Automated pricing and dynamic pricing in e-commerce

In the digital economy, automated pricing and dynamic pricing strategies are now part of everyday life. Whether for online shopping, flight bookings or transport services – prices can now be adjusted flexibly and in real time to demand, supply or even individual customer data. Companies use complex algorithms and artificial intelligence (AI) to determine optimal prices that maximize sales and profits.

Not only e-commerce giants, but also more and more small retailers and start-ups are turning to price automation: special SaaS tools for dynamic pricing are available on the market, and marketplace providers are increasingly offering their sellers algorithmic price recommendation systems. However, this practice raises a number of legal questions. When does dynamic pricing violate German law, in particular the Unfair Competition Act (UWG) or the Price Indication Ordinance (PAngV)? Where are the boundaries between permissible price differentiation and impermissible consumer deception? What transparency and information obligations must online providers observe – for example on marketplaces, in SaaS models, on blockchain platforms or for subscription services?

This comprehensive legal analysis examines the legal framework of automated pricing in Germany. Particular attention is paid to Section 5 UWG (misleading pricing), the requirements of the Price Indication Ordinance, the prohibition of unfair price discrimination and the transparency obligations for algorithmic pricing models. Relevant EU regulations – such as the Unfair Commercial Practices Directive (UCP Directive), the Digital Services Act (DSA) and the Digital Markets Act (DMA) – are also included and dovetailed with German regulations. A comparative look at international perspectives (including the USA and UK) shows how other legal systems deal with dynamic pricing.

In addition to the legal framework, the article considers moral and reputational aspects. Many consumers perceive personalized prices or extreme price fluctuations as unfair or even fraudulent. It discusses the consequences of customers recognizing the price variation – such as loss of trust, public shitstorms in social media or negative PR – and how companies can prevent such reputational risks. Finally, innovative start-ups, SaaS providers, platform operators and tech companies will receive practical guidelines: What is allowed, what is forbidden? How can AI-based pricing models be implemented in a legally compliant manner? What disclosure and information obligations exist? How does the legal framework for platform operators differ from that of retailers? The aim is to provide well-founded orientation in the area of conflict between price innovation, transparency, consumer protection and competition law.

Dynamic pricing: concept and practice

Dynamic pricing refers to dynamic pricing, i.e. the continuous adjustment of prices to current circumstances. In contrast to a fixed price list, which is the same for all customers at all times, dynamic pricing allows prices to be changed at short intervals. Reasons for price changes can be, for example, demand (high demand causes prices to rise, low demand lowers them), the available supply or stock, the time of day, the customer’s location or even individual characteristics and the customer’s previous behavior.

In practice, there are various forms of dynamic pricing. A classic example is yield management in the travel and airline industry: airlines and hotels constantly adjust their prices to booking occupancy and remaining capacity. When occupancy rates are high, prices rise (last-minute bookings are often more expensive), while discounts or low rates are advertised during slow periods. Surge pricing on ride-sharing platforms such as Uber also follows this principle – if demand rises sharply (e.g. when it rains or after major events), the algorithm automatically increases fares to balance supply and demand.

In addition to price adjustments based on time and demand, there is also personalized pricing. This involves tailoring the price to a specific customer or customer group. For example, algorithms can derive a user’s willingness to pay from their previous purchasing behavior or profile. It is therefore conceivable that regular customers see different prices than new customers or that users of a certain device type (such as iPhone users) are shown higher prices than users of a cheaper device. Such individual price differentiation aims to skim off the maximum price that each customer is still willing to pay (often referred to as first-degree price discrimination ).

Dynamic pricing and personalized prices are ultimately forms of price discrimination in a neutral sense: similar goods or services are offered to different customers at different prices. Economics and law distinguish between different levels of price differentiation:

• First-degree price differentiation: each customer is charged a price that is individually tailored to their willingness to pay. In practice, this can rarely be fully realized, but serves as a theoretical ideal in which the provider attempts to fully skim off the consumer surplus.
• Second-degree price differentiation: different prices depending on the purchase quantity or product variant. The customer can choose which package to buy (self-selection). Examples: Volume discounts, graduated prices, lower prices when buying larger units (“3 for 2” promotions) or the distinction between economy and business class tickets.
• Third-degree price differentiation: division of the market into customer groups that are treated differently. Criteria can be, for example, age (senior discount), status (student discount) or region (country prices). Every customer in a group pays the same price, but the group prices vary.

Dynamic pricing can contain elements of several degrees. In online retail, the concepts overlap – for example, a web store can vary prices in general depending on the time of day and also display special prices to a specific customer group.

Companies hope that automated pricing will enable them to better adapt to the market and increase profits. Modern software solutions analyze large amounts of data(big data) in fractions of a second in order to optimize prices in real time. Innovative business models in e-commerce, software-as-a-service (SaaS) or on digital platforms in particular are increasingly relying on such AI-supported pricing models. However, the opportunities are accompanied by risks: consumers react sensitively to price differences that are perceived as unfair, and the legal framework sets limits to prevent abuse. The following section explains the legal requirements that apply to dynamic pricing and how companies can act in a legally compliant manner.

Legal framework for pricing under German law

In Germany, competitive pricing is subject to the principle of freedom of contract – in principle, every provider is free to set its own prices. However, various laws restrict this freedom, in particular to protect consumers from being misled or exploited and to ensure fair competition. The most important legal bases in German law that must be observed in the case of automated pricing and dynamic pricing are explained below.

Checklist of important regulations for dynamic pricing in the consumer sector:

• § Section 5 UWG: Prohibition of misleading information about prices or price advantages.
• § Section 5a UWG: Obligation not to conceal material information (including price-related information).
• Annex UWG No. 5: Prohibition of enticing offers (price must be available for a reasonable time/quantity).
• Price Indication Ordinance (PAngV): Indicate total price incl. VAT; basic price for quantity units; for price reductions, indicate lowest price of the last 30 days (§ 11 PAngV).
• Art. 246a EGBGB: Information, if price personalized by automated decision.
• General Equal Treatment Act (AGG): No discrimination due to e.g. gender in mass transactions (exceptions possible, § 20 AGG).
• GWB/antitrust law: No price agreements, caution in the case of a dominant market position (no exploitation or obstruction through prices, Section 19 GWB).
• GDPR: Safeguard data protection when using personal data for pricing; if applicable, no exclusively automated price without notice/right to review (Art. 22 GDPR).
• Button solution (§ 312j BGB): Before clicking on “order with obligation to pay”, clearly display all price components (no hidden costs).

