Competition Law and Algorithmic Customer Allocation Concerns

1. Introduction

Algorithmic customer allocation concerns arise when competitors use algorithms, software platforms, artificial intelligence, shared data, or automated decision systems in ways that divide customers or reduce competition for particular customers.

Traditional customer allocation occurs when competitors agree that each will serve particular customers, territories, customer groups, or accounts and will refrain from competing for customers assigned to another competitor. U.S. competition authorities describe horizontal customer and territorial allocation among competitors as a classic form of market division and generally treat a naked allocation agreement as per se unlawful under Section 1 of the Sherman Act.

Algorithms do not fundamentally change that principle. The important legal questions are whether there is an agreement or other legally sufficient coordination between independent competitors, what role the algorithm plays, and whether the arrangement restricts independent competition.

2. What Is Algorithmic Customer Allocation?

Algorithmic customer allocation can occur when an automated system determines which competitor receives, targets, bids for, or avoids particular customers.

For example, competing firms could participate in a system that effectively divides:

• individual customers;
• customer accounts;
• geographic territories;
• categories of customers;
• sales opportunities;
• tenders or procurement contracts; or
• particular segments of an online marketplace.

The competition concern becomes especially serious where competitors knowingly use a common system to implement an understanding that they will not compete for one another's assigned customers.

Traditional customer allocation can allow each participant to exercise greater market power over its assigned customers because other cartel members no longer compete aggressively for those customers. DOJ economic analysis has explained that customer or territorial allocation can reduce the need for cartel members to compete on price and may require monitoring mechanisms to determine whether participants are respecting their allocations.

3. How Algorithms Can Facilitate Allocation

An algorithm can potentially perform several functions that previously required direct human coordination.

It may classify customers according to location, purchasing history, profitability or other characteristics and direct them toward particular sellers. It could also determine which competitors should bid for particular contracts, suppress solicitation of customers assigned elsewhere, or monitor whether participating firms are departing from an agreed allocation.

A more sophisticated system could detect when a customer moves from one supplier to another and automatically modify recommendations, bids, commissions, prices or customer-routing rules.

The legal significance depends on the surrounding facts. Merely using similar software or independently adopting algorithms does not by itself establish an unlawful agreement.

4. Agreement Versus Independent Algorithmic Conduct

This distinction is central.

Suppose Company A independently develops software deciding that it will concentrate on large corporate customers while Company B independently focuses on smaller customers. Parallel specialization alone does not necessarily establish a competition-law violation.

The position is very different if A and B agree to divide those customers and then use software to enforce their agreement.

Accordingly, investigators may examine evidence such as communications among competitors, common algorithm providers, instructions given to software developers, data-sharing arrangements, changes in solicitation patterns, account-routing rules, bidding records and whether competitors systematically stop pursuing particular customers.

The requirement for sufficient evidence of an agreement remains important. In BanxCorp v. Bankrate, for example, the court emphasized that allegations of market division require adequate factual support showing the alleged agreement rather than conclusory allegations alone.

5. Hub-and-Spoke Algorithmic Structures

An especially important modern issue is the hub-and-spoke model.

The structure can be represented as:

Competitor A → Common Algorithm ← Competitor B

Competitor C → Common Algorithm ← Competitor D

The software provider or platform acts as the hub, while competing businesses are the spokes.

A competition-law issue can arise where competing firms knowingly participate in a common arrangement through that intermediary and the legally required agreement or concerted coordination can be established.

However, the existence of a common software provider does not automatically prove that all users have entered into a horizontal customer-allocation agreement. Evidence about knowledge, communications, contractual arrangements and actual operation of the system remains important.

6. Allocation Through Automated Lead Distribution

Digital platforms frequently distribute leads among sellers.

That practice is not inherently unlawful. A platform may legitimately allocate leads to improve efficiency—for example, according to location, capacity, expertise or customer preference.

Competition concerns become much stronger where competing businesses themselves agree that an algorithm will divide customers and prevent them from competing for customers assigned to another participant.

The distinction is therefore between legitimate platform organization and a mechanism implementing horizontal market division.

7. Allocation Through Bid Algorithms

Customer allocation can also interact with bid rigging.

Imagine several competitors bidding for contracts. Software could theoretically identify a predetermined winner for each customer while instructing other participants not to bid or to submit noncompetitive bids.

The resulting arrangement could involve both customer allocation and bid coordination.

DOJ enforcement guidance identifies patterns such as competitors repeatedly avoiding particular customer groups or different groups of contractors consistently winning particular categories of customers as possible indicators requiring investigation.

8. Allocation Through Customer Data

Customer data can make algorithmic allocation easier to implement and monitor.

A common platform may possess information about:

customer identity → historical supplier → purchase volume → location → expected demand → competitor activity.

If competitors improperly coordinate through such a system, detailed data could help determine which customers belong to which participant and detect deviations from an allocation arrangement.

This monitoring function matters because cartels frequently require mechanisms for detecting cheating. DOJ economic analysis has specifically recognized monitoring as relevant to maintaining customer-allocation arrangements.

