AI adoption challenges: How to overcome resistance and build buy-in
AI adoption challenges are becoming more visible as organizations move from early experimentation to measurable, enterprise-wide results. According to McKinsey’s latest State of AI survey, nearly nine in ten respondents report regular AI use in at least one business function. Yet only 44 percent say AI is scaling across their enterprise, and just 37 percent attribute any EBIT impact to it.
The gap is not only a technology problem. AI adoption challenges can include unclear ROI, poor data readiness, security and governance concerns, unrealistic expectations, and employee resistance. Overcoming them requires a practical AI adoption strategy that connects AI to business goals while giving people a clear path to adopt it.
At a glance
AI adoption often stalls for reasons that go beyond the technology. Organizations need to address cost, data readiness, security, governance, employee trust, and change management while showing how AI supports measurable business goals. A practical AI adoption strategy starts with focused, lower-risk use cases, builds confidence through results, and expands as the organization’s skills, data, and governance mature.
Common AI adoption challenges and barriers to adoption
AI initiatives can lose momentum even when the technology works because organizations are not prepared for the process, data, governance, and people changes AI requires. Leaders may see the potential value but still struggle to decide where to start, how to measure success, and how AI should fit into day-to-day work.
Four AI adoption barriers come up again and again:
- Enthusiasm is not readiness. Wanting AI is not the same as being prepared for it. Leaders may be excited about the possibilities, but interest alone does not create a plan. Readiness means your data, governance, processes, and people can support the use case before it goes live.
- Many AI projects work differently from traditional IT projects. A standard ERP or IT rollout often follows a defined scope and planned go-live. AI projects can require more experimentation, testing, evaluation, and feedback. Results also depend heavily on the quality of the data, instructions, integrations, and controls supporting the solution.
- AI changes business processes, not just technology. AI affects how people complete their work, make decisions, and interact with business systems. If you treat AI as another application to install, adoption can suffer. It works best when you design it around the workflows people already use.
- AI still feels unfamiliar to many people. Leaders and employees can hold very different ideas about what AI can and cannot do. Some expect too much. Others do not trust it at all. Closing that knowledge gap is an organizational challenge in its own right.
Common AI adoption concerns by stakeholder
Many enterprise AI adoption challenges come from the fact that each person involved in a decision cares about something different. A CFO wants proof of payback, while an end user wants a tool they can trust. Recognizing these differences helps you address the real concern instead of giving everyone the same generic message.
In our experience, middle managers are often some of the strongest champions for AI. They understand daily workflows well enough to spot useful opportunities, and they have enough influence to help their teams adopt new ways of working.
How to overcome AI resistance and adoption challenges
The objections below come up in many AI conversations. Each one has a practical answer, and most point to a clear first step.
“We can’t justify the cost”
Executives are right to ask what they are getting for their AI spend. The answer is not that every AI project creates value. Value depends on focusing AI on a specific business objective and measuring whether it improves the outcome.
The honest answer on cost is “it depends.” Total cost can include implementation, development, data readiness, infrastructure, governance, change management, usage, and ongoing monitoring.
In our experience, data readiness is often one of the biggest variables affecting AI implementation cost. Clean, well-organized data can make later stages faster, while poor-quality or fragmented data may require additional work before the AI use case can deliver reliable results.
AI usage costs also vary by model and workload, and consumption can increase as adoption expands. Establish budget oversight early so you can track usage and understand which use cases are delivering enough value to justify their cost.
To prioritize spending, score each potential use case on its expected return, your confidence that it will work, and how easy it is to implement. Start with opportunities that score well across all three areas. Early wins help establish the business case for the next investment. For a deeper look, read our guide to forecasting AI costs and calculating ROI.
“Our data isn’t ready”
This is one of the most valid AI objections, and it deserves a direct answer. High-quality data is an important foundation for AI. When information is incomplete, duplicated, outdated, or poorly organized, AI can produce less reliable results that quickly weaken user confidence.
Treat this objection as a starting point rather than a roadblock. A data audit helps you identify where gaps exist and which issues matter to the use case you want to pursue.