Prohibition of misleading statements and true prices (Section 5 UWG)

The central regulation for the permissibility of pricing practices in e-commerce is the Unfair Competition Act (UWG). In particular, Section 5 UWG prohibits misleading business practices, i.e. deceptive practices in relation to the price of a product or service. According to Section 5 (1) sentence 2 no. 2 UWG, a commercial act is misleading if it contains untrue or deceptive information about the price or the way in which the price is calculated. The existence of a particular price advantage may also not be misleading. This standard expressly states: “A commercial act is misleading if it contains untrue statements or other misleading statements about the price or the existence of a particular price advantage.”

This means for dynamic pricing: Price adjustment in itself is not prohibited, but it must not lead to a false perception on the part of the customer. Classic examples of unfair behavior are sham discounts and price concealment. A sham discount occurs, for example, when a retailer increases the price at short notice in order to subsequently feign a “discount” – in this case, a supposed price advantage would be misleading. It would be similarly inadmissible if an algorithm regularly raises prices in order to then advertise with crossed-out prices and discount announcements without the higher “strike price” ever having applied for a reasonable period of time. In such cases, case law speaks of anti-competitive price gouging. As early as 1983, the BGH (judgment of 05.05.1983 – I ZR 46/81) ruled that a bait-and-switch offer was anti-competitive because the advertised price was in fact hardly available due to a lack of sufficient stock. Courts have also ruled that an originally higher price must have been seriously demanded for some time before a reduced promotional price may be advertised. Otherwise, the consumer is misled as to how significant the price reduction actually is.

Decoy offers are also inadmissible under Section 5 UWG. This refers to the advertising of a product at a conspicuously low price, even though this price is only available for a very short time or in limited quantities. Here, the customer is lured with a bargain that is hardly available in reality. Dynamic pricing could become a bait-and-switch offer if, for example, an algorithm initially sets the price extremely low, but drastically increases it after a few purchases or minutes, without the limitation being pointed out in the advertising. Such an approach would violate the principle of fair pricing. According to the UWG blacklist (Annex to Section 3 UWG, No. 5), an advertised price must be available for a reasonable period of time; if it is available for an extremely short period of time, there is the appearance of an unfair bait-and-switch offer. In a ruling, the Federal Court of Justice (BGH) decided that a minimum period of availability must be observed for promotional offers – e.g. an advertised special price in an electronics store must be available at least until the afternoon on the first day of the offer (BGH, judgment of 10.02.2011, I ZR 183/09). Stricter standards sometimes apply to everyday consumer goods.

Misleading information can arise not only through active deception, but also through the omission of important information (see Section 5a UWG). Therefore, if an essential price information detail is concealed that the consumer needs in order to make an informed decision, this may also constitute an unfair act. For example, a company that operates two different online stores with different price levels for different customer groups would have to clearly point this out to the consumer in order not to deceive by omission. Another example: Anyone advertising “guaranteed lowest price” must keep to this – an algorithm must not sometimes generate higher prices than the competition without qualifying the claim. Such unique selling point advertising would otherwise be misleading. Overall, it should be noted that every pricing model must be communicated transparently and clearly so that the customer is not misled about the price itself or the conditions of the price. Last but not least, enforcement and sanctions are an issue: infringements can result in warnings from competitors or consumer protection associations. In online retail in particular, there are specialized players who systematically check stores for PAngV and UWG violations. A single misleading price indication can therefore quickly result in a costly cease-and-desist order.

Information obligations and transparency

In recent years, the transparency obligations for online retailers have been significantly tightened. One major innovation – implemented by the EU Omnibus Directive 2019/2161 – concerns personalized prices. Companies are now legally obliged to disclose if they have personalized a price on the basis of automated decision-making. In Germany, this obligation has been included in the provisions on information obligations in electronic commerce (Art. 246a Section 1 (1) EGBGB, new version). Before the contract is concluded, it must be clearly and comprehensibly indicated that the displayed price has been personalized. A blanket reference somewhere in the general terms and conditions is not sufficient; the information must be provided specifically with the respective offer.

The background to this regulation is that consumers should be able to recognize whether they may receive a different price than another customer. Transparency should prevent customers from being secretly categorized according to their ability to pay or surfing behavior without knowing it. However, it is important to understand what exactly is meant by “personalized” pricing: Not every dynamic price change is personalized. If a retailer changes its prices in real time depending on general market demand or the time of day (i.e. the same for all customers), this is dynamic pricing, but not personalization within the meaning of the regulation. The disclosure obligation only applies if the price was calculated individually for the specific consumer based on their personal data or profile – for example, if an algorithm analyzes the purchasing behavior of the individual customer and adjusts the price accordingly. (Recital 71 GDPR explicitly mentions the automatic rejection of an online credit application or the automatic processing of data to set different prices as examples of such a decision).

If the required reference to personalized prices is omitted, this can be cautioned as anti-competitive behavior. § Section 3a UWG sanctions breaches of statutory information obligations (such as those from the Introductory Act to the Civil Code or the Price Indication Ordinance) as a so-called breach of law. In addition, the 2022 amendment to the UWG introduced a new sanctions regime: Consumer protection authorities can impose severe fines of up to 4% of the company’s annual turnover for far-reaching infringements. There is therefore a strong incentive for companies to take their information obligations seriously.

In the context of pricing, transparency also means not including any hidden costs or surprising surcharges. All price components (taxes, shipping costs, surcharges) must be clearly visible to the consumer before placing the order (Section 312j (2) BGB in conjunction with Art. 246a EGBGB). An algorithmic pricing model must not lead to additional fees only appearing in the very last step of the order, for example, which were previously concealed – such “drip pricing” would violate the clarity of the price information and could be punished as misleading by omission. Average price calculations or opaque subscription cost models that confuse customers should also be avoided. Instead, online providers should design their pricing models in such a way that the average consumer can understand what they are paying for and whether (and why) a price may fluctuate.

