New G2 Research: AI Is Reshaping How B2B Software Deals Are Won and Lost

Finding software has never been faster, but getting to final approval for the sale has never been more complex. That's the thesis of G2's 2026 Buyer Behavior Report: The Evaluation Maze, which is based on a survey of more than 1,000 B2B software buyers and decision-makers, paired with interviews from more than 50 B2B sales and marketing leaders.

The discovery phase has compressed from hours of website browsing and report reviewing to a single prompt in an AI chatbot. Eight out of 10 buyers have sourced software recommendations from tools like ChatGPT or Google AI Mode in the last two years, but once that research is done, they face an uphill journey. Approval isn't easy in today's environment because AI is often still categorized as risky, expensive, and opaque.

As G2 Chief Innovation Officer Tim Sanders notes in the report, AI has solved one constraint but created a new one. Now that finding software has gotten easier, the evaluation stage is bumpier than ever. Evaluation is where buyers compare finalists, validate proof, scrutinize pricing, assess security, pressure-test implementation, and ultimately decide whether to commit to the spend.

The surest way to see this change is to follow the money. Nearly half of software buyers said their CFO vetoed an approved deal in the last year, and 7 in 10 say the pace of AI innovation is pushing them to ask for shorter contracts. The question is no longer just whether a product works. It is whether the pricing model makes the risk of spending overruns worth taking.

Here's what the data shows. And more importantly, what winning brands are doing differently.

Evaluation is now the longest stage of the buying journey

AI has compressed the research and discovery phase of software buying, but it has also redistributed the friction rather than removing it. Buyers are getting to a shortlist faster than ever, but that shortlist now faces more internal scrutiny before a purchase is approved.

Evaluation is now the longest stage of the buying journey, surpassing research for the first time.

Once a vendor is selected, the internal champion now has more obstacles before reaching approval. IT security review is the single biggest source of delay, cited by 39% of buyers overall and rising to 50% among enterprise buyers. Budget approval (32%) and implementation planning (25%) follow close behind.

AI has added a new layer to that scrutiny. Concerns about internal resistance to AI adoption grew from 16% to 29% in a single year - the largest single-year shift in the entire study. Buyers want AI products, but the path to implementation is more complex now that they have to convince colleagues as well as stakeholders.

The shortlist forms faster with AI

By the time a seller gets a call, an email, or a demo request, a buyer's shortlist is often further along than sellers realize, and, increasingly, it's shaped by AI. Eighty-two percent of buyers sourced software recommendations from an AI chatbot in the last 24 months, and half of those buyers said AI had its greatest impact when narrowing and comparing options.

Review sites (38%) also rose to be the top source shaping which vendors make a buyer's shortlist, surpassing AI chatbots (37%) for the first time. That means vendors who aren't showing up in AI answers, and who have a thin presence on review sites, are losing deals before realizing they were in the running.

CFOs are reshaping how software deals get done

Finance involvement in software decisions jumped from 31% to 46% in a single year, and as the cost of AI becomes more important to understand, leaders are now taking a more active role in contract and budget negotiation. CFOs are reversing decisions that aren't right-sized for AI, and championing other purchases when value is clear.

The dynamic is even sharper among more AI-mature organizations. In companies with dedicated token or LLM budgets, the number of buyers whose CFO vetoed an approved purchase climbs to 54% - nearly double the rate of organizations without one (29%). The more committed a company is to AI spending, the more scrutiny individual purchases face.

That scrutiny is changing buyer expectations and reinventing contract composition. Three in four buyers who have experienced a late-stage veto now expect positive ROI within six months of signing. They also push for contract terms under 12 months at more than double the rate of other buyers (40% vs. 18%).

AI is rewriting the economics of SaaS

Finance scrutiny is reshaping individual deals, but it's part of a broader shift in how software gets priced, budgeted, and contracted across the board. Even buyers who haven't faced a CFO veto are rethinking contract terms. The pace of AI innovation is introducing new cost variables that are harder to predict and defend, and buyers are responding by demanding more flexibility from the start. Seventy percent say that pace is pushing them toward shorter contracts, and the traditional seat-licensing model is losing ground fast.

AI is the primary driver of this shift. Eighty percent of organizations now provide developers or technical teams with a dedicated token or LLM usage budget, introducing a new layer of spend that procurement teams are still figuring out how to plan for. That unpredictability is showing up in how buyers want to pay, and as a result, preference for outcome-based pricing more than doubled since 2025, from 11% to 23%.

The market is already adjusting. Half of buyers have already been offered a variable-cost pricing option in place of a traditional seat license or subscription, and another 42% have been told those changes are coming. That means 91% of buyers are either already navigating new pricing structures or about to be. For software companies still anchored to legacy pricing models, there is reason to lean into this change in preference, but the window to adapt is closing. Fifty-two percent of buyers we polled said variable pricing improved their perception of a vendor.

Agents are part of the conversation, but humans make the decision

Sixty-one percent of buyers currently use or plan to use AI agents as part of the buying process, and another 19% would consider them for select use cases. Interestingly, the most common use cases for AI are concentrated in the evaluation stage - things like understanding total cost of ownership (51%), building shortlists (51%), researching solutions (49%), and evaluating shortlisted vendors (46%).

Still, there's a clear line buyers aren't ready to cross. Less than half (47%) would allow an agent to conduct research and make recommendations while humans retain all final decision-making authority. Only 9% are comfortable letting an agent execute purchases within approved guardrails, and just 2% would allow purchases without pre-approval.

Agents are being welcomed into the evaluation process as researchers and analysts, not as decision-makers.

What software brands can do to win

Buyers arrive at evaluation informed and skeptical. They've done the research, built the shortlist, and in many cases, have a frontrunner in mind. The brands that win are the ones that make it easiest for buyers to say yes and defend that decision internally.

Four things to focus on:

1. Show up where buyers are looking, for humans and agents alike. Before a human is ever involved, buyers are researching, comparing, and shortlisting with AI. Make it easy for chatbots and agents to find you, understand what you do, and recommend you. Then, arm your human champion with security documentation, ROI estimates, and a clear explanation of how your AI features work.
2. Bring flexible pricing and a CFO-ready business case. Finance is more involved in software approvals than ever, and buyers are walking into evaluation expecting to be challenged on cost. Outcome-based pricing, flexible contract terms, and clear visibility into token usage are now the baseline expectation. Show up with them before you're asked, and you stand a greater chance of winning the deal.
3. Treat the shortlist as the starting line, not the finish line. Winning the deal means helping buyers navigate everything that comes after they've selected a preferred tool or service: security review, budget considerations, adoption, and value after signing. The brands that consistently provide proof at each stage will close more deals than the ones that stop selling after the demo.
4. Remember that evaluation never really ends. Buyers re-evaluate vendors they've already chosen, and a better alternative is always one AI query away. Protect renewals by assigning your strongest team members to evaluation, onboarding, and retention. A positive early experience is the best defense against being shortlisted out at renewal.

Welcome to the evaluation maze

Every shift in this report traces to AI as a catalyst. While it compressed software research into a prompting exercise that can be completed in a day, it simply redistributed that friction to evaluation, where cost, security, and internal trust now decide deals.

The buying journey has fundamentally changed, and the brands that adapt to account for that friction will give themselves the best opportunity to escape the maze and win the deal. The buyers in your pipeline are already shortlisting without you. The question is whether you'll be ready when they reach evaluation.

Explore all of this year's findings: G2 2026 Buyer Behavior Report: The Evaluation Maze.