The AI bubble and affiliate marketing: how to spot a product worth promoting

The money flowing into AI is extraordinary. Stanford’s 2026 AI Index estimates global private AI investment at $344.7 billion in 2025, including $170.9 billion for generative AI companies. Those figures measure investment, not the revenue or profit earned from it.

There is real use on the other side of that spending. The same report says 88% of surveyed organizations used AI in at least one business function in 2025. Yet agent deployment remained in the single digits across nearly all business functions. Adoption is broad; the depth and financial value of that adoption vary.

For an affiliate, neither figure picks a winning offer. A popular category can contain excellent tools, interchangeable products and companies spending heavily to acquire customers they struggle to keep. The label “AI” tells you little about which one you are looking at.

Before checking the commission, write down the buyer’s task in one sentence. “Generate social posts” is broad. “Turn a weekly product release into three usable ad concepts that a small team can edit and approve” is something you can test.

Use the product for that task more than once. The first attempt tests the demo. Repeating the work tests whether it fits a routine. Record how much preparation it needs, how often the output requires correction and whether another tool the customer already pays for can do the same job.

In a field experiment across 66 firms and 7,137 knowledge workers, users of a generative AI tool spent about two fewer hours a week on email during the latter half of the study. The researchers did not detect broader changes in the quantity or composition of workers’ tasks from individual access alone. That is a useful result for a defined task, not a license to claim every AI subscription transforms a business.

A review becomes more credible when it says what improved, what still needed human work and who would get little value from the product.

A tool can convert well because trying it costs nothing. That does not mean trial users become paying customers or stay subscribed.

Follow the buyer’s journey yourself. What can a new user accomplish before paying? Which feature triggers the upgrade? Are usage credits or output limits easy to understand? Does the price still make sense at the volume a real customer needs? Check the cancellation route and look for recent product changes that could alter the answer.

Then ask the program how it defines a commissionable customer. A “sign-up” may mean a free account, a qualified trial or a completed purchase. Those are different events with different values. Confirm whether refunds, cancellations and existing customers affect attribution. If the offer promises recurring commission, ask how long it runs and whether it stops when a plan changes.

A high payout per sale is less useful if the sale rarely qualifies. A lower commission attached to a product your audience can use every week may produce a better result for both reader and publisher.

Official program pages show why this matters. Semrush currently advertises a 120-day cookie and product-specific payouts; its listed AI Visibility Toolkit commission is $100 per sale. That is a concrete offer, but an applicant still needs to check current approval rules, eligible conversions and traffic restrictions before planning a campaign.

Access can also change. Canva says its Canvassador Program is now the only pathway to affiliate benefits and that existing affiliates must reapply and be accepted. An old article describing an open Canva affiliate program would give readers and publishers the wrong expectation.

Keep a dated copy of the terms you reviewed. Recheck them before updating a ranking, buying traffic or promising readers a discount. A commission figure without its qualification rules is an incomplete description of the deal.

Compare offers using the whole path from click to paid, approved customer. The payout alone hides two important variables: how many visitors buy, and how many sales the program approves.

Offer B wins in this example because more visitors become approved customers, despite its smaller headline payout. These numbers are hypothetical, not market benchmarks. Real results depend on your audience, the product, attribution and sample size. With only a handful of conversions, one refund can move the result sharply. Track trials, paid upgrades, approvals and reversals separately; don’t declare a winner from clicks alone.

For recurring offers, add another measure: how much commission a referred customer actually earns over several months. A promised lifetime percentage cannot be valued as “lifetime income” before you have evidence that customers stay.

One warning sign rarely settles the decision. Several together deserve a slower test:

A page titled “Best AI Tools” is easy to publish and hard to keep accurate. A page built around a specific task has a better reason to exist: it helps a reader decide whether to pay.

Show the inputs you used, the result you got and the work still required. Compare the tool with an alternative your audience might reasonably choose. Explain which buyer should skip it. Date price and program-term checks. When the product changes, update the parts of the review that affect the decision.

For affiliate marketers, that approach also limits exposure to the hype cycle. If one fashionable tool disappears, a genuinely useful guide to the underlying task can still help readers. The recommendation can change without the whole article losing its purpose.

AI can be a useful technology while individual companies attract more investment than their future profits justify. An affiliate cannot settle that market-wide argument by testing one subscription. They can make a sounder recommendation.

Choose the offer only when you can name the customer’s job, verify that the product handles it, understand the full commission rules and measure what happens beyond the first click. If any of those pieces is missing, the attractive payout is a reason to investigate further, not a reason to publish a recommendation.

Program terms checked November 13, 2026: Semrush affiliate program and Canva affiliate program guidance. Terms may change; verify them before promoting either product.