In summary, clear communication and openness are key: If AI-based algorithms are used, the company should ensure internally that their results can be made understandable to the customer. Although not every pricing algorithm needs to be disclosed in detail, the customer must not be left in the dark if they may be disadvantaged or subject to special conditions. In cases of doubt, more information is advisable in the interests of transparency – for example, information such as “price variable depending on capacity utilization” – in order to maintain consumer confidence.

Price Indication Ordinance: correct price indication and discount advertising

In addition to the UWG, the Price Indication Ordinance (PAngV) is a central set of rules for price labeling. It ensures that consumers find prices transparent and comparable. According to § 1 PAngV, total prices must always be indicated to consumers, i.e. the full final price including all taxes and unavoidable price components. A dynamically calculated price is not exempt from this – it must also be clearly visible as the final price. If a store changes prices depending on demand, the currently valid final price must be displayed at all times. Consumers must not be misled by incomplete price information, for example if an algorithm initially displays a net price and only adds VAT late in the purchase process.

The PAngV also contains regulations for special price indications: For example, Section 4 PAngV requires the indication of basic prices (price per unit of quantity, e.g. per kilogram or liter) for goods that are offered by weight, volume, length or area. If a retailer changes prices for such goods via dynamic pricing, the basic price must automatically be adjusted accordingly. Otherwise, an incorrect or missing basic price may also constitute misleading information.

The regulations on discount advertising are also highly topical. Since May 2022, Section 11 PAngV requires the lowest total price charged by the retailer in the 30 days prior to the price reduction to be stated when advertising price reductions. According to Section 11 (1) PAngV, the lowest total price charged by the trader in the 30 days prior to the price reduction must always be stated when announcing a price reduction. This rule is intended to prevent artificially inflated prices from being used as a comparison (keyword “50% discount” after a previous short-term price increase). For dynamic pricing, this means that if a retailer advertises with crossed-out old prices or percentage discounts, they must be able to prove in case of doubt that the higher price actually applied as the total price at least 30 days before the start of the promotion. Otherwise, the advertising violates Section 11 PAngV. The practice of rapid price changes must therefore not be misused to circumvent the new 30-day rule. Although there are exceptions – Section 11 (4) PAngV mentions, for example, individually negotiated price reductions, price reductions that are not generally announced, permanently low prices or temporary price reductions for perishable goods – as a rule, the greatest care must be taken with strike prices.

A violation of the PAngV can also have consequences under competition law outside of discount campaigns. For example, Section 1 (6) PAngV makes it clear that mandatory information must be clearly recognizable and easy to read. If prices in an online store change very frequently due to automated adjustments, this must not make it difficult for consumers to find their way around. Prices should be clearly displayed and changes explained where necessary to ensure transparency. In addition to warnings, infringements may also result in fines, as the PAngV contains a regulation on administrative offenses. (Warnings from competitors are frequent in practice, as incorrect price information can be asserted as an infringement of competition law). In addition, the aforementioned UWG regulations (in particular Section 3a UWG) apply, which classify violations of the PAngV as an unfair breach of the law. In short: Correct price information is essential for dynamic pricing models – despite technical innovations, the clarity and truth of price labeling remain inviolable.

Price discrimination and equal treatment

The term price discrimination sounds negative, but initially describes the different pricing of one and the same product for different customers or customer groups in a value-neutral way. In German law – unlike in public price law in regulated sectors – there is no general prohibition on charging consumers different prices. In principle, a supplier may therefore offer different prices to different customers (e.g. through individual discounts or negotiations). It has always been observed, for example, that goods can be priced differently in different branches of a retail group (depending on regional purchasing power or local competition). This is part of free competition and can even take on consumer-friendly forms, such as more favorable rates for socially disadvantaged groups or time-limited special promotions.

However, certain forms of price differentiation come up against legal limits if they are linked to prohibited grounds of discrimination. The General Equal Treatment Act (AGG) prohibits discrimination on grounds such as race, ethnic origin or gender in mass retail. Applied to pricing, this means that if an online store were to systematically display higher prices to female customers than to male customers, for example, this could constitute unlawful discrimination on the basis of gender under the AGG. Different prices based solely on the customer’s origin (nationality) would also be problematic – the Geo-blocking Regulation (Regulation 2018/302) also applies here at EU level, which prohibits unjustified discrimination against customers based on their nationality or place of residence (Geo-blocking Regulation, Art. 3 and 4). A provider may not automatically redirect a foreign EU customer to more expensive country sites or refuse their purchase at the advertised price simply because they come from another member state.

In practice, explicit discrimination of this kind is rare; most price adjustments are based on economic criteria (timing, demand, customer loyalty) and not on protected characteristics. However, the use of big data could indirectly lead to unequal treatment, for example if algorithms use characteristics that correlate strongly with gender, for example. Caution is required here – not only for legal reasons, but also from a reputational perspective. If it becomes known that a company is systematically discriminating against certain groups, public outcry is certain. In purely legal terms, personalized prices are permissible as long as they do not violate specific prohibitions on discrimination and are not unfairly deceptive. The EU has deliberately not imposed a ban on personalized prices, but instead relies on transparency as a regulatory measure: consumers should be informed so that the market can punish inefficient or unfair pricing.

Antitrust law makes an exception in the case of dominant companies: According to Art. 102 TFEU and Section 19 ARC, the abuse of a dominant market position can also lie in discriminatory pricing. This means that if a quasi-monopolistic supplier arbitrarily charges certain customer groups higher prices, this could be considered an exploitative abuse. However, this scenario does not usually affect start-ups and most e-commerce providers as long as they do not amount to a monopoly. In fact, the Higher Regional Court of Stuttgart (judgment of 12.08.2019 – 10 U 15/17), for example, awarded compensation to a man because he was charged a higher admission price in a discotheque compared to women – a violation of the AGG.