Important Case Laws and Enforcement Precedents

There is still a distinction between traditional customer-allocation precedents and cases specifically involving modern AI-driven allocation. Consequently, established market-allocation cases provide the legal foundation, while newer digital and algorithmic matters illustrate how those principles can apply to technology-assisted conduct.

1. United States v. Topco Associates, Inc., 405 U.S. 596 (1972)

This U.S. Supreme Court decision is one of the foundational market-allocation cases.

Topco was an association involving independent supermarket operators. Restrictions allocated territories in which members could sell certain products.

The Supreme Court treated agreements between competitors at the same market level allocating territories to minimize competition as a classic per se violation of Section 1 of the Sherman Act.

Importance for algorithms

If competitors use an algorithm simply as the technological mechanism for implementing a horizontal territorial or customer division, automation would not ordinarily transform the underlying restraint into something fundamentally different.

Topco therefore supplies an important doctrinal foundation for analyzing digitally implemented market allocation.

2. United States v. Suntar Roofing, Inc., 897 F.2d 469 (10th Cir. 1990)

This case directly concerned customer allocation.

The Tenth Circuit held that an agreement to allocate or divide customers between competitors in the same horizontal market constitutes a per se violation of Section 1.

The defendants therefore could not escape the applicable legal rule merely by attempting to demonstrate that their allocation was reasonable.

Algorithmic relevance

The principle can apply where software performs the allocation mechanically:

Competitors agree → customers are divided → algorithm implements division.

The fact that software rather than employees performs the operational allocation does not eliminate the underlying competition concern.

3. United States v. Cooperative Theatres of Ohio, Inc., 845 F.2d 1367 (6th Cir. 1988)

This case involved competitors and customer allocation in the motion-picture exhibition context.

The Sixth Circuit characterized customer allocation as the kind of naked restraint capable of triggering per se treatment.

Algorithmic relevance

The case reinforces the proposition that competition law focuses on the economic nature of the agreement.

Consequently, competitors cannot necessarily avoid established customer-allocation principles merely because allocation occurs through automated software rather than through manually prepared customer lists.

4. United States v. Cadillac Overall Supply Co., 568 F.2d 1078 (5th Cir. 1978)

This decision is another significant customer-allocation precedent.

The case involved competitors in the industrial garment-supply business and supported the treatment of horizontal customer allocation as a serious Sherman Act violation.

It was subsequently cited in Suntar Roofing as authority concerning the per se status of customer-allocation agreements.

Algorithmic relevance

Modern technology can automate conduct that historically required sales managers to maintain lists identifying which customers belonged to which competitor.

The underlying legal question remains whether competitors agreed to eliminate or materially restrict competition for those customers.

5. Hammes v. AAMCO Transmissions, Inc., 33 F.3d 774 (7th Cir. 1994)

This case is particularly interesting from a technology perspective.

The dispute considered an alleged mechanism involving customer allocation through telephone-call routing. Later courts described the arrangement as involving automatic call forwarding from supposed dealers located around boundary areas.

The concern was that customers who otherwise might have had meaningful choices between competing dealers could instead be automatically allocated.

Algorithmic relevance

This provides a useful bridge between traditional allocation and modern automated systems.

Automatic routing technology can influence which competitor receives a customer. Modern recommendation engines, digital marketplaces and AI-driven lead systems can perform a much more sophisticated version of that function.

The Eighth Circuit later relied on Hammes when discussing the established treatment of customer-allocation agreements.

6. In re Wholesale Grocery Products Antitrust Litigation / D&G, Inc. v. SuperValu, Inc.

The litigation concerned allegations involving grocery wholesalers and an arrangement that allegedly divided customers and geographic areas.

The Eighth Circuit concluded that evidence could permit a reasonable jury to find an agreement dividing territory and customers geographically. If such an agreement were established, it could constitute a per se antitrust violation.

Importantly, the appellate court did not itself determine that the defendants had committed the alleged violation; factual issues remained for determination.

Algorithmic relevance

The case demonstrates that customer allocation need not always involve an explicit permanent prohibition stating that one competitor can never serve another's customers.

Courts examine the actual competitive arrangement and evidence surrounding it.

7. United States v. Kemp & Associates

This prosecution concerned companies providing heir-location services.

The indictment alleged an agreement under which competitors allocated certain customers or estates rather than competing fully against each other.

The Tenth Circuit reiterated the established principle that agreements dividing customers between competitors in the same horizontal market are per se unlawful. The unusual structure of an industry does not by itself prevent established customer-allocation doctrine from applying.

Algorithmic relevance

This principle matters for new technology markets.

A defendant cannot necessarily argue:

“Our industry uses AI, so ordinary allocation principles do not apply.”

Courts can instead examine the economic substance of the arrangement.

8. EU Cathode Ray Tubes Cartel

European competition enforcement provides another useful example.

Evidence in the cathode-ray-tube cartel showed competitors discussing market shares and allocations concerning individual customers. European Commission findings described detailed discussions involving market shares for major customers and arrangements concerning customer allocation.

