that it is in this area that the greatest benefit of an AI can be developed most quickly. Irrespective of the industry, – the revenues and earnings can be significantly increased, for example, through the realization of the two influencing factors competition and target groups. Nevertheless, the added value of dynamic pricing is higher in some industries than in others. The more influencing variables are considered, the simpler the interfaces between the systems, the lower the disruptive factors and the more data is available for each influencing variable, the more successful dynamic pricing becomes.

Data acquisition in dynamic pricing

It is no longer a secret that numerous data should be available for training an AI. The more data about the respective influencing variables are available, the more precise the AI becomes in dynamic pricing. Past projects have already provided us with experience regarding the required amount of data per influence quantity. Mathematical stochastic methods, which we have electronically mapped in algorithms, permanently monitor the capabilities of the AI and determine the remaining data quantities required during the AI training phase until the end of the training phase.

Necessity of research and development alliances for dynamic pricing

The conflict between data quantity/data quality/data topicality and dynamic price adjustments as well as data protection/antitrust law is wide. The renunciation of dynamic pricing is not a solution. Competition, which is more advanced in its maturity level, would quickly overtake one’s own company. Accepting the challenge of “dynamic pricing” is not a luxury decision – dynamic pricing can quickly become a question of existence.
p align=”LEFT”>The bitter cost-causing pill “data protection and antitrust law”, especially with regard to the new EU data protection directive, should be courageously swallowed.

And also the answers to the questions how much data will be required for which influencing variables are no longer rocket technology nowadays. The key questions are rather:

  • Where do the masses of good quality data for each influencing variable come from?
  • How are the processes designed and how are the processes mapped digitally in or with the historically grown systems?
  • What resources are required after the provision of the dynamic pricing system for ongoing operation and further development?
  • With which standard technologies currently traded on the market can dynamic pricing be set up? Which standards have a future and which system and software suppliers are (still) likely to be successful in the future?
  • How high will the ROI be and when will the Break Even Point (BEP) be reached?
  • Who pays for everything?

Dynamic pricing should be considered holistically and taken into account in the corporate strategy. However, quick successes in the project business are necessary for the project team as well as for the stakeholders and investors. Therefore, two to three influencing variables should be defined at the beginning and their pricing dynamized.

Contrary to the disdainful prophecies of doom from innovators in love with technology, we have found that a high degree of artificial intelligence is not yet required for the first development stage of the dynamic pricing system. The project milestone of machine learning can thus also be established at a later stage.

Sophisticated mathematics cast in algorithms paired with standard technologies and standard processes can already carry out complex processes within dynamic pricing very quickly and automatically. The implementation of expensive novel high technologies, for which there are hardly any personnel resources in Germany, can take place as soon as economic successes flush more finances into the coffers through dynamic pricing.

Nevertheless, a dynamic pricing project will not be a cost-effective undertaking. Not only medium-sized but also large companies are afraid of costs. Data collection is also a big challenge for most companies. In most cases, sufficient data is available for dynamic pricing for some influencing variables, but not for all.

This data acquisition problem is not only present in Germany or Europe, but also in the USA or China. Therefore, the companies there form data and research alliances.

The companies in Europe and Germany should also form R&D and data alliances, in the first step within and in the second step across industries. Within the framework of these alliances, dynamic pricing systems could be developed to conserve resources (complete result with shared costs).

In the travel industry, alliances between tour operators, airlines, Amadeus and hotel chains would be possible in order to cost-effectively develop a dynamic pricing system. In a second step, for example, the insurance groups (travel insurance products) currently under strong pressure could join the R & D alliance and benefit from the advantages of a dynamic pricing system through additional R & D payments.

In Germany, too, there are experts recognised throughout Europe in legal matters.

In the area of data protection and the new EU data protection directive, for example, it is the law firm Spirit Legal with offices in Leipzig and London.

Since dynamic pricing, in particular the important influencing factor “competition price screening” (actually price agreements), has an antitrust component, the Faculty of Law of the University of Würzburg with Prof. Florian Bien has a capacity recognised throughout Europe.

In most cases, such developments as dynamic pricing result in numerous valuable by-products that can be used throughout the industry.

The outsourcing of R&D services is currently becoming more popular. The reasons for this are the high costs for innovation and product development and the lack of natural scientists and (software) engineers.

Example:
HighPots has developed a recurrent artificial neural network (rKNN) for autonomous driving in recent years in cooperation with the Chinese automotive group SAIC Motors and Huawai. The aim was to achieve level 3 autonomy. The Chinese search engine giant Baidu now wants to develop a procedure for level 5 autonomy for SAIC Motors (Interview Baidu Wall Street Journal ). Level 5 autonomy can only be achieved if cars no longer need a permanent Internet connection for autonomous driving. Since the functioning of assistance systems such as Siri or Cortana on smartphones is also important offline, SAIC and Baidu use Huawai technologies for this purpose. A good example of meaningful R&D alliances.

By the way, there are also development and data alliances between the apparent competitors Microsoft, Google, Facebook and Amazon. Alliances are also common in the hardware production of smartphones, for example. Companies such as Samsung, for example, supply displays to other smartphone manufacturers.

For German companies such alliances are sometimes still unpopular. Perhaps the reason is that German companies still think and act too little globally and too locally. As a result, they often only see competition in their own country and only in their own industry.

Dynamic pricing is predestined to be developed in an alliance or to establish an R&D alliance/cooperation. The separation between “general validity” and individual “company internas” is easily possible. For example, the threshold values for product-based profit and loss accounts can remain in the companies’ internal financial systems, while the system for the individual automated matching of offers to demand (and vice versa) can be made available to all cooperation partners.