Survey: 77% Reevaluate AI Vendors at Least Every Six Months

A survey of 150 enterprise decision-makers found that most regularly reconsider their AI vendors, even as spending on AI continues to grow.

Enterprise AI budgets are growing. That doesn’t mean vendors can count on keeping the business.

Research released in August by Madrona, a Seattle-based venture capital firm, found that 77% of the enterprises it surveyed reevaluate their AI vendors at least every six months or on a rolling basis.

The survey doesn’t tell us how many of those reviews result in a switch. But it does show how often AI providers may have to prove their value after winning the account.

AI Budgets Keep Growing

Companies are still spending on AI. They’re also still deciding which vendors deserve that spending.

Madrona surveyed 150 senior enterprise decision-makers. Seventy-four percent expect to increase their AI spending during the next 12 months. Only 1% expect to cut it. Forty-five percent now have a dedicated AI budget.

Vendor choices are still unsettled.

Twenty-nine percent said they reevaluate AI tools on a rolling or as-needed basis. Another 14% do so monthly, 21% quarterly and 13% every six months.

Companies have money to spend. They’re still working out where AI fits in the business and which providers they want to keep.

Companies Are Struggling to Move AI Pilots Into Production

Getting an AI pilot into regular use is another challenge. Madrona found that 83% of respondents moved fewer than half of their AI pilots into production during the previous 12 months. Thirty-five percent moved fewer than one in four.

The technology itself wasn’t the biggest problem.

Integration problems were the most commonly cited reason pilots failed to move forward, selected among the top reasons by 55% of respondents. Security, privacy or compliance requirements followed at 52%. Thirty-three percent cited difficulty proving enough financial value.

Only 27% included “did not work as promised” among their top reasons. It ranked sixth. A product can work and still have trouble becoming part of a customer’s day-to-day operation. It may be difficult to fit into existing systems. Security or compliance concerns may get in the way. People may not use it enough to justify the cost.

For the provider, winning the contract is only the beginning.

Insider Take

These findings come from 150 enterprise decision-makers. They shouldn’t be treated as a measure of every AI buyer or every business built around annual recurring revenue.

AI is a particularly unsettled market right now. Companies are increasing their budgets while they work out which tools they need and where AI belongs in their operations. At the same time, new vendors keep entering the market and existing products keep changing.

Buyers have plenty of reasons to reconsider choices they made only a few months ago. An annual contract may secure the business for a year. It doesn’t guarantee the customer will renew. In AI, customers appear to be reconsidering those choices unusually often.

The pilot findings show how much has to go right before an AI product becomes part of the way a company works. Integration can get in the way. So can security concerns or difficulty proving enough value to keep spending.

This doesn’t mean recurring revenue across the subscription economy has suddenly become less dependable. Established subscription businesses may have years of renewal and retention history that show how dependable their revenue has been. Many AI providers are still building that history. Their customers are still deciding which products belong in the business and which vendors deserve to stay.

Winning the contract brings in the customer. Keeping the revenue depends on whether that customer continues to see enough value to renew.

Sources

• Madrona, “Harnessing Enterprise Value: The ROI of AI,” August 13, 2026

• Madrona, “Harnessing Enterprise Value: The ROI of AI,” Full Report, August 2026

• TechCrunch, “Startup ARR Is Less Secure Than Ever, New Research Shows,” September 3, 2026