Published on the 05/06/2025 | Written by Heather Wright

And GenAI’s growing buying influence…

Companies need to start factoring in behavioural outcomes when it comes to their AI purchases and become as rigorous about those outcomes as they are about the technology and business outcomes.

That’s according to Mary Mesaglio, Gartner distinguished VP analyst, who says for too long – even pre-AI, organisations have failed to consider behavioural outcomes to the detriment of workers, customers and businesses.

“Behavioural outcomes are the outcomes most overlooked. And they will matter most in terms of getting the business outcomes you want.”

A 2024 survey by Gartner showed just 20 percent of CIOs say they focus on mitigating the potential negative impacts of GenAI on employee wellbeing.

“If you’re serious about AI, you and your enterprise have to care about three kinds of outcomes: Business outcomes (does it deliver ROI, some value), technology outcomes (is is secure, scalable, is the integrity of data really good) and behavioural outcomes,” Mesaglio said during a recent webinar.

While businesses instinctively know they need to be rigorous about business outcomes –  doing ROI analysis, building business cases, tracking benefits – and tech outcomes, when it comes to the behavioural side, most businesses don’t know how to be rigorous, or even who should be monitoring it.

“Behavioural outcomes are the outcomes most overlooked. And they will matter most in terms of getting the business outcomes you want.”

She outlined three AI deployment scenarios – defend, extend and upend – noting each of them presents its own behavioural challenges.

For defend deployments – those where AI is used in everyday work as an assistant to remove drudgery through summarising meetings or documents, for example.

“The truth is there isn’t an ROI here [in this AI use case] but that’s not something executives want to hear,” Mesaglio says.

They’ve spent money on a licence. They want an ROI. And there is reward, but it’s not in financial terms, she says. Instead, it’s in employee wellbeing – or return on employee – helping them feel more engaged, productive and less disenfranchised.

Here, Mesaglio says the behavioural challenge is adoption.

The fix, however, isn’t all that complex: “Don’t take a tools and technology-based approach, but instead going to a skills and learning-based approach.”

Instead of asking if staff are using Copilot, saying they aren’t allowed to use a particular tool, or hoping they’ll spend their free time learning AI, she urges companies to focus on micro-skills and microlearning – a two minute video on how GenAI can help with writing or how to do parallel prototyping, or microbursts of learning from peers, an area she says is proving ‘enormously effective’ in getting people to adopt AI tools.

In the extend category, where companies are using AI to extend, enhance, improve or sustain something they’re already doing, ROI is an effective measure of initiatives and the return will focus on traditional business objective measures.

But, Mesaglio warns of a design challenge when it comes to behavioural challenges in this area, noting the benefits of GenAI are ‘totally uneven’, with some users gaining considerably more benefit from using AI than others, and not everyone getting the same access.

“Also, not enough organisations are asking if AI is doing a portion of someone’s job, and it’s better at it than them, what is their new job?”

She cautioned about ‘accidental decision delegation’ where users unthinkingly delegate decisions to the machine (case in point, blindly following your navigation system), and ‘experience compression’ – an area where new recruits, such as those in a call centre, can garner the same results using AI as someone who has spent five years – without AI – in the job.

“From a tech outcomes perspective, that’s absolutely a good thing. They’re providing great answers, people are learning fast. From a business outcome the answers are accurate and creating great ROI, getting people up to speed much more quickly and you don’t have to train them as much.

“But on the behavioural side? If you learn to be as good as me in three months and I’ve been here five years with my blood sweat and tears, how do I feel? Not so good. Threatened, what is my career path? How do we manage those two things – should you be as rewarded the same as I even though I’ve been here five years? What’s our talent strategy?

“This notion of experience compression is really cool from business outcome perspective, but turns out to be complicated from behavioural perspective.”

Also on the watch list: Skills atrophy. Just like the computer has meant penmanship is no longer a required skill for most, or Google Maps means map reading is necessary for most of us, with AI there are skills that are atrophying and companies should consider having a skills baseline set employees have to know – doing the first five negotiations for themselves, or writing the first five briefs for example, to ensure they have the baseline skill level and baseline level of discernment to know whether what the machine is providing is good or not.

Mesaglio says the big takeaway is the need to be intentional.

In the upend deployment area, where companies are making bets on AI that they don’t really know will work, the ‘return on the future’ magnifies the behaviour challenges for AI and the human.

She noted the currently massively hyped area of agentic AI.

“When you are in the world of return on the future in genAI innovation, you are in the world of relationship creation. You are not just creating a business solution, you are not just creating a tech solution, you are creating a relationship between a human and machine.”

She urged companies to think about the depth, duration and dependence of the relationship now and how it might evolve over time.

“Is it supposed to engender deep emotional connections or is it just trancactional and relatively superficial. For duration, is it one off or something you will have long term, getting to know someone, modifying its behaviour in order to know them? How will it evolve?

“And when considering dependence, what is the nature of the work being done by the machine? Is it removing drudgery, doing something the human can’t? Doing something the human will rely on for work or life? Does it know secrets? Does the agent depend on the human?

“Just asking these three questions – depth, duration, dependence – gets you more focused on the relationship lens and less on just the business and tech outcomes.”

She cited a non-AI example of a company using a fully automated check-in process where the machine could only speak ‘red light, green light’ when customers placed their baggage on the bag drop machines. Customers were simply provided with either a green light or a red light – but no information about how overweight the baggage was. As a result, customers were clamouring to remove contents of bags, retrying only to receive a red light again and again.

“From strictly financial outcomes does it make sense to automate process? There’s headcount reduction,  saving money, cost efficiencies, check, check, check. From the tech outcome, it works, it’s accurate, weights the bag and zips it off on conveyor belt. Perfect.

“Does it make sense from behavioural outcome? What relationship are you creating between machine and human? With red light green light one the machine has all this power, but no humanity,” she says.

The changing face of business buyers

Mesaglio’s comments come as a report from another analyst firm – Forrester – highlights the role of GenAI in in IT buying.

The State of Business Buying in Asia Pacific 2024, which draws from extensive survey data across the region, shows local tech buyers, now largely under the age of 45, are leaning heavily into generative AI to help them with their buying journey

The report shows 71 percent of APAC business buyers are now under the age of 45, and 68 percent are using generative AI to research and compare vendors, signalling a growing preference for self-service discovery.

But those younger buyers are also more likely than their older counterparts to be dissatisfied with their winning provider, citing reasons ranging from technical issues with the vendor’s ability to handle implementation through to cultural concerns about the lack of dedicated DEI programs.

At the same time APAC buyers are less likely than European and North American counterparts to rank price as a primary factor in their decision – at least until it’s time to seal the deal, when pricing rises to the top with budget and price the key factors cited for deals stalling. They’re also less likely to have a fixed vendor in mind at the start of the buying process, making for a more open playing field, potentially enabling companies outside the incumbents and those with high brand reputation to win deals.