Summary
- Criteo's best case is not that old tracking returns. It is that retailer data, commerce audiences, closed-loop measurement and AI bidding can make advertisers accept an independent intermediary between their budget and measurable sales.
- The hard test is margin durability. Criteo's 2025 consolidated revenue was $1.945 billion, but $770 million went to traffic acquisition costs, and Q1 2026 exposed how two retail-media client scope changes could erase $27 million of contribution ex-TAC.
- The judgment is cautious: Criteo has a real commerce-data asset and a credible independent retail-media position, but it has not yet proved that customer concentration, platform rules and traffic costs will leave it with lasting pricing power.
Advertisers Buy Sales Evidence, Not Screen Time
The economic incentive starts with a practical change in how marketing buyers defend budgets. A retailer, travel brand or consumer-electronics seller can buy reach almost anywhere. It can buy search from Google, marketplace placement from Amazon, social distribution from Meta, programmatic reach through independent demand platforms, or retail-media access directly from a merchant network. Criteo only deserves a margin in that chain if it proves that its data, bidding, measurement and supply access convert spend into revenue better than those alternatives.
That is why the company describes itself as a commerce intelligence platform rather than a simple retargeting vendor. The sales pitch is built around shopper intent, product-level signals and performance measurement. On its own website Criteo claims access to 2.5 billion users and more than $1 trillion in yearly sales data, alongside more than 17,000 global clients.
Those numbers are management claims, not independent proof of value, but they frame the economic thesis: Criteo wants to replace cross-site tracking privilege with a commerce graph that advertisers can use to reach likely buyers and that retailers can use to monetize digital shelves.
The downside is just as clear. Criteo is trying to sit between parties that increasingly want to own the data, the budget and the customer relationship themselves. Large retailers are building in-house retail-media networks. Google, Amazon and other platform operators control enormous surfaces where shoppers already search, watch, compare and buy. Browsers and mobile operating systems decide which identifiers are allowed. Regulators decide whether consent, transparency and joint-controller arrangements are enough.
A performance-ad company can be technically sophisticated and still lose economics if the upstream owners of data and inventory decide to keep more of the value.
The right question is therefore not whether Criteo can grow reported revenue in a strong advertising cycle. Revenue can expand while value creation weakens if traffic costs, client concessions, investment spending or retailer concentration rise faster than the contribution retained by the company. The right question is whether Criteo can convert commerce data into a defensible contribution ex-TAC margin after paying for media access, engineering privacy-compliant addressability, funding research, and absorbing the bargaining power of large customers.
Management's answer is that Criteo is moving from a narrow retargeting identity to a broader commerce media position. The market will pay for that answer only if the underlying economics show that the company is more than a rented link between advertisers, retailers and media owners.
The Entity Is A French Technical Footprint Inside A Listed Group
The assigned directory entity is Criteo Technology SAS, a French company name tied to Criteo's technical and network-resource footprint. The public audited financial statements, however, are for Criteo S.A. and its consolidated subsidiaries. That distinction matters. The public evidence does not support treating Criteo Technology SAS as a standalone telecom operator, a retail-media subsidiary with separately disclosed revenue, or an independent connectivity seller. It is better understood as part of the French-origin operating structure behind the listed Criteo group.
Criteo S.A. was incorporated in France, lists American Depositary Shares on Nasdaq under CRTO, and gives its principal executive office as 32 Rue Blanche in Paris in the 2025 Form 10-K and Q1 2026 Form 10-Q. The same filings report the consolidated business in two operating segments: Retail Media and Performance Media. Retail Media covers revenue from brands, media buyers and retailers for retail-media digital advertising inventory, audiences and services. Performance Media covers targeting, supply and ad-tech services. Those are advertising-technology segments, not telecom-service segments.