From there, you can:
- Define data ownership
- Establish standards for entering and maintaining information
- Address duplicate or incomplete records
- Review which data sources the AI solution needs
- Decide who should have access to sensitive information
You do not need to fix every data problem across the business before trying AI. Focus first on the data required for the use case you want to implement. Learn more about preparing ERP data for AI without creating data risk.
“Is it secure and compliant?”
Security, IT, legal, and risk leaders tend to ask the same core questions:
- Can we protect sensitive data used by AI?
- Are AI agents properly identified, permissioned, and auditable?
- Can we detect new AI-related threats?
- Are we meeting our regulatory and contractual obligations?
- Can we put controls in place without making AI unusable?
Strong answers start with practical controls. Classify data by sensitivity, apply least-privilege access, protect sensitive information, assign clear ownership, and review permissions regularly. Organizations should also establish rules for approved AI tools and train employees on what information they should and should not share.
No single governance policy fits every organization. Your approach should reflect your industry, regulatory requirements, business processes, culture, and risk tolerance. Our data governance and AI readiness services help organizations assess these areas before expanding AI.
“Our people won’t trust it”
Trust is one of the biggest barriers to AI adoption. Employees may question the accuracy of AI outputs, worry about losing control of decisions, or wonder how AI will change their roles.
Hands-on experience can help build trust when employees start with low-risk use cases and learn how to validate AI outputs. Give users a safe way to experiment before asking them to rely on AI for business-critical work.
When the solution supports it, show users the sources, data, or context behind an AI response. Teach employees to validate important outputs rather than accepting an answer because it came from AI. Keep people involved in higher-risk decisions so accountability remains clear.
You should also be specific about how AI may change work. Explain which tasks AI will support or automate, where people remain responsible for decisions, and what new skills employees may need.
That is more credible than simply promising that AI will not change anyone’s job.
“Will the results be reliable?”
AI does not always produce one fixed result the way traditional rules-based software does. Outputs can vary based on the model, instructions, available data, context, and level of autonomy.
That does not mean organizations should accept inconsistent results. Define what a successful output looks like before launch. Set clear boundaries around what the AI should and should not do, test important scenarios, and decide when human review is required.
For AI agents, also define which actions they may take independently and which require approval. Start with narrow tasks where you can evaluate the result easily. Expand the scope as the solution proves reliable and users gain confidence.
“Do we even need AI for this?”
Not every process needs AI, and saying so builds credibility with skeptics.
Traditional automation usually works well for predictable processes that follow predefined rules. AI becomes more useful when a process requires interpretation, unstructured information, pattern recognition, content generation, or responses that change based on context. Do not add AI for the sake of adding AI. Match the technology to the business problem, and your investments will be easier to defend.
“We don’t know where to start”
This may be the most common objection of all. Organizations see hundreds of possible AI use cases and struggle to decide which one should come first. The answer is not to deploy everything at once. Build your AI adoption strategy in stages.
How to build your AI adoption strategy in stages
AI adoption works best as a journey rather than a single launch. Each stage can build the skills, trust, data readiness, and governance needed for more advanced use cases.
Crawl. Many organizations start with an AI assistant that helps employees find information, summarize meetings, draft content, or analyze documents. This gives users hands-on experience while keeping the work easy to review.
Walk. Once teams become comfortable with AI, they can explore AI features and agents inside the business applications they already use. Some ERP, CRM, and productivity platforms now include built-in AI features or agents for defined tasks, such as processing documents, assisting with reconciliation, summarizing information, or supporting sales activities. These can be a practical next step because they fit into familiar applications and existing workflows.
Run. As confidence and governance mature, departments can design agents around more specialized processes. Leadership may also sponsor cross-system use cases where AI analyzes information, coordinates multiple steps, and takes approved actions within clear guardrails.
Adoption is not linear. Certain departments and people will move faster, and they often uncover some of the strongest use cases. Identify these superusers and local champions early. Their results can give more cautious teams practical examples of how AI works.
Across every stage, organizations need to consider data governance, AI literacy, security, and change management.
How to identify and prioritize your first AI use cases
The strongest first AI use cases are usually specific, measurable, and connected to a business problem.
