AI Adoption Is a Motivation Design Challenge

WRITER’s 2026 Enterprise AI Adoption survey found that 79% of organizations face significant AI adoption challenges. 29% of employees admit to actively sabotaging their company’s AI strategy. 75% of executives concede their AI strategy is more performative than operational. Only 29% of companies report significant ROI from generative AI, despite individual super-users being 5x more productive than non-adopters.

Meanwhile, somewhere inside those same companies, an employee built a custom GPT over the weekend that cuts a 36-minute task to 11 minutes. Their colleagues love it. Leadership has no idea it exists.

A recent HBR article by BBVA’s AI adoption team and academics from LSE and UC3M documented the pattern at scale: while only 40% of companies have purchased official LLM subscriptions, employees at over 90% of those companies report regular personal AI use for work. The demand is there. The motivation is there. Companies are designing against it.

The technology delivers value. The deployment model destroys it.

What BBVA Did Differently

The HBR article is worth reading closely because BBVA, one of Europe’s largest banks with 125,000 employees across 25 countries, designed a motivation architecture rather than a technology rollout. The results: 83% weekly usage rates. Over 4,800 custom GPTs built internally. 99% adoption in the audit function. An insurance-advisory GPT in Mexico that cut response time by 92%.

Here is what they actually did, mapped against the Octalysis Framework’s eight Core Drives of human motivation:

They created scarcity with purpose. Only 3,000 licenses were distributed initially, allocated to motivated individuals, not senior ones. A “use it or lose it” policy meant low users would lose their license to eager colleagues. This activated Core Drive 6 (Scarcity & Impatience) in a productive direction. Access felt earned, not imposed. The feeling shifted from “you must use this” to “you get to use this.”

They formalized internal champions. 200 “Wizards” were selected for enthusiasm and skill, becoming peer-to-peer experts. Their incentives were a mix of intrinsic enjoyment and reputational status within a domain the organization clearly valued. Core Drive 5 (Social Influence & Relatedness) plus Core Drive 2 (Development & Accomplishment).Early adopters got social recognition and a meaningful role.

They built a Community of Practice. It became the most active internal forum in the company’s history. Employees shared custom GPTs, troubleshot problems, discovered innovations from other departments. An engineer in Mexico built a sentiment analysis tool. Other departments adopted and adapted it. Core Drive 3 (Empowerment of Creativity & Feedback) plus Core Drive 5. Innovation spread horizontally, not through top-down directives.

They trained managers through their own work. A mandatory five-hour workshop for the 250 most senior managers, including the CEO, focused on practical use: preparing investor days, drafting communications, writing performance feedback. Not a feature demo. Not a theoretical overview. Core Drive 4 (Ownership & Possession). Managers felt they were leading the adoption, not being swept along by it.

They preserved human agency. AI is an assistant, not an autonomous agent. Employees own the output. No direct writes to core systems without human validation. Core Drive 4 again, plus protection against the avoidance motivation that fear of replacement triggers.

The pattern: BBVA gave employees a journey (from initial scarcity through expanding access to community participation), agency (they chose their own use cases and built their own tools), and social context (peer networks, visible recognition, shared learning). These are the ingredients of a well-designed experience.

What Most Companies Do Instead

Most AI adoption programs follow the same linear sequence. Procurement selects a tool. IT configures and secures it. HR creates a training module. A launch email goes out. Usage is tracked by login frequency. Success is declared, or failure is blamed on “change resistance.”

This is how you deploy software. Changing behavior requires designing a journey, not a launch event.

The default approach has three structural flaws that show up clearly when you compare it side-by-side with what BBVA did:

Linear when adoption is not. Human motivation does not move in a straight line. People need a reason to start (Discovery), early wins that build confidence (Onboarding), deepening engagement as they find personal value (Scaffolding), and eventually a sense of mastery and identity (Endgame). These are the Four Experience Phases that anchor the Octalysis approach. Skip any of them and adoption stalls. Most AI programs invest everything in announcement and training, then wonder why usage plateaus after week two.

Function-focused when it should be human-focused. Employees do not wake up wanting to “use an AI tool.” They wake up wanting to finish a report without staying late, or stop answering the same customer question 40 times a day, or finally have time for the strategic work that keeps getting crowded out. When you design around functions, you get a product demo. When you design around human needs, you get adoption.

Strips users of agency. Someone else chose the tool. Someone else decided the use cases. Someone else set the KPIs. The employee’s role is to comply. People engage deeply with systems where they have meaningful choices. Strip that away and you get exactly what the data shows: shallow adoption, sabotage, and shadow IT.

The Behavioral Mechanism

A study in Frontiers in Psychology found that organizational AI adoption triggers two competing motivational responses simultaneously. Approach motivation: excitement, curiosity, expanded capability. Avoidance motivation: anxiety about obsolescence, fear of looking incompetent, stress about learning curves. Every employee experiences some mix of both.

The design of the rollout determines which response wins.

Top-down mandates and “use it or lose your job” messaging (60% of companies plan to lay off non-adopters, per WRITER) amplify avoidance. They activate what the Octalysis Framework calls Black Hat Core Drives: Loss & Avoidance (CD8), Scarcity & Impatience (CD6), and Unpredictability (CD7) in its anxious form. These drives produce action. People will open the tool if their job depends on it. But Black Hat motivation is exhausting and unsustainable. It generates compliance, not commitment. Compliance does not build custom GPTs, share best practices, or redesign workflows.

