The Ad-Tech Choice Hiding Inside Your Media Budget

Somewhere between a marketer approving a budget and a person seeing an ad, a machine makes a decision in a fraction of a second. It weighs the person, the page, the price and the odds of a useful result. The Trade Desk and Criteo both sell access to that machinery. Put their product pages side by side and the verbs begin to blur: plan, buy, target, optimize, measure. The distinction becomes sharper when you stop reading the verbs and follow the money.

The Trade Desk comes to the transaction as a demand-side platform, or DSP. Its clients, largely agencies and advertisers, use it to buy inventory across connected television, video, display, audio, native placements and digital out-of-home. The company says it earns a platform fee generally based on a percentage of total client spend, along with revenue tied to data and other services. It also stresses that it does not own the media it recommends. Its incentive story is simple: help the buyer make better decisions across the open internet and earn more as useful spending moves through the platform.

Criteo arrives with different muscle memory. It became familiar to consumers through dynamic retargeting: look at a pair of shoes, leave the store, see the shoes again. That system required a close reading of shopping intent, product catalogs and conversions. Criteo has since expanded into acquisition, retail media, supply-side tools and a self-service commerce DSP. Its pitch now rests on turning retailer data and product-level signals into advertising decisions, then connecting exposure back to sales.

The useful comparison matches platform incentives to campaign objectives.

Start with the invoice

In its 2025 annual filing, The Trade Desk said $13.4 billion in gross spend passed through its platform and that it recorded $2.90 billion in revenue. Gross spend includes inventory, data, value-added services and the platform fee; revenue is generally reported net of money paid to suppliers because the company acts as an agent. The numbers are a reminder that the DSP is a toll collector on a much larger river of media.

That model creates a buy-side alignment worth taking seriously, but not worshipping. A percentage-of-spend fee can align the platform with the buyer's desire to manage more media in one place. It can also reward more spending. Buyers still need independent incrementality tests, clear fee disclosure and guardrails around high-margin data or add-on products. Alignment is a design choice, not a substitute for procurement.

Criteo's economics require a different map. The company reports two operating segments: Performance Media and Retail Media. In performance campaigns, it helps marketers translate activity on websites and apps into audience and bidding signals. In retail media, it helps brands reach shoppers near a digital point of sale while helping retailers monetize their audiences and inventory. Criteo says brands typically pay a negotiated DSP fee, while retailers pay a negotiated supply-side or technology fee. It can sit near both ends of a commerce-media transaction.

That is a strength when the job is to connect a brand, a retailer, a product and a verified sale. It is also a reason to ask more questions about auction mechanics, take rates, inventory access and data rights. A connected system can remove handoffs and measurement loss. It can also make each internal toll harder to see unless the buyer insists on transparency.

The retail aisle became a data network

Criteo's retail relationships did not appear overnight. Its 2016 acquisition of HookLogic brought a network that connected brands with major ecommerce sites through sponsored product ads. HookLogic's partners included retailers and marketplaces such as Walmart, Target, Best Buy and Expedia at the time of the deal. That acquisition added the retailer-monetization half of the equation to Criteo's performance-advertising roots.

The current platform is broader. Criteo says its commerce data covered roughly $1 trillion in online sales during 2025 and mapped more than 5 billion stock-keeping units. Commerce Max lets brands and agencies buy retailer-site placements and open-internet media, build audiences from retailer first-party data and measure product-level sales. Performance Media continues to include dynamic retargeting, but also covers discovery and acquisition. Calling Criteo only a retargeter now misses much of the product. Forgetting the retargeting origin misses the source of its decisioning culture.

The same bid through a different lens

Imagine a kitchenware brand launching a new espresso machine. The Trade Desk can help an agency allocate budget across streaming television, premium publishers, audio and display, use first-party or partner data, control frequency across channels and compare paths through the supply chain. Its advantage appears when the brief asks for broad reach, cross-channel coordination and an independent view of inventory.

Criteo can approach the same campaign through shoppers who browsed coffee equipment, product-level retail inventory, sponsored listings and offsite ads, with sales reporting from participating retailers. Its advantage appears when the brief asks for a short line between shopping intent and a transaction. The target is not just an audience profile. It can be a relationship between a person, a product and a retailer.

What a buyer can steal from both

The platforms are not sealed boxes. The Trade Desk has retail-data partnerships and OpenPath, which creates more direct connections to publishers. Criteo has an open-internet DSP, supply-side technology and media access far beyond a retailer's own site. Their product surfaces overlap more each year. Origins still matter because they shape the data each company collected, the clients it learned to serve and the metrics its systems learned to improve.

A useful evaluation therefore begins with four questions: Who pays the platform? Which data enters the bid? Where can the ad run? How is the result counted? Ask for each fee in the chain, not only the headline platform rate. Separate retailer-reported return on ad spend from incremental sales. Check whether identity and frequency controls survive movement between channels. Make data portability and log-level access part of the commercial discussion, not a technical afterthought.

Then resist the clean but lazy ending in which one vendor takes the whole budget. If both can plausibly perform a job, give each a job-shaped test. Keep the audience, creative, geography and conversion window as comparable as the systems allow. Exclude existing buyers when measuring acquisition. Hold out a control group. Reconcile platform reports against retailer, site analytics and finance data. A dashboard can prove that a campaign found buyers; only a careful test can suggest that it created them.

1. Name the job. Choose reach, acquisition, retargeting, retailer monetization or sales measurement before choosing the interface.
2. Draw the fee stack. Include platform, data, measurement, supply-path and retailer charges in the working media calculation.
3. Protect the test. Control overlap, frequency, attribution windows and existing-customer exposure.
4. Judge a business result. Compare incremental reach, profit or sales, not only clicks, platform ROAS or cheap impressions.

Choose the job, then choose the machine

The Trade Desk is the cleaner first call when an agency wants an independent buying layer across many open-internet channels, especially when connected television, supply-path control and flexible data activation lead the brief. Criteo is the sharper first call when retailer relationships, product-level shopping signals, sponsored placements, dynamic retargeting or closed-loop commerce measurement lead it.

Neither description is permanent. The Trade Desk keeps moving closer to premium publishers and retail data. Criteo keeps moving outward from conversion into full-funnel buying. The practical advantage belongs to the marketer who treats those road maps as evidence, not destiny. A platform should not win because its category label sounds right. It should win because its incentive, data and measurement design fit a defined piece of work.

That is the ad-tech choice hiding inside the media budget. One line item says “programmatic,” but the dollars beneath it can be buying very different things: neutral access to a broad market, privileged knowledge of commerce, or some negotiated blend of both. Make that choice visible and the rest of the plan gets easier to defend.