Look for opportunities where AI could:
- Reduce cycle time or manual effort
- Lower operational cost or risk
- Help employees make faster, better-informed decisions
- Improve customer or employee response times
- Increase capacity without adding the same amount of manual work
- Solve a defined process bottleneck
Keep a human in the loop for early or higher-risk use cases, and make sure you can measure the outcome.
To find opportunities, ask your team:
- Which processes require the most manual effort?
- Where are the biggest delays or bottlenecks?
- Which decisions depend on information that is difficult to find or analyze?
- Which teams could benefit quickly from AI assistance?
- Where could AI improve the customer experience?
- Is the required data ready and accessible?
- How would we know if the use case succeeded?
Depending on the process and platform, common AI-assisted starting points can include invoice capture, account reconciliation support, demand forecasting, meeting recaps, document creation, proposal writing, and RFP responses. Generic examples can help you start thinking, but the biggest opportunity may come from a process that is specific to your business. To see how AI can work alongside core business systems, read whether AI will replace ERP software.
Common AI implementation challenges and how to prepare for them
Buy-in can disappear quickly when actual results do not match expectations. Clear communication about how an AI project will unfold helps reduce that risk.
- Plan for a prototype, an MVP, and then production. A minimum viable product lets you test the core use case before a full rollout. Define the success criteria early so business and technical stakeholders agree on what “working” means.
- Treat go-live as a milestone, not the finish line. AI solutions often require continued evaluation and improvement. As your business, data, applications, or market changes, you may need to update instructions, data sources, integrations, evaluations, permissions, or, in some cases, model tuning or retraining.
- Watch for common causes of delay. Poor data quality, scope creep, unclear ownership, and unrealistic goals can slow an AI project before the technology itself becomes the issue.
- Expect small technical details to matter. We have also seen projects stall over surprisingly small technical details. An agent may not run because an application is one update behind or a required billing setting is not connected. These issues can take minutes to correct when you know where to look, but they can cost a team days of frustration when you do not.
A frustrating first experience can quickly turn a supporter into a skeptic, which is why testing and experienced implementation support matter.
Build your AI roadmap with confidence
Our facilitated workshop helps your leaders identify high-value AI use cases, address adoption barriers, prioritize opportunities, and leave with a practical action plan for moving forward.
Six AI adoption best practices for long-term buy-in
The crawl, walk, run model explains how AI capabilities can mature. Sustaining adoption also requires ongoing attention to people, processes, and governance.
1. Align on goals
Agree on what AI should achieve and how success will be measured. That might include saving employee time, reducing errors, improving response times, increasing capacity, or lowering operating costs. Without shared goals, teams may evaluate the same AI initiative differently. Effective AI change management helps keep people, processes, and expectations aligned.
2. Build a communication plan
Explain what will change, when it will happen, and what employees will need to do differently. Be clear about what AI can and cannot do, where human review remains important, and what outcomes are realistic. Visible leadership support also helps reinforce that AI adoption is a business priority rather than a short-lived technology experiment.
3. Create a feedback loop
Develop AI solutions in manageable increments and collect feedback from the people who use them. Employees can uncover confusing steps, missing information, weak instructions, or opportunities to improve the solution. Involving them also creates greater ownership of how AI fits into their work.
4. Invest in AI skills
Employees need to know how to use AI responsibly, not just how to access it. Training should cover giving clear instructions, evaluating outputs, protecting sensitive data, recognizing potential errors, and knowing when human review is required. It should also reflect how different roles use AI in practice.
5. Maintain data standards
Focus data management efforts on the information that supports your priority AI use cases. Define ownership, quality expectations, access rules, and processes for keeping important information current. You do not need perfect data everywhere, but the data supporting an AI-enabled process needs to be reliable enough for the intended use. For a primer, read what data governance is and why it matters.
6. Review governance and security regularly
AI governance should evolve as your use cases expand. Regularly review permissions, sensitive data access, approved tools, agent actions, and new security or compliance requirements. As adoption grows, some organizations formalize these responsibilities through an AI center of excellence or cross-functional governance team. This can also give employees an approved path for experimentation and help reduce the use of unapproved AI tools.
Why organizations choose Rand Group for AI adoption
Rand Group helps organizations move from AI interest to practical, measurable use cases. Our team uses AI in our own daily work, so our guidance comes from hands-on experience as well as technical expertise.