The bottom line is this: Price differentiation is generally permitted in competition and is widespread. It is only prohibited in very limited cases, such as violations of the AGG or EU regulations on customer discrimination based on nationality. Nevertheless, companies should proceed sensitively and ask themselves how price differences are perceived by customers. It is often wiser to design price differentiation openly (e.g. as official discounts for certain groups) instead of secretly letting algorithms decide on sensitive characteristics.

Competition and antitrust law aspects

Algorithmic pricing can also pose challenges from the perspective of competition and antitrust law. On the one hand, dynamic pricing increases competitive pressure – providers undercut each other in real time. On the other hand, there is a risk that algorithms may unintentionally or intentionally lead to price alignment between competitors. Under antitrust law, any agreement or concerted practice to fix prices is strictly prohibited (Art. 101 TFEU, Section 1 ARC). Therefore, if companies use algorithms that communicate with each other or react to the same signals, this can lead to the formation of a cartel, even without a classic “round table” agreement. A well-known example from practice is a case in the USA in which online providers of poster art used price algorithms to coordinate their offers, which was classified as illegal collusion by the authorities. Companies must be aware that they are liable for the behavior of their algorithms: If they are programmed in such a way that they effectively maintain a parallel price or eliminate price competition, they could face severe penalties from antitrust authorities.

The issue of algorithmic collusion is the subject of intense debate. In 2019, the German Federal Cartel Office published a study on algorithms in competition together with the French competition authority, which warned of possible collusion mechanisms. Even without direct collusion, similar self-learning price algorithms in an oligopoly could tend to tacitly maintain a higher price level because any automated reduction would be immediately recognized and countered by competitors. Legally, this is a gray area between permitted parallel behavior and prohibited coordinated behavior. There have been few court rulings to date, but the antitrust authorities – such as the German Federal Cartel Office – are keeping a close eye on such developments. When using external price optimization software, companies should check whether it may share data with competitors or algorithmically promote cooperative behavior. If an external service provider makes the same algorithm available to several competitors, there is a risk of a hub-and-spoke cartel (a central player – the software provider – acting as a hub between competitors). Recently, such cases have also been pursued legally.

Another aspect is the restriction of competition through contract-based price maintenance on platforms. If, for example, a marketplace operator requires its sellers not to offer a lower price anywhere (so-called “price parity clauses”), this may be illegal under antitrust law. In the past, Amazon had such clauses with Marketplace sellers, but had to abandon them following investigations by the authorities. For algorithmic pricing, this means that a platform operator may not use algorithms to suppress cheaper offers outside its platform. Dynamic pricing algorithms should always reflect real competition and not artificially eliminate competitive alternatives.

Finally, reference should also be made to the UWG with regard to aggressive business practices. § Section 4a UWG prohibits impairing the consumer’s freedom of choice through inappropriate, unobjective influence. This also includes taking advantage of weak situations or predicaments. If, for example, an algorithm recognizes that a customer is in urgent need of a product (e.g. due to an emergency or time pressure) and drastically increases the price in a targeted manner, this could be considered an unlawful aggressive practice. Taking advantage of a predicament to charge exorbitant prices is also immoral under civil law in accordance with Section 138 BGB. Although such scenarios are extreme, they show the limits: Even with automated price control, the decency of business dealings must not be abandoned. A special measure in Germany is the “market transparency office” for fuel prices: Petrol stations change their prices, sometimes several times a day; to make it easier for consumers to keep track, all price changes must be reported in real time so that comparison portals can display them. This shows a regulatory reaction that focuses on transparency rather than prohibition and thus at least makes dynamic pricing more controllable.

Special contractual features and consumer rights

Automated price adjustments also affect general civil law principles of contract and consumer law. First of all, the moment the contract is concluded is important: In online retail, the purchase contract is usually concluded when the retailer accepts the customer’s order (for example, by confirmation email). A displayed price can change up to this point. This means that, in principle, the customer is not entitled to a price previously displayed on the website being permanently valid as long as they have not yet placed an order. In practice, however, consumers expect a certain price continuity during their purchase. If a price in the shopping cart were to suddenly increase, the customer could feel deceived. Companies should therefore – also for reasons of trust – avoid abruptly increasing prices during an ongoing session, or at least clearly indicate this (e.g. “Price update due to interim change”).

If it transpires after an order that an algorithm has displayed an obviously incorrect price (e.g. €100 instead of €1000 due to a technical error), German law does offer correction options – the retailer can contest the contract on the grounds of error (Section 119 of the German Civil Code). However, this “emergency brake” should not be included as part of the business model. On the one hand, the error must be recognizable to the customer (this may be the case for blatant errors such as 90% below the market value, but not for smaller differences). On the other hand, there is a risk of negative customer reactions and loss of trust if errors are contested on a large scale. It is therefore essential to carefully monitor the price algorithms and set plausibility limits to prevent gross mispricing.

In the case of continuing obligations (subscriptions, memberships), the question arises as to the extent to which dynamic price changes are permissible. The following applies here: Agreed prices are binding for the duration of the contract, unless the contract contains a valid price adjustment clause. A clause that grants the company a unilateral right to increase the fee at will after conclusion of the contract would be invalid under the law on general terms and conditions (Section 307 of the German Civil Code (BGB)) because it unreasonably disadvantages the customer. However, index or cost clauses (such as linking to a published price index) or the right to increase prices under certain circumstances are permissible, provided that the customer is granted a right of termination. For example, a SaaS provider can stipulate in its terms of use that the subscription fee may be adjusted annually, but must then inform the customer in good time and give them the opportunity to terminate the contract before the increase comes into effect if they do not agree. Particularly in the B2C sector, the law strengthens consumer protection here – for example through Section 41 (3) TKG (for telecommunications) or new requirements in the energy sector – which can serve as an analogous benchmark.

