Published on the 20/05/2025 | Written by Heather Wright

Businesses revamp purchasing in response to AI…

IT deal sizes are getting smaller and decisions are being made faster – and with less people at the buyer’s table – but when it comes to buying AI, well, there’s a stricter set of requirements and embedded AI is on everyone’s list.

That’s according to software review platform G2, which surveyed nearly 1,200 B2B decision makers globally earlier this year.

“A growing segment opts to pay strictly for measurable outcomes.”

It shows nearly three out of four companies are imposing stricter requirements when they’re evaluating AI-powered software, versus non-AI solutions, with enterprises of 1,000-5,000 employees setting the highest standards for AI software procurement.

The report does hint at growing confidence and maturity in managing AI-related risks with the year-over-year data showing buyers in fact moving from a ‘much stricter’ to ‘somewhat stricter’ approach for AI.

And if we still have some concerns about AI, that’s not stopping the demand, with two out of three respondents now actively considering AI capabilities when selecting software. For self-identified power users – 36 percent of respondents – AI functionality is now a must-have to make their shortlist, leading the report authors to dub AI ‘always included’.

But vendors had better be able to demonstrate clear value or productivity gains from the AI added into their offerings if they want to charge extra. Nearly half of enterprise buyers said they had switched to software with better AI over the past year, specifically to gain access to those AI features. Midmarket respondents were also making the move (42 percent), and even SMBs were on the move (32 percent).

As to what they were seeking, it was better insights and decision-making (22 percent) that lured them across to an AI product, followed by operational efficiency and cost saving (15 percent). Lagging behind in third was innovation/future proofing, at 6 percent.

The report also shows the evolution of deal sizes and scope, clearly showing that multi-year, mega-deals are shrinking. Instead, companies, including larger enterprises facing macroeconomic headwinds and increased scrutiny or ROI, are downsizing the deals, looking for smaller contracts tied directly to performance.

That’s a trend long seen for smaller businesses, who unsurprisingly have long favoured more modest deals aligned with their cash flow management priorities. But the mid-market too, is embracing the ‘do more, but spend less per deal’ mindset too, and enterprises are clustering their purchases in the $100,000 to $150,000 range, the report shows.

Usage-based or short-term software is also a driver in the shrinking deal sizes. The report says generative and agentic AI are ‘reshaping’ pricing models – taking companies away from the more traditional subscriptions or fixed fees – with more than one in three buyers now preferring variable pricing.

“Some pay based on API calls, others are billed by agentic actions, and a growing segment [10 percent] opts to pay strictly for measurable outcomes.”

In what is perhaps a warning for users however, the report notes: “This evolution to a pay-as-you-go approach directly contributes to smaller initial deals, prompting sellers to start small and strategically accelerate usage to expand account revenues over time.”

Companies are also making the decisions about what they want more quickly, and using smaller decision making teams to drive those decisions: The once dominant five-to-eight member buying committees have shrunk, with quartets and trios on the rise, even among larger companies, with the exception of companies of 1,000 to 5,000 employees, who still strongly favour a five-to-eight person team. Interestingly, those over 5,000, while still favouring the five-to-eight person team (24 percent) were much more likely to embrace a four person team (17 percent) than those in a 1,000-5,000 employee company.

“Frequently, champions now work closely with just a handful of colleagues to swiftly evaluate and select the best solution,” the report notes.

The smaller deal sizes, along with increasing urgency to modernise quickly are driving the move to efficiency over exhaustive consensus-building.

AI is also playing a role in our decision making, with the research showing we’re placing plenty of trust in genAI chatbots to provide confidence in our final decisions – and more so in APAC than elsewhere, with 19.2 percent of APAC respondents rating the genAI chatbots as their most trusted source of information. That’s well ahead of peers and colleagues at just 11.9 percent or software review sites at 12.4 percent. Interestingly, vendor sites were ranked higher than either of those options, at 13 percent in APAC.

“For buyers navigating complex evaluations, generative AI simplifies the process by synthesising extensive information into a clear recommendation, streamlining their path to a confident decision.”

“Trust in generative AI has grown significantly, thanks to enhancements like verification links in chatbot responses and buyers providing clear context to minimise errors. This has elevated generative AI from a research assistant to a trusted advisor.”