This boundary prevents two common errors. First, RIPE membership should not be inflated into proof that Criteo sells connectivity. A company can hold number resources, operate internet-facing services, run data systems, peer or manage infrastructure without being an ISP to the public. Second, consolidated Criteo S.A. results should not be presented as if they are the statutory accounts of Criteo Technology SAS. The available public facts support an economic analysis of Criteo's group-level commerce media model, with the French technology entity treated as a relevant local and network-resource anchor.
That boundary also clarifies why Criteo is relevant to a telecom economics lens. The importance is not that the company competes with carriers. It is that Criteo depends on internet infrastructure, cloud and data locality choices, IP-number governance, browser policy, API access, retailer data flows and media-owner supply. Digital advertising is not usually called network economics, but its margin is shaped by who controls identifiers, traffic, data exchange and delivery surfaces. Criteo's cost base and strategy sit directly on that control surface.
The company is therefore best judged as a digital-service intermediary. It sells outcomes in commerce advertising, relies on data and infrastructure to deliver those outcomes, and must keep enough independence from the largest platforms to justify its role. The French technical footprint is part of that operating reality, but it is not evidence of a public network-service business.
The Model Is A Take-Rate Business With Two Different Segments
Criteo's consolidated 2025 numbers show a business that is profitable, material and still exposed to the quality of its retained spread. Revenue was $1.945 billion in 2025, up only slightly from $1.933 billion in 2024 and below the $1.949 billion reported in 2023. Gross profit was healthier, rising to $1.049 billion in 2025 from $983 million in 2024 and $863 million in 2023. Contribution ex-TAC, Criteo's key operating measure after deducting traffic acquisition costs, reached $1.175 billion in 2025, up 5% reported and 3% at constant currency.
That pattern is important. Reported revenue is not the cleanest guide to Criteo's economics because a large part of revenue is offset by traffic acquisition costs. Contribution ex-TAC is closer to what Criteo keeps before the rest of its operating expenses. Management uses it to allocate resources and evaluate segment performance. The measure is not a substitute for GAAP income, and it is not perfectly comparable with other advertising companies, but it is the right starting point for judging whether the company is retaining value after buying or accessing traffic.
The segment split is asymmetric. In 2025, Performance Media generated $1.681 billion of revenue and $914.9 million of contribution ex-TAC. Retail Media generated $263.9 million of revenue and $259.7 million of contribution ex-TAC. Almost all reported traffic acquisition cost sat in Performance Media: $766.1 million in 2025, compared with only $4.2 million in Retail Media. Retail Media therefore looks much cleaner on a contribution ex-TAC basis, but it is also smaller and more exposed to the bargaining power of a limited number of large retail partners.
That is the central trade-off. Performance Media is the larger engine and still pays heavily for traffic. Retail Media is strategically attractive because retailer data and ad inventory can sit closer to the point of purchase, but Criteo does not fully control the retailers that supply the inventory and measurement. The margin quality of the group improves if Retail Media can scale broadly across many partners without one or two large customers resetting scope or pricing. It weakens if retail media is a high-expectation segment controlled by a few large counterparties.
Q1 2026 made that risk visible. Revenue fell 6% year over year to $425 million, contribution ex-TAC fell 5% to $250 million, and adjusted EBITDA fell 30% to $65 million. Criteo said Retail Media contribution ex-TAC fell 32% at constant currency because of previously communicated scope changes with two specific Retail Media clients. Excluding that $27 million headwind, the underlying retail partner base grew contribution ex-TAC 24%. Both facts matter. There is growth under the surface, but the surface can be moved by a small number of powerful clients.
Traffic Costs Still Decide How Much Revenue Becomes Economics
Traffic acquisition cost is the economic toll Criteo pays to access advertising inventory and reach. In 2025, TAC was $770.3 million. It was lower than $811.8 million in 2024 and $926.8 million in 2023, which helped contribution ex-TAC improve even though reported revenue barely moved over the three-year period. That trend is constructive, but it does not eliminate the risk. A business that pays hundreds of millions of dollars each year for traffic is still sharing economics with publishers, exchanges, media owners and other supply holders.
