Walk into any CMO office this quarter and you’ll find a variety of vendor proposals based on 12-month payback periods, with vendors fairly confident about the first year and a little less confident about the third year. This is where a successful pilot project turns into technical and operational debt.
The proposals of trading through agents are starting to get confusing. One rewrites the funnel, another consolidates the stack, and a third replaces the tools the CMO purchased 12 months ago. The acronyms are present (UCP, MCP, A2A, etc.) and the demos are impressive. But like any emerging category, there are big unknowns: acquisitions, evolving protocols and pricing changes that can upend a three-year roadmap.
To evaluate these investments, it is useful to consider a channel that has already survived decades of technological revolution: direct mailing.
As unfashionable as it may be to say it, direct mail will outlast most agent commerce startups pitching to you today. That’s a strange thing to say in a year where AI agents influenced 20% of all Cyber Week orders in 2025 and generated global sales of approximately $67 billion, according to Salesforce. The ability is real, and so is the money behind it.
Direct mail, meanwhile, was expected to die with the arrival of email, and then again for banner ads, programmatic media, social and retail. It’s still on budget, having survived four rebuilds of everything around it.
Despite all the attention-grabbing headlines to the contrary, direct mail endures because brands own what makes it work. This is a useful test for any investment in agent trading you are considering today.
Direct mail has survived because brands own the resources, focus on durable features, and measure results consistently. These same characteristics provide a useful framework for evaluating investments in agent trading.
Your customers search everywhere. Make sure your branding introduces himself.
The SEO toolkit you know, plus the AI visibility data you need.
Start your free trial
Start with
1. The brand owns the asset base
Direct mail runs on an infrastructure that is unlikely to be disrupted anytime soon, as postal addresses are a public utility. Customer files, response history, and accumulated knowledge of what works for a particular audience become more valuable over time. Everything remains with the brand.
Much of the last decade in digital has offered a slow (and expensive) lesson in how little of the underlying infrastructure brands actually own, and how that lack of ownership translates into revenue for the platforms that claim it first.
Instead, advertising inventory is rented from platforms whose auction logic changes frequently. Public access is rented by networks that have narrowed targeting more than once, often with little transparency. Marketers have experienced the “end” of third-party cookies, the ATT, and the decline of organic reach on social platforms.
They are now being asked to build business infrastructure for agent trading. Protocols evolve, LLMs change, and many companies are still searching for sustainable profitability. Data clauses are often reminiscent of social media contracts signed in 2011, when the industry was still in its infancy.
Marketing teams need to ask tough questions during platform demos so they understand where they stand if a vendor changes course, protocols change, or another technology disruption emerges.
First-party structured data and know-how that can be transferred to another system qualify as assets worth investing in. Vendor lock-in and high switching costs, however, can outweigh any initial gains, even if they help beat competitors to the market.
2. The feature provides a solid foundation for new formats
Direct mail has evolved from simple origins to include postcards, self-mailers, catalogs, dimensional packages, variable data printing tailored to CRM segments, informed delivery previews, and even QR codes that insert you into an email sequence.
While the delivery mechanism may evolve every few years, the underlying purpose does not change. Direct mail delivers an addressable message to a known household and measures the response. This was true in 1968 and will probably be true in 2035 as well.
Many new marketing investments follow the opposite pattern, where formats and function remain unclear, especially as specifications, industry alliances and technologies continue to change. Chatbots became conversational AI, then co-pilots, then agents, then agent workflows, and each shift triggered another sourcing cycle.
By 2030, AI itself could fade into the background, as has happened to the information superhighway. If so, platforms whose core strength is AI will have to demonstrate tangible value beyond the label.
Teams that joined the chatbot wave of 2017 can rarely point to a single feature that has survived every rebrand. They can point to four purchases, each of which was eventually replaced.
Gartner expects Moreover Over 40% of AI projects will be canceled by the end of 2027, due to rising costs, unclear business value and poor risk controls. The company also estimates that only about 130 of the thousands of self-described agent sellers are real. The rest is agent scrubbing, labeling older assistants, and decades-old robotic process automation (RPA).
Require vendors to describe their product in terms of foundational work that remains necessary even if the underlying model, interface, and delivery mechanism change completely within three years.
For example, “Accurately representing our offering in machine-mediated purchasing environments” is an enduring function. “Orchestrating multimodal generative experiences across the omnichannel surface” is a format masquerading as function.
Buying new technology is often the right choice. The discipline is identifying the work you would still need to do if the platform disappeared overnight. Direct mail marketers answer this question without hesitation. Agent sellers who can usually deserve your attention.
3. Consistent measurement allows for constant improvement
Direct mail measurement has gradually evolved as everything around it turmoiled. With direct mail, the list, offer, date, cost, and response window are known entities. The tests are based on a control cell, and marketers have been measuring incremental response the same way for at least 40 years.
Digital advertising and engagement, on the other hand, have never achieved the same stability. Click attribution gave way to engagement, engagement to attention, attention to incrementality, and somewhere along the way, media mix modeling came back from the dead. Each wave came with new vocabulary and the suggestion that the previous approach was naive. Agent vendors now sell “agent ROI” and “outcome attribution,” terms that vary by platform and may not survive this rapidly evolving environment.
When evaluating an agent commerce platform, look beyond ease of implementation and consider how benchmarks and metrics will hold up over time. In a situation where agents handle discovery, recommendations, and checkout, and brands appear everywhere agents refer buyers, two questions help determine whether you’re making a long-term investment or getting short-term results:
- What does the brand preserve? This can include product data, interaction history and trading policies that can be migrated to another system and remain as proprietary assets. Compare this to configurations locked within a vendor’s platform.
- What questions will it still answer in 2029? “Did the officer’s recommendations revoke the qualified conversion against a resistor?” is a valuable question. Achieving a high score on a vendor’s proprietary index is not.
Answering these questions helps determine whether an investment is durable or whether the contract terms should reflect a more temporary arrangement while additional build-versus-buy scenarios play out and the market matures.
If a platform’s core capacity is valuable but the surrounding infrastructure remains leased, renegotiating the scope of an initial procurement to secure greater long-term ownership may prove more strategic than accepting the original terms or walking away and allowing competitors to gain a short-term advantage.
Considerations for the next few months
Between the pressure to keep up with the competition and rising customer expectations, the appeal of early adoption is clear. Even so, inaction can be cheaper than plunging onto shaky foundations. The biggest discipline is maintaining consistency by moving quickly, because speed on a leased basis only accelerates the need for migration.
Before your next contract or renewal, define the version of this capability you plan to operate in 2029. If that vision includes resources, knowledge and measurements that belong to the brand rather than the vendor, the investment can increase over time. If you cannot define such proprietary assets, you are likely purchasing a short-term feature with a lifespan of 12 months or less, and the contract should reflect this reality.
Vendors will continue to commercialize future innovations. The strategic choice is to define what you intend to own and invest accordingly.