The alternative is designing for White Hat Core Drives, which is what BBVA did:

• Epic Meaning & Calling (CD1): employees connect AI use to a purpose larger than task completion, like the BBVA insurance advisor that reduced client wait times by 92%.
• Development & Accomplishment (CD2): they feel a growing sense of mastery, like the 4,800 custom GPTs that turned employees into visible builders.
• Empowerment of Creativity & Feedback (CD3): they can experiment and iterate, like the engineer in Mexico whose sentiment-analysis tool spread across departments.
• Ownership & Possession (CD4): they feel the AI-enhanced workflow belongs to them, reinforced by BBVA’s hard rule that employees own the output.

White Hat motivation is what separates the employee who builds a GPT that saves their department 19 hours per week from the one who completed training and never logged in again.

Wharton professor Ethan Mollick made a parallel point in his 2026 keynote at Valence’s AI & The Workforce Summit: the most advanced AI users are already inside your organization, using AI in secret. The bottleneck is not capability. It is organizational design. Mollick’s framework: “leadership, lab, and crowd.” Leaders model AI use. A lab validates applications. The crowd scales what works. Leadership modeling activates CD1 and CD5. The lab activates CD3. The crowd activates CD2, CD4, and CD5. Different drives for different stages. The design has to match the journey.

A Five-Step Redesign Playbook

If your AI rollout is stuck in the failure pattern, here is what BBVA’s approach translates into operationally:

Step 1: Discovery — Find your hidden innovators. Do not start with mass training. Start by identifying 50-200 motivated early adopters across the organization, regardless of seniority. The signal is not their job title. It is whether they would build something with AI given the chance. They are already there. They are using AI in secret. Make it safe to surface.

Step 2: Onboarding — Give them real problems, not feature demos. Distribute access competitively. Make it scarce and earned. Pair early adopters with concrete business problems they care about: cutting their own report time, automating their own repetitive work. The first 90 days should produce visible wins owned by the people who built them.

Step 3: Scaffolding — Formalize the power users. Identify the top 5-10% as “Wizards” or “Champions.” Give them a name, a community, and recognition. Their job is peer mentorship, not enforcement. They become the horizontal channel through which innovations spread. This is the move that scales adoption from a pilot to a culture.

Step 4: Endgame — Build the Community of Practice. A persistent forum where people share custom GPTs, troubleshoot together, and discover innovations from other departments. This becomes the most active forum in your company if you let it. It also becomes your competitive moat: peer-built solutions that competitors cannot copy because they emerged from your specific workflows.

Step 5: Governance — Human in the loop, always. AI is an assistant, not an autonomous agent. No direct writes to core systems without human validation. Employees own the output. This is not just a compliance posture. It is the rule that protects Core Drive 4 (Ownership) and prevents the fear-of-replacement spiral that kills adoption.

What We Have Seen in Practice

We have applied this kind of thinking across industries for over a decade, working with more than 170 organizations.

When Procter & Gamble needed to drive sales performance across 19 countries, the challenge was not giving reps better tools. It was making them want to use those tools daily. We designed a gamified platform that transformed the sales process into an experience where reps could see their progress, compete meaningfully, and earn recognition from peers. The project won a Best Gamification Award.

When Microsoft needed to deepen learning outcomes on its education platform, the issue was not content quality. It was user engagement. We designed journeys that gave learners a sense of progress and autonomy, reducing dependence on first-line support.

When CAIXA Econômica Federal needed company-wide engagement across Brazil’s largest public bank, the answer was not another top-down initiative. Daily active usage hit 92.61%. Annual net profit increased by $1 billion in six months. CAIXA became Brazil’s #1 public bank.

When Astana Hub needed to transform a function-focused innovation portal into something its startup community would actually use, we redesigned the experience around motivation, not features. Active users increased by 504%.

The common thread: the technology was never the constraint. The constraint was always the human experience surrounding it. Fix the experience, and adoption follows.

The Design Shift That Matters

AI adoption will not be solved by better training decks, stricter mandates, or more sophisticated dashboards tracking login frequency. It will be solved by treating it as what it is: a product experience design challenge.

The Octalysis Framework provides the structure to do this systematically. The 8 Core Drives map the full range of human motivation. The Four Experience Phases ensure the design adapts as users mature. Over a decade of application across Fortune 500 clients and startups has proven that the approach moves hard KPIs, not just sentiment scores.

The organizations that figure this out will not just adopt AI. They will build cultures where AI adoption is self-sustaining, where employees drive innovation from the inside, and where the gap between individual productivity and organizational ROI finally closes.

The ones that keep treating adoption as a technology rollout will keep producing the same statistics. 80% failure rates. 29% sabotage. Millions invested, marginal returns delivered.

The choice is a design choice.

The Octalysis Group is the world’s leading gamification and behavioral design consultancy, co-founded by Yu-kai Chou and Joris Beerda. We help organizations design human-focused experiences that drive measurable results. Explore our case studies or contact us.