Finally, consumer rights such as withdrawal and warranty must be observed. Although they do not directly affect pricing, they do indirectly: a dissatisfied customer who feels they have paid an inflated price could make use of the 14-day right of withdrawal for online purchases and cancel the purchase. Even if this does not imply a breach of law by the retailer, it does have economic consequences. All the more reason for reputable providers to have an interest in customers accepting their price as fair. In this respect, legal compliance and good customer experience management complement each other when it comes to pricing. Incidentally, a consumer who suspects that they have received an unfairly inflated price could contact the consumer advice center or – if a discrimination characteristic is affected – assert an AGG claim for compensation. Such claims are rare, but the possibility acts as a disciplinary sword of Damocles.

Data protection and automated decision-making

Where prices are personalized, the processing of personal data is often involved – such as purchase history, location, device type or usage behavior. This is where the General Data Protection Regulation (GDPR) comes into play. Companies must have a legal basis for data processing (usually based on legitimate interests, Art. 6 para. 1 lit. f GDPR, as personalized prices can be used to optimize business activities). However, careful consideration is required: the legitimate interests of the company in price differentiation must be weighed against the interests of the data subject in fair treatment. In view of the element of surprise in personalized prices, this is not a foregone conclusion – the less transparent the practice, the more likely it is that a data protection supervisory authority could raise doubts about its legality. Consent (Art. 6 para. 1 lit. a GDPR) as a basis would be theoretically possible, but difficult to implement in practice (hardly any customer would actively agree to pay more than others).

Furthermore, personalized pricing can be considered an automated decision in individual cases (Art. 22 GDPR) if it is made without human intervention and has a significant impact. Whether a different price already constitutes a “significant impact” is a matter of judgment – if the amounts involved are small, this is probably not the case, but it is in the case of larger differences (Recital 71 GDPR also mentions the automatic rejection of an online credit application or the setting of individual prices based on data as examples). In such cases, data subjects would have a right not to be subject exclusively to such a decision and could, if necessary, request that a person review the decision. The right to information (Art. 15 GDPR) also includes information about the logic involved in an automated decision: a customer could therefore ask why they received a certain price. Companies should be prepared to provide a comprehensible explanation (e.g. “You have received a discount for new customers” or “Your price was based on the current very high demand”). Black box AI models whose results cannot be explained could run into compliance problems here.

In general, it is also advisable to observe data protection principles such as data minimization and privacy by design in AI-based pricing systems. Only data that is relevant and permitted for price optimization should be used. Particularly sensitive data (e.g. ethnic origin, health, income as such) must never be used in a discriminatory manner – this would not only be ethically wrong, but also legally inadmissible. If, for example, an AI indirectly deduces income or willingness to pay from place of residence or user data, great care must be taken to ensure that no hidden discrimination takes place. In practice, transparency is the best approach here too: being open about the use of data for pricing purposes in the data protection notice and towards customers creates trust and reduces the legal risk.

European requirements and international perspectives

The German regulations on fair trading law and price transparency are strongly influenced by European requirements. Many provisions are based on EU directives that apply similarly in all member states. It is also worth taking a look across the border – particularly in the USA and the UK – to see how dynamic pricing is dealt with there.

EU law: UCP Directive, Digital Services Act, Digital Markets Act

The basic guidelines in consumer law are set by EU law. The UCP Directive 2005/29/EC (Unfair Commercial Practices Directive) forms the basis for the UWG and ensures that misleading or aggressive practices are prohibited throughout Europe. Specifically on the subject of pricing, the directive does not contain an explicit ban on dynamic prices, but it does impose an obligation to maintain truth and clarity in the presentation of prices – this is reflected in Section 5 UWG and the annex (Black List). The Omnibus Directive (EU) 2019/2161, which was implemented in 2022, modernized these requirements: Retailers must now indicate personalization in the case of personalized prices and, in the case of price reductions, indicate the previous price of the last 30 days throughout the EU. This harmonization ensures that similar information obligations also apply in other EU countries such as France, Italy or Spain, so that companies with international business have uniform standards here. (Back in 2018, the EU Commission conducted a mystery shopping exercise to investigate how often personalized prices occur online. The result showed only a few cases of genuine price personalization, but the legislator wanted to proactively prevent abuse).

The Digital Services Act (DSA) and the Digital Markets Act (DMA), in force since 2022, have further indirect effects: The DSA, as a new EU law for online platforms, specifically prohibits “dark patterns” (see Art. 25 DSA) – i.e. manipulative design elements that entice users to take unwanted actions. In terms of pricing, this means that misleading countdown timers (“Only 5 minutes left at this price!”) or constantly changing prices designed to confuse users are viewed critically. Online platforms must ensure that their user interfaces are transparent and not misleading. Although the DSA is primarily directed against disinformation and unsafe content, the requirement for a fair, non-manipulative design should also include pricing mechanisms.

The DMA in turn addresses large gatekeeper platforms (e.g. large marketplaces or app stores) and stipulates fairness in dealing with business customers and end customers. For example, a gatekeeper may not use the data it receives from third parties on its platform to gain an unfair advantage in competition (see Art. 6 (5) DMA). In terms of pricing, this means that a platform operator such as Amazon, which sells products itself, may not use confidential market data from competing retailers in order to systematically set its own prices lower with the help of algorithms and outbid the competition. The DMA also prohibits discriminatory terms and conditions, such as the aforementioned price parity clauses. A gatekeeper must also allow the use of other sales channels – retailers should be free to offer other (including lower) prices outside the platform. All of this helps to ensure that competition for the best price is not distorted by positions of power. For innovative companies that are not gatekeepers themselves, EU law as a whole signals that dynamic pricing remains permissible, but transparency and fair competition must be guaranteed.

United States (USA)

Dynamic pricing is widespread and widely accepted in the USA. Famous examples are the airline and hotel industry with their yield management and transportation services (Uber & Co.) with surge pricing. There is no general law prohibiting personalized or dynamic pricing to consumers at the federal level. US law relies more on the free market: as long as there is no deception or fraud, companies are allowed to set their prices flexibly. However, general consumer protection rules also apply here. For example, the Federal Trade Commission (FTC) can take action against deceptive pricing – i.e. when consumers are deceived by false information about price reductions (comparable to the bogus discount ban in the Unfair Competition Act). The FTC has also identified “drip pricing” as a misleading practice if mandatory fees are only added at the end of the ordering process.