Algorithmic relevance

Modern software could make this kind of monitoring dramatically easier.

Instead of cartel participants manually comparing sales figures, an automated platform could potentially calculate customer shares, identify deviations and generate allocation recommendations. The legal concern would still depend on proving the prohibited coordination required by applicable competition law.

9. Recent Indian Illustration: HP India and Reseller Conduct

A recent Competition Commission of India matter concerning HP India and resellers illustrates customer-allocation concerns in electronic procurement.

The investigation considered communications, tender material and other evidence concerning coordination among participants. The Commission's decision described designated government-department accounts and customer allocation alongside bid-rigging concerns, ultimately finding contraventions of Sections 3(3)(d) and 3(1) for the conduct within the investigated period.

Although this was not principally an autonomous-AI customer-allocation case, it demonstrates how digital procurement systems do not prevent traditional allocation and bidding theories from applying.

10. EU Competition-Law Position

Under EU competition law, horizontal agreements allocating markets or customers can constitute serious restrictions of competition under Article 101 TFEU.

The analysis should nevertheless distinguish horizontal cartel arrangements from legitimate vertical distribution systems.

For example, EU rules recognize circumstances in which suppliers may allocate exclusive territories or customer groups to distributors. Such vertical arrangements require their own legal analysis and should not automatically be equated with a cartel among competitors. EU guidance even provides examples where exclusive customer allocation can potentially satisfy the conditions for exemption where appropriate efficiency and competitive conditions exist.

Thus:

Horizontal competitor allocation
Competitor A ↔ Competitor B
→ particularly serious concern.

Vertical distribution arrangement
Manufacturer → Distributor
→ requires contextual analysis under vertical-restraint rules.

11. Tacit Algorithmic Coordination

One of the hardest questions arises where there is no proven agreement.

Imagine independent competitors each using self-learning algorithms. The algorithms observe competitors' behavior and eventually settle into patterns under which each firm concentrates on different customers.

Economically, the result might resemble customer allocation.

Legally, however, similarity of outcome does not automatically establish the agreement required by provisions such as Section 1 of the Sherman Act.

Authorities therefore have to distinguish among:

explicit algorithmic cartel → competitors agree and use software to execute it;

algorithm-assisted agreement → humans establish coordination while algorithms monitor or implement it;

common intermediary arrangement → several competitors use the same platform or software provider, raising questions about whether legally sufficient horizontal coordination exists;

independent algorithmic parallelism → independently operating systems produce similar conduct without a proven agreement.

These categories can produce significantly different legal analyses.

12. Potential Evidence

Investigators examining suspected algorithmic customer allocation may therefore focus on the combination of human and technical evidence rather than treating an algorithm itself as proof of collusion.

Potentially relevant material includes communications among competitors, software specifications, instructions to developers, customer-routing rules, API configurations, shared datasets, bidding histories, account restrictions, audit logs and evidence showing whether businesses intentionally stopped pursuing customers allocated to rivals.

A repeated pattern can be economically important, but competition authorities still have to satisfy the applicable legal requirements for establishing prohibited coordination.

13. Competitive Harm

Algorithmic customer allocation can produce several traditional cartel harms.

Customers may face fewer meaningful suppliers, weaker price competition, reduced incentives to improve quality, fewer promotions and reduced innovation.

Automation can potentially make an allocation arrangement more stable because software can process large amounts of market information and detect deviations faster than manual monitoring.

That does not create a completely new competition-law offense. Rather, algorithms can change the speed, scale, transparency and enforceability of conduct already familiar to competition law.

14. Compliance Implications

Businesses using pricing, bidding, recommendation or customer-routing algorithms should maintain independent competitive decision-making.

Particular attention is warranted where competitors share commercially sensitive customer-level information, jointly establish customer-assignment rules, allow common software to determine which customers each competitor should pursue, or create mechanisms discouraging participants from competing for another firm's customers.

Independent software use and legitimate vertical distribution arrangements should not be automatically characterized as cartel conduct; the contractual structure, evidence of agreement, economic context and applicable jurisdiction matter.

Conclusion

Algorithmic customer allocation represents the intersection of a traditional competition-law prohibition and modern automated decision-making.

Cases such as United States v. Topco Associates, United States v. Suntar Roofing, United States v. Cooperative Theatres, United States v. Cadillac Overall Supply, Hammes v. AAMCO, D&G/SuperValu, and United States v. Kemp & Associates establish the underlying principles governing horizontal customer and market allocation. European cartel enforcement similarly demonstrates that allocation down to individual customers can constitute serious anticompetitive coordination.

The central point is therefore that an algorithm is not a legal shield. If independent competitors enter an unlawful customer-allocation agreement, implementing or monitoring that arrangement through AI, automated routing, shared software or another digital system does not inherently change its competitive character. At the same time, algorithmic similarity or parallel conduct alone should not automatically be treated as proof of an agreement; the legally required coordination must still be established under the relevant competition regime.