This is where revenue growth and value creation separate. If Criteo wins more spend only by paying more for inventory or accepting lower retained spread, shareholders get volume without pricing power. If Criteo uses commerce data and prediction to buy media more efficiently, then even flat revenue can produce better contribution and operating income. The 2025 result leans toward the second interpretation, but Q1 2026 shows how fragile the story can become when a few large retail-media arrangements change.
The company also reports media spend, defined as working media spend allocated to Retail Media campaigns and media spend activated on behalf of Performance Media clients. In Q1 2026, media spend surpassed $1.0 billion and was up 8% year over year at constant currency. For the last twelve months, it was $4.4 billion. That is the scale of budget Criteo is helping move. But the conversion of media spend into contribution ex-TAC is not a simple published take rate. The measure mixes Retail Media and Performance Media activity, and it does not disclose enough about pricing by product, channel or customer type to calculate true unit economics.
The practical question is whether advertisers see Criteo's retained margin as a cost worth paying. An advertiser can buy performance media directly from platforms with enormous first-party data. A retailer can run its own sponsored-products or display media program. A publisher can connect to several demand sources. Criteo has to prove that its bidding, commerce audiences, identity matching, creative optimization and measurement raise the advertiser's return enough to justify the fee.
Traffic costs also determine operating flexibility. In a downturn, advertisers can reduce spending quickly. Criteo's media access costs may fall with activity, but its research, engineering, sales and compliance costs do not reset as fast. That makes contribution ex-TAC the pressure gauge. If it remains stable while revenue moves, Criteo can defend the model. If it falls because traffic costs or client concessions rise, the business becomes an expensive services layer between stronger platforms.
Retail Media Gives Better Data But Less Control
Retail Media is Criteo's clearest answer to the decline of third-party tracking. The logic is strong. Retailers know what shoppers browse, put in baskets and buy. Brands want to reach shoppers near the digital shelf. Closed-loop measurement can show whether an ad drove an online or in-store sale. If Criteo can connect brands, media buyers and retailers across many networks, it can offer an alternative to buying only from Amazon or from one retailer at a time.
Criteo's own positioning reflects that ambition. The company markets Retail Media as a way to reach and convert shoppers near the point of sale, Commerce Max as a demand-side access point for retail media beyond a single retailer, Commerce Yield as a monetization system for retailers and marketplaces, and Commerce Grid as a commerce supply-side platform for media owners. On the home page it says it provides access to more than 200 retailers and premium media owners and demand from thousands of brands and hundreds of media buyers.
The numbers are company claims, but they show the intended market structure: Criteo wants to be a neutral commerce-media layer.
The problem is control. Retailers own the shopper relationship. They decide which data can be shared, which measurement claims can be made, how much inventory is available, and how much margin to retain. The largest retailers can build or buy their own systems. Smaller retailers may value Criteo's technology and demand aggregation, but they may also lack the scale to offset losses from one major partner. That is why the Q1 2026 $27 million Retail Media headwind from two clients is more important than it looks. It is not just a quarterly variance.
It is evidence that contract scope can alter the economics of the segment Criteo presents as its strategic growth engine.
That contract structure also explains why the near-zero Retail Media TAC line should be read carefully. In Performance Media, the toll is visible as traffic acquisition cost. In Retail Media, the toll can be embedded in commercial scope: which retailers permit off-site activation, which categories are available, which measurement windows count, whether in-store sales are attributed, who owns campaign service, and how revenue is split among the merchant, Criteo and demand-side buyers. Public filings do not disclose those terms by customer.
The Q1 2026 disclosure therefore matters because it turns an abstract concentration warning into a dollar test: two client scope changes were large enough to reduce Retail Media contribution ex-TAC by $27 million in one quarter, even while the rest of the partner base was growing. That is downside allocation in practice. Criteo carries the cost of product investment and advertiser demand aggregation, but a retailer that controls shopper data and inventory can still change the addressable surface.
The more Criteo sells itself as a technology layer for retailers, the more its economics depend on contract renewal, module adoption and retained rights to measure sales.