Price gouging laws exist at state level in the USA. These laws – enshrined in many states – prohibit extreme price increases for essential goods during emergencies (e.g. natural disasters). For example, the price of gasoline, drinking water or generators may not suddenly increase many times over just because a hurricane is approaching. These regulations aim to protect consumers from unscrupulous exploitation of exceptional situations. Uber, for example, came under fire when it charged exorbitant surge prices during a snowstorm; the company vowed to do better and has since set upper limits in emergencies. Outside of such extreme situations, however, there is no equivalent to price fixing – for example, online retailers are generally allowed to display different prices to different customers.

However, American lawyers and consumer advocates are also discussing the implications of personalized prices. The main issues here are discrimination against certain groups and competition. Unlike in the EU, there is no comprehensive ban on discrimination in B2C business – although the Civil Rights Act prohibits discrimination in public places of business on the basis of skin color or religion, for example, pricing is hardly covered as long as no one is excluded from making a purchase. Nevertheless, a company that varies prices based on, for example, residential districts (which often indirectly represent ethnic or income-related segments) would risk being legally and reputationally attacked. In the context of competition, a recent case made the headlines: Several large property managers used software (RealPage) that algorithmically optimized rental prices for apartments, allegedly leading to inflated rents as competition between landlords was reduced. The US Department of Justice has therefore launched an investigation into possible antitrust violations. This shows that even in the USA, algorithmic price control has its limits where it leads to collusion or exploitation. Although the Robinson-Patman Act of 1936, which prohibits anti-competitive price discrimination between resellers, applies to business transactions between companies, it is of little relevance to end consumer prices.

Overall, the approach in the USA is pragmatic and market-based: price discipline is achieved through competition and consumer behavior. If customers perceive a provider as unfair, they switch to a competitor or public pressure is exerted. Legal intervention occurs selectively (in cases of fraud, emergencies or genuine abuse of power). For international start-ups, this means that those serving the US market have a largely free hand when it comes to pricing, but should adhere to the general guidelines (no deception, no illegal exchange of information with competitors, no exploitation of disasters) and consider the impact on their own reputation.

Great Britain (UK)

Since leaving the EU, the UK has had its own rules, although these are largely based on the previous EU standards. British consumer law (Consumer Protection from Unfair Trading Regulations) prohibits misleading and aggressive business practices in line with the UCP Directive. Dynamic pricing itself is not prohibited in the UK either. British companies use it just as intensively as on the continent – be it in online retail, for ticket prices for events or in local public transport (keyword “peak/off-peak” fares). Here, too, it is important that there is no misleading information. For example, the UK Competition and Markets Authority (CMA) has investigated cases in which online retailers operated with constantly changing discount prices, which impaired transparency.

The CMA (Competition and Markets Authority) published a study on personalized online prices in 2018. It found that genuine personalization at individual customer level has been relatively rare in UK online retail to date – more common are segmented offers (such as special discounts for members, personalized voucher promotions or regional differences). However, the authority emphasized that the same legal limits apply in principle: Price differentiation must not be deceptive or violate discrimination laws. It is interesting to note that the UK also has mechanisms in its data protection legislation (similar to the GDPR, which applied there until 2020) that could restrict extremely personalized profiling. Although the UK has not adopted the new EU transparency obligations on personalized prices 1:1 after Brexit, British retailers are also advised to disclose when prices are personalized as a matter of good business practice. In 2021, for example, the British Financial Conduct Authority (FCA) decided to ban “price walking” in the insurance industry – existing customers are no longer allowed to systematically pay higher premiums than new customers when renewing a policy.

In terms of competition law, UK antitrust law largely corresponds to EU law. The discussion about algorithmic price coordination is also taking place here. The CMA has warned in reports that pricing algorithms could enable new forms of collusion, and it is monitoring markets with little competition (e.g. fuel prices, online marketplaces) very closely. To date, there has been no precedent in the UK of a company being punished for purely algorithmic collusion – but the authorities have clearly indicated that they would intervene if such a case were to arise. UK courts have also considered price discrimination to be a breach of contract in individual cases where, for example, regular customers were disadvantaged compared to new customers and this was contrary to the terms of the contract. This shows an interesting difference in expectations: in the UK, there is a strong focus on customer fairness, often without explicit legislation, because the media and the public keep a watchful eye.

This means for companies with UK business: Essentially, dynamic pricing can be handled in a similar way to the EU. However, you should stay up to date, as the UK could go its own way in terms of regulation – for example, the government is planning a Digital Markets Act that will regulate large online companies more closely (similar to the EU DMA). Anyone operating platforms or providing pricing tools should therefore also follow British developments. Reputation with British customers is just as valuable: transparency and fairness are soft factors that determine success on the market – a shitstorm in the British press can be just as damaging to business as a warning letter in Germany.

Moral and reputational aspects

Many things may be legally permitted – but what is technically feasible and legally permissible is far from being perceived as fair by consumers. Dynamic pricing has a particular impact on customers’ sense of fairness. If two customers pay different prices for the same product, many instinctively feel that they are being treated unfairly. In a survey, most consumers would probably agree that “same price for all” seems fair, even if the reality has always been different (discounts, special offers, etc.). In fact, a 2018 study by the German Council of Consumer Experts found that around 90% of respondents felt that personalized prices were unfair. The awareness that today algorithms may only show “the best price” for some can fuel mistrust. Companies should not underestimate this psychological component.

Cases of personalized pricing have already led to public outcry, so-called shitstorms, on several occasions. Ticketmaster also caused frustration in the event sector in 2022 when it offered concert tickets with a dynamic pricing model and prices rose sharply when demand was high – fans accused the company of profiteering at the expense of real fans. An early example is Amazon in 2000: the company tested variable DVD prices at the time and showed different customers prices between 23.24 and 26.24 US dollars for the same DVD. When this practice became public, customers were outraged by the “secret prices”. Amazon publicly apologized and refunded the difference to affected buyers – and vowed not to conduct any more such experiments. This incident is still cited today as a warning of how quickly reputations can be damaged if customers get the impression that they are being deceived.

