Algorithmic Pricing and UK Competition Reform 2026: What Businesses Must Do Now

Executive Summary, What Businesses Must Do Now

Algorithmic pricing in the UK has moved from a policy discussion topic to an active enforcement priority. The Competition and Markets Authority (CMA) has signalled, through its March 2026 blog on AI and collusion, that pricing algorithms creating risks of coordinated outcomes are firmly within its crosshairs. The 2026 competition reform package gives the regulator enhanced information-gathering and interim-measure powers that apply directly to algorithm-driven conduct. For in-house counsel, compliance officers and pricing teams, the compliance window is narrowing fast.

Three immediate actions every business using pricing algorithms should take now:

1. Run an algorithmic risk assessment. Map every pricing algorithm in use, whether built in-house or supplied by a third-party vendor, against the red/amber/green criteria set out later in this guide. Document the results and preserve the underlying evidence.
2. Update vendor contracts and internal governance. Insert competition-law compliance warranties, audit rights, algorithm-change notification obligations and rollback clauses into all pricing-software agreements. Establish a human-in-the-loop escalation protocol for pricing decisions.
3. Prepare an investigation-readiness plan. Assemble a response team, implement a legal-hold protocol for algorithm logs, and prepare template responses for CMA information requests. Businesses that can demonstrate proactive compliance will be in a far stronger position if the regulator comes calling.

What Changed in 2026, Competition Reform Bill and CMA Algorithm Powers

The 2026 competition reform package represents the most significant expansion of CMA algorithm powers since the Competition Act 1998. As of 2 June 2026, the reforms strengthen the CMA’s toolkit in three critical areas relevant to algorithmic pricing in the UK.

First, the CMA now has broadened information-gathering powers that explicitly cover algorithm design documents, training data sets, pricing logic and decision logs. Where previously the authority might have requested pricing records, it can now compel the production of the underlying algorithmic architecture, including source code and model parameters, under enforceable notice.

Second, the reforms introduce enhanced interim measures. The CMA can impose temporary restrictions on specific algorithmic conduct while an investigation is ongoing, including ordering businesses to disable particular repricing features or revert to manual pricing in defined product categories. Industry observers expect these interim powers to be used early in investigations to prevent further consumer harm while evidence is gathered.

Third, the digital markets competition framework, operating alongside the competition reform package, creates additional obligations for platforms designated with strategic market status. Platform pricing regulation under this framework can require transparency about algorithmic ranking and pricing mechanisms, creating a dual-track compliance obligation for marketplace operators.

The CMA’s March 2026 blog post on AI and collusion confirmed that algorithmic pricing creates risks related to dynamic pricing, where prices move rapidly in response to changing market conditions. The regulator has made clear that the use of an algorithm does not provide a shield against liability, the same competition law principles that prohibit human price-fixing apply with equal force to automated systems.

Quick Timeline of Key Developments

• November 2024. CMA publishes blog post defining pricing algorithms and outlining competition law risks for the first time in practitioner-accessible terms.
• October 2025. Multiple law firm analyses confirm that algorithmic pricing has emerged as an enforcement priority for both the European Commission and the CMA.
• March 2026. CMA publishes “AI and collusion: frontiers, opportunities and challenges” blog, explicitly flagging algorithmic pricing risks and enforcement intentions.
• Q2 2026. Competition reform package provisions on enhanced information powers and interim measures take effect.

How Pricing Algorithms Work, Technical Primer for Counsel

Before assessing legal risk, in-house teams need a working understanding of how pricing algorithms operate. The CMA defines a pricing algorithm as a system that sets prices or recommends prices to be set, usually based on current and past data about market conditions. In practice, these systems range from simple rule-based tools to sophisticated machine-learning models.