Amazon is the strategic reference point. Criteo's own annual report states that about 75% of U.S. retail-media ad spend is concentrated on Amazon, while Amazon accounts for roughly one third of ecommerce gross merchandise value. That gap creates an independent retail-media opportunity: brands want commerce data outside Amazon because shoppers buy across many merchants. But it also shows why Amazon remains the default competitor. Amazon owns demand, data, inventory, measurement and merchant relationships at a scale no independent intermediary can easily match.
Criteo's opportunity is therefore in fragmentation. It can help retailers that do not want to build everything themselves, and it can help brands avoid managing hundreds of separate retail-media buying points. Its risk is the same fragmentation. Every integration, retailer rule, data-sharing boundary and measurement standard adds operating complexity. Retail Media becomes attractive if Criteo can standardize that complexity and keep a small but durable share of many transactions. It becomes less attractive if the largest retailers in-house the margin and smaller ones cannot produce enough volume.
Identity Infrastructure Is The Operating Surface
Criteo's old advantage depended heavily on recognizing users across sites and serving relevant ads when they returned to the open web. That world has changed. Safari, Firefox and privacy-focused browsers have long restricted third-party cookies. Chrome's direction has shifted repeatedly, moving from planned deprecation to user-choice and then a narrower privacy roadmap. Mobile platforms have also restricted tracking through consent frameworks and device-identifier limits.
The exact rules can change, but the economic trend is stable: identity is moving from broadly available web tracking toward permissioned, first-party and platform-controlled signals.
Criteo's response is a multi-part addressability strategy. Its product pages emphasize first-party data onboarding, hashed email matching, contextual signals, commerce audiences, retailer data, predictive bidding and privacy-adapted AI. The Shopper Graph page describes a system that connects shopper IDs and commerce data for campaign optimization. Commerce Growth says targeting can use behavioral, contextual and commerce data without relying on third-party identifiers. Those claims matter because the product must continue to work when old identifiers are missing or restricted.
This is not merely a privacy-compliance story. It is a bargaining-power story. Whoever controls identity controls who can measure, attribute and optimize. Retailers control logged-in shopper data. Platforms control user accounts and closed measurement environments. Browsers control client-side storage and cross-site signals. Publishers control logged-in audiences and consent interfaces. Criteo's role is to stitch together enough permissioned signals to make independent performance advertising work without owning the whole surface.
That is a hard engineering and commercial task. Hashed email and first-party data can be valuable, but they depend on scale, consent quality, matching rates and customer willingness to share. Contextual targeting is privacy-resilient, but usually weaker for conversion prediction than deterministic commerce history. Retailer data is close to purchase, but retailer access is negotiated. AI bidding can improve decisions, but only if the input signals remain available, clean and representative.
The company has one credible asset: long experience with commerce prediction and large volumes of shopping signals. Its 2025 research release around CriteoPrivateAd, a public anonymized bidding dataset for private advertising research, shows that Criteo has production data and technical interest in privacy-constrained ad systems. That does not prove a moat, but it supports the view that Criteo is not simply hoping old cookies survive. It is investing in ways to model, bid and measure when data is delayed, aggregated or less granular.
The investor question is whether this identity infrastructure produces pricing power or merely keeps Criteo in the game. If it allows advertisers to measure incremental sales across fragmented commerce surfaces, it is valuable. If it becomes a maintenance cost required to offset privacy restrictions, the benefit may accrue more to retailers and platforms than to Criteo.
The Network Evidence Is Real But Narrow
The RIPE NCC member evidence for Criteo Technology SAS is useful because it confirms a real network-resource and internet-governance footprint. It should be interpreted narrowly. RIPE membership records the presence of a member in the regional internet registry ecosystem; it is not proof that the company sells ISP service, IP transit, hosting, cloud infrastructure or managed network services to customers. For Criteo, the evidence fits an ad-tech company with digital infrastructure needs, not a carrier.