Similarly controversial was the debate surrounding a proposal by Coca-Cola in the 1990s to introduce vending machines that automatically charge higher prices in hot weather. The idea was torn apart in the media as exploitation of thirsty customers, so the company never put it into practice. And the introduction of surge pricing at Uber has dogged the company for years: Although Uber factually justified its price increases when demand was high by saying it wanted to get more drivers on the road, in extreme cases (such as price explosions on New Year’s Eve or during crisis situations) the mood changed. In New York, for example, following criticism, Uber signed an agreement not to raise fares above the local price gouging limit in emergencies in order not to lose the public’s trust.

In Germany, too, reports of personalized prices have repeatedly made headlines. Whether it’s the railroad, which has drastically different ticket prices depending on the time of booking, or online stores where users suspect they are seeing more expensive offers with Mac computers – the suspicion alone can lead to negative headlines. (For example, the advice that users should delete their browser cookies or surf in incognito mode before booking online in order to get a better price persists to this day – a clear indication that many users suspect personalized price control). Most larger retailers have therefore officially denied using personalized prices in individual cases. They fear a loss of trust if customers believe that the price could get worse with every click. However, trust is a valuable commodity in e-commerce: start-ups and new platforms in particular first have to build up customer loyalty and can hardly afford public outrage. A single tweet or post accusing a company of “ripping you off” can go viral and generate immense pressure. The only way to prevent this is to avoid making such accusations in the first place.

To prevent moral risks, companies should follow certain guidelines for fairness. Firstly: Transparency towards the customer – anyone who openly communicates why a price fluctuates (e.g. “time-of-day tariff” or “high demand – price increases”) is more likely to gain understanding than someone who changes prices without comment. Secondly: No personal discrimination against loyal customers – if regular customers feel that they are being taken advantage of (e.g. paying higher prices than new customers), they turn away disappointed. This phenomenon of the “loyalty penalty” has been sharply criticized in the UK, for example, in the case of insurance and telephone contracts. Thirdly: Moderation in the event of price jumps – extreme and rapid price fluctuations should remain the exception. A sudden massive price increase without a comprehensible reason leaves a bad impression. Fourthly: Goodwill and communication in an emergency – if displeasure does arise, a proactive approach (explanations, price adjustments or vouchers as compensation if necessary) helps to win back trust.

Finally, social media also plays a role: complaints and price comparisons now spread in real time. Companies must expect clever customers to uncover price differences and share them publicly (e.g. screenshots from different devices or accounts). In 2012, the Wall Street Journal reported that the travel platform Orbitz prioritized more expensive hotel offers to Mac users based on the knowledge that they spend more on average. Although Orbitz emphasized that no one got the same room more expensively than others, the case fueled the perception that device- or data-based price differences are a reality. In the long term, dynamic pricing will only be successful if it is accepted by customers as reasonably fair. This requires sensitivity, communication and sometimes the willingness to subordinate short-term profit maximization to longer-term relationship building with the customer.

Guidelines for companies: Implementing dynamic pricing in a legally compliant and fair manner

Finally, we summarize the most important findings and provide practical recommendations. When introducing AI-based pricing models, companies – whether start-ups or established companies – should not only focus on profit, but also on legal compliance and customer trust.

What is allowed, what is forbidden?

In principle, automated pricing is permitted as long as it does not violate specific prohibitions. The following are permitted, for example:

• Flexible price changes according to demand, time or stock levels (classic dynamic pricing), provided that the applicable final price is displayed correctly.
• Different prices on different sales channels or for different customer segments (2nd and 3rd degree price differentiation), provided there is no discrimination on prohibited grounds (e.g. gender, origin).
• Personalized discounts or individual offers (e.g. loyalty discount, personalized vouchers), because these put the customer in a better position, not a worse one.

In contrast, the following are not permitted or risky:

• Misleading price information: Fictitious crossed-out prices, fictitious discounts, hidden surcharges or inconsistent information (e.g. price without VAT to consumers) are prohibited.
• Breach of information obligations: Failure to provide information on personalized prices or failure to indicate the 30-day advance price for reductions can result in warnings and fines.
• Discriminatory pricing: Price differences that are directly linked to protected characteristics (gender, ethnicity, etc.) violate the AGG and fundamental legal principles. Such criteria must not play a role in algorithms.
• Violations of antitrust law: Any agreements or concerted algorithms with competitors to jointly control prices are strictly illegal. Implicit collusion via shared software is also a minefield.
• Exploiting predicaments: Extreme price increases in emergency situations (usurious prices) are inadmissible and also ruin reputations.

In short, market and demand-oriented price adjustments are permitted within the framework of the law. Deception, discrimination or anti-competitive agreements are not permitted. Where the line is drawn has been explained in detail above – if in doubt, you should opt for the more conservative, transparent option.

Implementing AI-based pricing models in a legally compliant manner

Companies should take an interdisciplinary approach to the technical implementation of algorithmic pricing: Bring lawyers, data scientists and sales experts to the table. The following best practices can help:

• Compliance by design: Take legal requirements into account as early as the development stage of the pricing software. For example, provide parameters that ensure that price reductions/increases are made in accordance with the PAngV (keyword: 30-day rule) and that no impermissible criteria are included.
• Transparency modules: The software should make it possible to provide the customer with information at the appropriate point (for example, automatically displaying “Your price has been personalized” if appropriate logic has been used). Transparency is also needed internally: the company needs to understand why the algorithm sets certain prices – explainable AI is a plus.
• Monitoring and audit: Algorithmic systems must not run completely uncontrolled. Set up limits (e.g. max. price X, min. price Y, max. change frequency per day) and check the price data regularly. This allows you to quickly identify anomalies (e.g. prices that are unusually high or low). An internal audit can also ensure that certain customer groups are not systematically disadvantaged.
• Documentation & FAQ: Record the criteria used by the pricing algorithm. This will enable you to answer customer queries and prove to the authorities that no prohibited criteria are used. A FAQ section on the website can proactively clarify frequently asked questions about pricing.
• Training and awareness: Management and the pricing team should be trained to know the legal rules of the game. Do not trust blindly, especially when dealing with AI: Ultimately, the entrepreneur bears the responsibility and should be allowed to ask critical questions of the technology.