Common Algorithm Types and Where Legal Risk Arises

• Rule-based repricing. The simplest form: “if competitor X lowers price by 5%, match within 2 hours.” These algorithms execute pre-programmed instructions without learning. Legal risk is moderate but increases sharply when rules are calibrated to track and match specific competitors, potentially facilitating tacit coordination.
• Machine-learning dynamic repricing. These algorithms ingest historical sales data, competitor prices, demand signals and inventory levels to optimise pricing in real time. Legal risk is elevated because the algorithm may independently converge on supra-competitive price levels without explicit human instruction, a phenomenon the GOV.UK economic report on pricing algorithms identifies as a key concern.
• Reinforcement-learning models. The most advanced category: algorithms that learn pricing strategies through trial-and-error interactions with the market environment. These systems can develop cooperative pricing strategies without being explicitly programmed to collude, making them the highest-risk category from a competition law perspective.

From a vendor perspective, most businesses acquire pricing algorithms through one of three models: licensed SaaS platforms (e.g., repricing tools for e-commerce), API-integrated pricing engines embedded within enterprise systems, or bespoke algorithms developed by in-house data science teams. Each model creates different liability exposure and different compliance obligations. Vendor-supplied systems demand contractual protections; in-house systems require documented design principles and audit trails.

Practical next step: Catalogue every pricing algorithm in your technology stack, classify it by type, and identify the data sources it consumes, particularly whether it ingests competitor pricing data.

When Algorithmic Pricing Becomes Illegal, Collusion, Signalling and Facilitation

Not all algorithmic pricing creates competition law risk. When used lawfully, pricing algorithms can lead to greater competition, reduced cost and reduced barriers to market entry. The legal line is crossed when algorithms become the mechanism through which competitors coordinate pricing, whether deliberately or through design choices that produce the same effect.

Under UK competition law, the key prohibition is contained in the Chapter I prohibition of the Competition Act 1998, which mirrors Article 101 TFEU. This prohibits agreements and concerted practices that have the object or effect of preventing, restricting or distorting competition. The critical question for algorithmic pricing in the UK is whether algorithm-driven price convergence constitutes a concerted practice or merely reflects lawful parallel conduct.

Red-Flag Behaviours and the Collusion Spectrum

The CMA and leading competition practitioners have identified several categories of algorithmic conduct that present escalating levels of risk:

Price Gouging and Related Offences

A common question is whether price gouging is illegal in the UK. Unlike some US jurisdictions, the UK does not have a specific price-gouging statute. However, excessive or exploitative pricing by a dominant undertaking can constitute an abuse of a dominant position under the Chapter II prohibition. Algorithmic pricing that drives prices to exploitative levels could therefore attract enforcement action where the business holds a dominant market position, even absent any element of coordination with competitors.

Practical next step: Review every pricing algorithm for the five conduct categories above. Any algorithm falling into the “High” or “Very high” risk categories requires immediate remediation or legal review.

Algorithmic Pricing Risk Assessment, Template and Scoring

An algorithmic risk assessment is the single most important compliance step a business can take. It creates a documented record of proactive compliance, provides the foundation for remediation, and demonstrates good faith to the CMA in the event of an investigation. The following template provides a structured framework that in-house counsel can adapt to their organisation.

Risk Assessment Criteria, Red/Amber/Green Scoring

Step-by-Step: Conducting the Assessment

1. Inventory. Identify every pricing algorithm in use across the business, including tools embedded in marketplace seller accounts, ERP pricing modules and standalone SaaS subscriptions.
2. Data mapping. For each algorithm, document every data source it consumes. Pay particular attention to competitor pricing feeds, third-party data aggregators and any data shared with or received from the vendor.
3. Stakeholder interviews. Interview the product managers, data scientists and commercial leads who configured or manage each algorithm. Document the business rationale, the pricing parameters set, and any manual overrides applied.
4. Scoring. Apply the red/amber/green criteria above to each algorithm. Any algorithm scoring “Red” on two or more factors should be flagged for immediate legal review.
5. Evidence preservation. Preserve all configuration files, pricing logs, vendor contracts, data-sharing agreements and internal communications relating to algorithm design and operation. These documents will be critical in any CMA investigation.
6. Remediation plan. For each red-flagged algorithm, develop a remediation plan with specific actions, timelines and responsible owners. Prioritise disabling features that create the highest collusion risk.