That narrow reading is still economically relevant. Performance advertising requires low-latency bid decisioning, data ingestion, creative delivery, attribution, fraud controls, consent handling, API integrations and reporting. Even when much of the physical infrastructure is rented from cloud or data-center suppliers, the company must operate internet-facing systems reliably across regions. If latency rises, data flows are restricted, identifiers are blocked, or delivery systems fail, the business impact can be immediate because advertisers buy outcomes in near real time.
The 2025 annual report names systems and infrastructure failure as a risk. It also warns that Criteo's ability to generate revenue depends on collecting significant amounts of data from multiple sources and that this access may be restricted by consumer choice, clients, publishers, retailers, browsers, software changes, laws, regulations and industry standards. Those risks describe the practical infrastructure layer of the business. It is a networked data company even though it is not a telecom provider.
Cloud dependency also matters. Criteo does not disclose a simple supplier concentration metric for cloud or data infrastructure in the public filings used here. That absence is itself part of the uncertainty. The investor can see research spending, property and equipment, cash, receivables, payables and segment contribution, but not enough to separate owned systems, leased infrastructure, public-cloud commitments, third-party data costs and partner API dependence. A company can be profitable while its future margin is exposed to supplier price changes or technical access terms that are hard to see from outside.
Data locality and sovereignty are also part of the control surface. Criteo operates across regions, handles personal data and commerce signals, and is headquartered in Europe. It must satisfy GDPR and local consent rules while serving global advertisers and retailers. The French network-resource footprint is therefore not just administrative trivia. It is one piece of a broader operating reality in which data, infrastructure, jurisdiction and privacy constraints shape the cost of selling digital advertising.
The important judgment is restraint. The network evidence confirms digital-service seriousness. It does not create a telecom revenue thesis. The economic story remains advertising technology, with network resources as supporting evidence of operating capacity.
Research Spending And Working Capital Are The Price Of Relevance
Criteo is not a capital-heavy infrastructure owner in the way a fiber operator or data-center landlord is, but it is not a light marketing agency either. The 2025 consolidated income statement shows research and development expense of $283.3 million, or 15% of revenue. Sales and operations expense was $394.4 million, or 20% of revenue. General and administrative expense was $168.9 million. Together, those operating expenses consumed $846.6 million against gross profit of $1.049 billion. The company produced $202.8 million of operating income, but the margin depends on keeping a large technical and commercial organization productive.
That spending is not optional if Criteo wants to remain relevant. Privacy changes, AI bidding, retail-media integrations, creative automation, identity matching, fraud controls, reporting and APIs all require continuing engineering work. The company is trying to sell performance in a market where platforms with far larger budgets also invest heavily in ad systems. If Criteo cuts research too hard, it risks becoming a legacy retargeting vendor. If it spends aggressively without clear contribution growth, it protects the product while diluting shareholder returns.
Q1 2026 showed that tension. Operating expenses increased 12% year over year to $212 million, mostly because of planned growth investments, while contribution ex-TAC fell 5%. Adjusted EBITDA margin fell to 26% of contribution ex-TAC from 35% a year earlier. Management still guided full-year adjusted EBITDA margin to approximately 32% to 34% of contribution ex-TAC, but the quarter shows the near-term cost of investing through a softer revenue environment.
Working capital is the other hidden constraint. The Q1 2026 balance sheet reported $448.3 million of trade receivables and $448.5 million of trade payables, both down from much higher year-end balances after the seasonal fourth quarter. In an advertising business that handles large media budgets, receivables and payables can be large relative to quarterly profit. Criteo must collect from advertisers and brands while paying publishers, retailers and other supply partners. That timing risk is manageable in normal conditions, but it becomes more important if customers slow payments, large clients change scope, or advertising demand weakens.
The balance sheet gives Criteo room to operate. At March 31, 2026, it reported $371 million of cash and marketable securities and about $889 million of total financial liquidity, including available revolving credit capacity and treasury shares reserved for acquisitions. It also deployed $31 million for share repurchases in Q1 2026. That liquidity supports investment and capital returns, but it does not answer the strategic question. Share repurchases create value only if the core business can produce durable free cash flow after funding research and absorbing working-capital swings.