Platform operator vs. retailer: differences in the legal framework

Platform operators (e.g. operators of online marketplaces or brokerage platforms) face the challenge of satisfying and legally protecting both their end customers and their affiliated retailers. Key differences compared to retailers:

• Consumer relationship: If the platform itself is a party to the contract with the consumer (such as Uber as the provider of the transportation service), all consumer rights apply directly to it. If, on the other hand, the platform is only an intermediary between the trader and the customer (such as eBay or a marketplace), it must above all comply with the statutory information obligations (e.g. indicate which provider is selling and display the correct price). According to the DSA, marketplaces must ensure that retailers provide lawful information on their platform – e.g. that they post real prices and not prohibited bait-and-switch offers.
• Rules for retailers: Platform operators should set out clear guidelines on pricing in their terms of use. For example, a ban on advertising prices on the platform that are then immediately raised (to avoid warnings due to bait-and-switch offers). It should also be made clear that all prices must be final prices including taxes and that discount campaigns must comply with legal requirements. Although the platform is not obliged to check every price manually, it could also be held responsible in the event of systematic violations by a retailer (e.g. constantly unclear prices).
• Own pricing: Many platforms have two types of pricing: the retailers set their product prices and the platform charges additional fees (e.g. commission, service fee for buyers). These platform fees must be communicated just as transparently. A typical mistake is when only the retailer price is displayed at the beginning and the platform’s “service fee” only appears in the last order step – this could lead to accusations of drip pricing. It is better to clearly indicate the additional costs at the beginning.
• DMA/antitrust law: Large platforms with gatekeeper status must be particularly careful not to exploit their market power. They should not use tools that centrally control the prices of all retailers (risk of price harmonization) and should not misuse retailers’ data to make their own offers cheaper. Fairness towards the affiliated providers is not only legally required, but also necessary to maintain the ecosystem.

Special features of SaaS and subscription models

SaaS providers and subscription services often have recurring revenue and individual customer relationships. Here are some tips:

• Communicate a clear price structure: Especially with subscription models, customers expect transparency about ongoing costs. If dynamic prices apply depending on usage or time, this should be made clear in the contract and marketing (e.g. “User-dependent billing: If your usage is higher, your price for the following month will increase by X”).
• Price adjustment clauses: For longer-term contracts (annual subscription, SaaS contract with a term), it is essential to include a fair price adjustment clause if you want to make future price changes. This should state objective reasons (inflation, scope of services) and grant the customer a right of termination. Without such a clause, you are bound to the originally agreed price for the fixed term.
• A/B testing with caution: Startups like to try out different price points. This is legitimate as long as new customers receive different offers – but existing customers should not suddenly find that others are paying much less. So if you do market tests, the spread should not seem unfair or you should be prepared to adjust if complaints accumulate. This is less critical in the B2B SaaS sector (individual offers) than in the B2C mass market.
• Appreciate regular customers: Don’t annoy long-standing users with sudden price jumps. On the contrary, it can make sense to offer them stable conditions or bonuses instead of just favoring new customers. This is how you promote loyalty.
• Communicate added value: Customers are price-sensitive when it comes to subscriptions. Dynamic pricing can be communicated positively here by emphasizing the added value – for example, that the customer also pays less in weak months (if the model is variable both upwards and downwards). Those who only increase prices dynamically but never offer discounts risk resentment.

Blockchain-based platforms and decentralized markets

Blockchain platforms and decentralized marketplaces face the dilemma that they often want to operate outside of traditional regulatory frameworks, but the real laws still apply. For example, if a platform allows users to set dynamic prices for digital assets (such as NFTs or tokens) via smart contracts, the same principles apply in principle: Clarity and truth of pricing information, no deception. Even if “code is law” is the motto, operators (if identifiable) should ensure that users are informed about important features of pricing (e.g. that a smart contract automatically increases the price after each purchase – this principle must be explained in advance). Example: If an NFT marketplace offers a so-called “Dutch auction” (price drops over time), the way the price drop works must be clearly explained to bidders. Furthermore, decentralization must not serve as a cover for legal violations: If a seemingly decentralized marketplace is operated or significantly controlled by a company in the background after all, this company is liable for breaches of consumer law. Regulatory authorities are increasingly looking at crypto platforms, also with regard to fraud prevention and consumer protection.

Those who offer innovative blockchain pricing models (such as dynamic auction mechanisms) have the opportunity to build trust through self-regulation: For example, through voluntary guarantees (a kind of “smart contract auditing” or insurance in case something goes wrong) and by complying with known standards from traditional e-commerce. In this way, customers can be made to feel secure even in novel environments.

Conclusion

Automated and dynamic pricing in e-commerce is a double-edged sword: it opens up considerable sales potential for companies and can benefit consumers in the form of flexible offers, but places high demands on legal compliance and fairness. In Germany and the EU, a tightly meshed network of UWG, PAngV, consumer protection and competition law forms the framework within which dynamic pricing must operate. International examples from the USA and UK show that transparency and trust are the decisive currencies – legal leeway should never be selfishly pushed to the limits of what is permissible without keeping the customer perspective in mind.

For innovative start-ups, SaaS providers, platform operators and tech companies, one thing is certain: Legal certainty and customer acceptance go hand in hand. If you want to establish an AI-based pricing model, you should seek legal advice early on, create internal compliance structures and communicate your pricing strategy openly. Dynamic pricing is most successful when customers feel that they are benefiting from dynamic offers rather than being left out. Ultimately, it boils down to responsible use of the available technology. Companies that think long-term use dynamic pricing in such a way that customers perceive it as a service (e.g. a favorable price at times of low demand) and not as a rip-off. This sometimes requires restraint on the part of management: the algorithm may generate more profit in the short term if it raises prices to the maximum – but the long-term costs of a loss of trust can wipe out these gains. A balance needs to be struck between efficiency and fairness. If this can be achieved, nothing stands in the way of a successful, innovative and customer-oriented pricing policy.