Our AI capabilities are strengthened by decades of ERP and CRM consulting experience across Microsoft, Oracle NetSuite, and Sage solutions. Because AI increasingly works within and across core business systems, our team can help organizations identify opportunities that connect AI with finance, operations, sales, service, and other business processes.
We also focus on the factors around the technology. Data readiness, governance, security, user adoption, and change management can have just as much influence on the outcome as the AI solution itself.
Rand Group has served more than 1,200 clients across North America, with a 90% client retention rate and more than 3,000 successful engagements.
Hear what our clients say
“Rand Group has become a resource we can lean on when we are trying to improve or digitize a process. Because of the time spent with our business, the team can go beyond the superficial layers and help us get to solutions that have a material impact on operations.” – Scott Rolston, President, Stucchi USA
“Everything we asked, Rand Group was able to accommodate. The process was extremely organized, the implementation plan was well thought out.” – Sam Rentz, Corporate Controller, Dairy Products Inc.
Key takeaways
- AI adoption challenges often involve business processes, data, governance, security, expectations, and people, not just technology.
- Different stakeholders have different concerns, so organizations should address the specific objection instead of using one message for everyone.
- A strong AI business case connects a specific use case to a measurable operational or financial outcome.
- Data readiness can affect AI cost, reliability, and the ability to scale a use case.
- A crawl, walk, run approach lets organizations increase AI capabilities as their experience, trust, and governance mature.
- Change management, communication, AI literacy, feedback, and ongoing governance help sustain buy-in after launch.
Frequently asked questions about AI adoption challenges
What are the biggest AI adoption challenges?
The biggest AI adoption challenges often include unclear ROI, poor data quality, security and compliance concerns, weak governance, unrealistic expectations, and low employee trust. Organizations may also struggle when they treat AI as a traditional technology rollout instead of preparing for the process and workforce changes that come with it.
How do you get executive buy-in for AI?
Tie the AI use case to a specific business objective and define how you will measure success. Start with a focused opportunity that can demonstrate measurable value without requiring a large enterprise-wide rollout. Be transparent about cost, risk, data requirements, and ongoing support so leadership can make an informed investment decision.
How do you overcome employee resistance to AI?
Give employees practical experience with AI in lower-risk scenarios and teach them how to evaluate the results. Be clear about which tasks AI may change, where employees remain accountable, and which skills they may need. Involve users in feedback and improvement so they have a role in shaping how AI fits into their work.
What is an AI adoption strategy?
An AI adoption strategy is a plan for introducing and expanding AI based on your organization’s business goals and readiness. It prioritizes use cases, defines success measures, and addresses data, security, governance, skills, and change management. A staged approach can help organizations build confidence before moving to more complex AI use cases.
How do you measure the ROI of AI?
Start with a specific business outcome, such as hours saved, lower processing costs, faster cycle times, increased capacity, fewer errors, or improved customer response time. Establish a baseline before implementation, measure the result after launch, and include both implementation and ongoing operating costs when evaluating the return.
What is the difference between AI and automation?
Traditional automation follows predefined rules and works well for predictable workflows. AI can interpret unstructured information, generate content, recognize patterns, or respond to changing context. Use automation when the process follows consistent rules and consider AI when the work requires interpretation or flexibility.
How long does it take to see value from AI?
Organizations can sometimes see value from simple, ready-to-use AI scenarios within weeks, especially when the tool, data, and security requirements are already in place. Integrated or custom AI projects usually require more planning, testing, and iteration. The timeline depends on the use case, data readiness, integrations, risk, and scope.
Overcome AI adoption challenges with a practical strategy
AI adoption challenges are not a reason to avoid AI, but they are a reason to approach it deliberately. Cost, data readiness, security, reliability, and employee concerns all deserve clear answers before an organization attempts to scale.
Start with a business problem you can define and measure. Put the right guardrails around the use case, involve the people who will use it, and expand based on evidence rather than enthusiasm alone. A practical AI adoption strategy can turn early experimentation into sustainable business value. If you are ready to identify the right place to start, contact Rand Group to discuss your AI goals.