The cost-base test is therefore simple: each dollar of research, sales and operations spending must either increase contribution ex-TAC, defend retention, lower traffic costs, or open new surfaces where Criteo can retain margin. Strategy without resource allocation is marketing. Resource allocation without visible returns is just cost.
Competition Comes From Retailers, Walled Gardens And Independent DSPs
Criteo competes against three different substitute models. The first is the platform bundle. Google, Amazon, Meta and other large platforms offer demand, logged-in audiences, inventory, measurement and buying tools inside environments they control. They can make buying easy, absorb privacy changes internally, and offer scale that most advertisers cannot ignore. Amazon is especially important in retail media because it combines search intent, marketplace transaction data and sponsored-products inventory. Criteo can argue that brands need the open internet and non-Amazon retailers, but it cannot pretend Amazon is just another channel.
The second substitute is retailer self-supply. Large retailers can build their own media networks, hire ad-tech teams, partner with cloud providers, and sell directly to brands. They may still use Criteo for parts of the stack, but they have incentives to own data, pricing and measurement. If retail media becomes a core profit center, more retailers will ask why an external intermediary should keep a meaningful share. Criteo's answer has to be demand aggregation, technology speed, cross-retailer standardization and better performance than a retailer can deliver alone.
The third substitute is independent programmatic infrastructure. Demand-side platforms, supply-side platforms, clean-room providers, identity vendors and measurement companies all claim pieces of the same budget. Some focus on agencies and media buyers. Others focus on publishers. Others promise privacy-safe collaboration between brands and retailers. Criteo's advantage is that it combines commerce data, bidding, retail-media access and performance heritage. Its disadvantage is that each piece of that bundle has specialized competitors.
Competition also changes the customer conversation. If Criteo sells only as a managed service, advertisers may pressure fees and compare it with agencies or platform tools. If it sells self-service software, it needs ease of use and automation. If it sells retail-media infrastructure to retailers, it faces procurement scrutiny and in-house build comparisons. If it sells supply monetization to media owners, it competes for yield against many demand sources. A broad platform story is attractive, but only if the pieces reinforce one another.
This is why management's emphasis on Criteo GO, self-service performance advertising and AI-driven onboarding is economically meaningful. Self-service can lower sales friction and widen the customer base beyond large enterprise accounts. But self-service also exposes the product to direct comparison with simpler ad-buying systems. Small and midsize advertisers may value speed, yet they churn quickly if performance disappoints.
The company therefore needs a narrow definition of winning. It does not need to beat Google or Amazon everywhere. It needs to be the best independent way for brands to turn commerce signals outside the largest closed ecosystems into measurable sales. That is a plausible niche, but it must be defended with data access, product quality and partner breadth, not rhetoric.
Privacy Risk Is Economic, Not Just Legal
Privacy risk for Criteo is often described as legal exposure, and that is part of the story. In 2023, the French data protection authority CNIL fined Criteo 40 million euros over personalized advertising and consent-related GDPR failures. For an ad-tech company whose business depends on user-level or pseudonymous data, that kind of sanction is not a side issue. It affects trust with retailers, publishers, advertisers and regulators. It also raises the cost of compliance, documentation and partner assurance.
But the larger privacy issue is economic. Consent rules decide which data can be used. Browser settings decide whether identifiers survive. Retailers and publishers decide how much data they are willing to share. Consumers decide whether to opt out. Each restriction can reduce match rates, weaken attribution, delay reporting, or push advertisers toward platforms with more logged-in data. Even if Criteo follows the rules, the rules can still reduce the value of its product.
The company acknowledges that risk directly in its filings. It warns that data access can be restricted by consumer choice, clients, publishers, retail partners, browsers, software changes, technology developments, law and industry standards. It also warns that regulatory, legislative or self-regulatory developments regarding internet or online matters could hurt the business. These are not generic boilerplate risks for Criteo. They are the operating conditions under which Criteo sells.