Practical next step: Complete the algorithm compliance checklist within 30 days. Schedule the first review cycle for 90 days after initial completion, and annually thereafter.

Governance, Contracts and Vendor Management, What to Change Now

Compliance with algorithmic pricing rules requires changes to both internal governance structures and external vendor relationships. The competition reform package makes it clear that businesses cannot outsource compliance responsibility to technology providers.

Recommended Contract Clauses for Pricing Algorithm Vendors

Every agreement with a pricing-algorithm vendor should include the following provisions:

• Competition compliance warranty. The vendor warrants that the algorithm does not facilitate, and is not designed to facilitate, any agreement, concerted practice or exchange of competitively sensitive information between the licensee and any other client of the vendor.
• Data segregation obligation. The vendor undertakes to segregate the licensee’s data from that of all other clients, and confirms that no aggregated, anonymised or benchmarked data derived from other clients operating in the same market is used to train or calibrate the algorithm.
• Audit rights. The licensee has the right to audit the algorithm’s design, data inputs and decision logic, including through independent third-party review, upon reasonable notice and at least annually.
• Change notification and rollback. The vendor must notify the licensee in writing before implementing any material change to the algorithm’s pricing logic, data sources or optimisation objectives. The licensee has the right to require rollback of any change that creates competition-law risk.
• Investigation cooperation. The vendor must cooperate fully with any competition authority investigation, including providing access to source code, training data and decision logs on request.

Internal Governance, AI Policy and Human-in-the-Loop Controls

Beyond vendor management, businesses should establish an internal governance framework that includes:

• Pricing algorithm policy. A board-approved policy that defines permissible algorithm configurations, prohibited features (such as competitor-price matching or retaliation logic), and escalation procedures for anomalous pricing outcomes.
• Human-in-the-loop requirement. A defined threshold above which all algorithmic pricing decisions must be reviewed by a human decision-maker before execution. Industry observers expect the CMA to view businesses with robust human oversight more favourably in enforcement contexts.
• Competition compliance training. Mandatory training for all staff involved in pricing algorithm design, configuration or management, covering the legal risks of algorithmic collusion and the red-flag indicators identified above.
• Board reporting. Regular reporting to the board or audit committee on algorithmic pricing risk, including the results of risk assessments and the status of remediation actions.

Practical next step: Review all existing vendor contracts against the clause list above within 60 days. Initiate renegotiation for any contract that lacks data-segregation or audit provisions.

Investigation Readiness, If the CMA Opens a Probe

The CMA’s enhanced powers under the 2026 competition reform package mean investigations can move quickly. Businesses that have not prepared will find themselves scrambling to assemble evidence, preserve data and coordinate responses under intense time pressure. The following investigation readiness checklist should be completed now, before any investigation materialises.

Ten-Point Investigation Readiness Checklist

1. Appoint a response team. Identify the individuals, legal, IT, data science, commercial, who will form the investigation response team. Ensure each member understands their role and has current contact details on file.
2. Implement a legal hold protocol. Prepare a template legal hold notice that can be issued within 24 hours of receiving a CMA enquiry. The hold must cover all algorithm logs, pricing data, configuration files, vendor communications and relevant internal messages.
3. Preserve algorithm decision logs. Ensure that all pricing algorithm decision logs are retained for a minimum of six years from the date of each pricing decision. Configure logging to capture the data inputs, decision logic and output for every price change.
4. Suspend auto-learn features. Prepare a protocol to suspend reinforcement-learning or auto-calibration features immediately upon receiving a CMA information request, to prevent the algorithm from altering its behaviour during the investigation.
5. Notify the board. Establish a board notification protocol that triggers within 48 hours of any formal or informal CMA contact regarding algorithmic pricing.
6. Map document categories. Pre-identify the categories of documents the CMA is likely to request: algorithm design documents, vendor contracts, pricing parameter settings, competitive intelligence reports, internal communications about pricing strategy, and board papers discussing pricing approaches.
7. Prepare initial response templates. Draft template responses for common CMA information requests, including an initial holding response acknowledging receipt and confirming the timeline for substantive compliance.
8. Engage external counsel. Identify and brief competition law counsel in advance so they can be engaged immediately if needed. Advance engagement avoids the delay of a cold-start briefing during a live investigation.
9. Privilege review. Establish a process for privilege review of algorithm-related documents, including legal advice on algorithm design and compliance assessments. Ensure privileged documents are clearly marked and segregated.
10. Test the plan. Run a tabletop exercise simulating a CMA dawn raid or section 26 notice. Identify gaps in the response plan and address them before they become real-world problems.