Google's changing approach to third-party cookies and privacy APIs illustrates the instability. When the largest browser owner can alter the future of tracking, attribution and ad targeting, independent ad-tech firms must spend years preparing for possible futures that may not arrive as expected. Chrome's shifts did not remove the pressure on Criteo; they showed that platform dependence is a permanent condition. Safari, Firefox, mobile operating systems and regional privacy regimes still constrain addressability. Retailers and publishers still control consented data. Advertisers still demand measurable results.
The strongest version of Criteo's strategy turns privacy pressure into a reason to buy from Criteo. If the company can offer privacy-adapted commerce audiences, retailer measurement and cross-channel performance without relying on prohibited identifiers, privacy becomes a product advantage. The weaker version turns privacy into a tax. Criteo spends more on compliance and engineering while platforms with direct user relationships keep the best signals.
The evidence is mixed. Criteo has real technical assets, public privacy research activity and a commerce-data strategy. It also has a record of regulatory sanction and continuing dependence on rules it does not control. The conclusion should not be that privacy will destroy Criteo. It should be that privacy is one of the main reasons the company's margin must be earned again each year.
What Would Change The Judgment
The current judgment is that Criteo is investable as an independent commerce-media intermediary, but not yet proven as a durable margin compounder. The company has scale, profitability, cash, research depth, a real retail-media opportunity and a credible response to identifier loss. It also has concentrated customer risk, large traffic costs, dependence on retailers and platforms, and limited public disclosure on unit economics by product, channel and customer cohort.
Several facts would improve the judgment. First, Retail Media contribution ex-TAC would need to grow after removing the two-client Q1 2026 scope headwind, not merely recover because comparisons get easier. Broad-based growth across many retailers would show that Criteo is not dependent on a few large counterparties. Second, Performance Media would need to keep improving contribution ex-TAC while controlling traffic acquisition costs, proving that the older business can fund the transition rather than drain it.
Third, management would need to show that self-service and AI-assisted campaign creation lower sales friction without reducing retention or pricing. A larger long-tail customer base is valuable only if acquisition cost, support cost and churn remain under control. Fourth, Criteo should disclose enough about media spend, contribution ex-TAC, customer concentration and product mix to let investors distinguish high-quality retained margin from volume bought through expensive traffic.
Fifth, regulatory and platform developments would need to validate Criteo's identity strategy. Higher consented match rates, stronger retailer data partnerships, privacy-safe measurement adoption, and less dependence on legacy identifiers would all strengthen the case. A new wave of browser restrictions, a major consent ruling, or a retailer decision to withdraw data access would weaken it.
The facts that would most damage the thesis are also clear. If more major retail-media clients reduce scope, Retail Media becomes a supplier-dependent services business rather than a scalable platform. If traffic acquisition costs rise faster than contribution ex-TAC, Performance Media loses operating leverage. If research spending keeps rising without reaccelerating growth, the AI and commerce-data story becomes a cost center. If advertisers shift incremental performance budgets back to Google, Amazon, Meta or in-house retail systems, Criteo's independence becomes a disadvantage.
Criteo's strategic choice is sensible. It is trying to move toward the part of advertising where dollars are closest to transactions and where retailers outside Amazon need monetization technology. The company also has enough cash and operating income to keep investing. But the market should not give full credit for the transition until the economics are visible. Commerce data can replace lost tracking privilege only if Criteo keeps access to the data, proves incremental sales, and retains enough margin after paying everyone who controls the inputs.
That is the position: Criteo is not a failing legacy tracker, and it is not yet a proven owner of a new commerce-media toll road. It is a competent intermediary in a market where the most valuable inputs are controlled by others. The next evidence should be judged less by slogans about AI and more by contribution ex-TAC growth, Retail Media diversification, traffic-cost discipline, free-cash-flow conversion and the company's ability to make privacy-resilient measurement pay.