Typical CMA Investigation Timeline

Based on precedent and published CMA guidance, a typical algorithmic pricing investigation is likely to proceed through the following phases:

• Weeks 1–4: Initial information request (section 26 notice or voluntary request); legal hold; document preservation and collection.
• Months 2–6: Document production; CMA analysis of algorithm design and pricing data; possible site visits or interviews with algorithm designers and commercial staff.
• Months 6–12: CMA issues statement of objections (if infringement suspected); business prepares response; possible access to the CMA file.
• Months 12–24: Oral hearings; final decision; possible commitments or settlement discussions.

Practical next step: Complete the investigation readiness checklist and run a tabletop exercise within 90 days. Store the completed checklist alongside the risk assessment as evidence of proactive compliance.

Remediation and Safe Harbour Options, Practical Remediation Playbook

Where a risk assessment identifies high-risk algorithmic pricing conduct, businesses face a strategic decision: remediate proactively and consider voluntary notification, or adopt a defensive posture and prepare for potential enforcement.

When to Self-Report Versus Defend

Early indications suggest that the CMA will look more favourably on businesses that identify and correct algorithmic pricing risks before enforcement action begins. Priority remediation steps include:

• Disable high-risk features immediately. Any algorithm feature that enables automated retaliation, competitor-price matching or signalling should be disabled pending full legal review.
• Introduce human-in-the-loop controls. For algorithms scoring “Red” in the risk assessment, implement mandatory human review of all pricing decisions until the algorithm has been redesigned.
• Switch vendors or renegotiate. Where a vendor cannot demonstrate data segregation or provide adequate audit access, migrate to a compliant alternative or renegotiate the contract with enforceable compliance provisions.
• Establish ongoing monitoring. Implement a quarterly review cycle for all pricing algorithms, with documented sign-off by the compliance function and regular reporting to the board.

Voluntary notification to the CMA should be considered where the risk assessment reveals conduct that may already constitute an infringement. Self-reporting may qualify the business for leniency, including significant fine reductions, but the decision to self-report requires careful legal analysis of the specific facts. The likely practical effect of early engagement with the CMA is a more cooperative investigation process and reduced penalties.

Obligations by Entity Type, Comparison Table

Different types of businesses face different levels of enforcement risk and compliance obligation under the algorithmic pricing rules. The following table summarises the key distinctions:

Preparing for What Comes Next

Algorithmic pricing in the UK is no longer a theoretical concern, it is an active enforcement priority. The CMA has the tools, the mandate and the stated intention to pursue businesses whose pricing algorithms produce anti-competitive outcomes. The compliance window for remediation is open now, but it will not remain open indefinitely. Businesses that complete their algorithmic risk assessment, update their vendor contracts and governance frameworks, and prepare an investigation-readiness plan will be in the strongest possible position, whether the CMA comes knocking or not. Those that wait risk facing enhanced penalties, intrusive interim measures and the reputational damage that accompanies a public enforcement action in a rapidly evolving area of competition law.

Sources

1. CMA Blog, AI and Collusion: Frontiers, Opportunities and Challenges
2. CMA Blog, Pricing Algorithms and Competition Law: What You Need to Know
3. GOV.UK, Pricing Algorithms Economic Report
4. Akin Gump, CMA Highlights Real World Risks of Algorithmic Pricing
5. Paul, Weiss, Enforcement on Algorithmic Pricing Starts to Bite in Europe
6. Morgan Lewis, Algorithmic Pricing Emerges as Enforcement Priority for EU and UK Antitrust Regulators
7. Shepherd and Wedderburn, Pricing Algorithms and Competition Law in the UK and Beyond