Boris Kontsevoi is a technology executive, President and CEO of Intetics Inc., a global software engineering and data processing company.
Covid-era digitization moved faster than anyone expected. According to research from McKinsey, surveyed companies “accelerated the digitization of their customer and supply-chain interactions and of their internal operations by three to four years” in just a few months during the pandemic.
This left many midsize companies with systems to stabilize, data to reconcile and processes to simplify. Even now, years of technology investments still await measurable returns. Generative AI was expected to become the next major technology wave, but it arrived before many organizations had realized returns from their earlier transformation efforts.
Yet, AI adoption does not automatically create business impact. McKinsey reports that only 39% of organizations report enterprise-level EBIT impact from AI, so many initiatives still need to prove their financial contribution beyond usage and experimentation. In the middle market, RSM reports that 92% of companies using generative AI face rollout challenges with data quality, security, governance and internal skills.
For midsize businesses, access to AI tools is no longer the limiting factor. The challenge is deciding which business problems should be prioritized first so that AI delivers measurable revenue growth, productivity gains, operational visibility or risk reduction within months rather than years.
How To Prioritize
Prioritization should begin with the places where the business already feels pressure: lost sales opportunities, margin leakage, customer churn, manual work, operational delays, unreliable data, dependence on key employees, technology drag or rising compliance risk. These are better AI starting points than broad transformation programs because the business issue is already visible, accountability is easier to define and progress can be measured.
For midsize businesses, your first AI investments should improve work that already exists: make it faster, more accurate, more predictable or less dependent on scarce resources.
10 Areas Where AI Can Deliver The Greatest ROI
Not sure where to start? Consider the following places where I’ve found AI can make the greatest impact:
1. Revenue Expansion Opportunities: AI can analyze CRM history, account behavior, win/loss patterns and buyer intent to help sales teams prioritize the right accounts, prepare proposals faster and follow up at the right time. Agentic AI can update CRM records, monitor pipeline movement and trigger next-best actions with oversight. The value here is higher conversion, shorter sales cycles and less time wasted on low-probability opportunities.
2. Margin Protection Initiatives: Many midsize companies I’ve observed lose profit not because they lack demand but because quoting, discounting and cost visibility are inconsistent. AI can flag margin leakage, recommend price corridors, identify excessive discounting and connect pricing to inventory, demand and customer history. This makes AI a margin-management tool, especially where small pricing improvements across many transactions can materially affect profitability.
3. Customer Retention And Service Excellence: AI can identify churn signals, summarize account history, prioritize service cases and recommend retention actions before customers leave. In service operations, agents can resolve routine requests, prepare responses and route exceptions. I’ve found that the immediate opportunity here can lead to fewer escalations, faster resolution and better protection of recurring revenue.
4. Document-Intensive Operations: Invoices, purchase orders, contracts, claims, onboarding files and compliance forms still consume time. AI can extract data, validate fields, compare documents, route approvals and identify exceptions. This is a clear capacity use case that can lead to fewer handoffs, less rework, faster cycle times and more output.
5. Operational Visibility And Process Efficiency: Many post-Covid systems produce data but not always decisions. AI can connect ERP, warehouse, manufacturing, service and finance signals to highlight delays, inventory imbalances, late orders, quality deviations or workload spikes. This can give managers earlier visibility into exceptions before they affect customers or margins.
6. Industrial And Manufacturing Intelligence: AI can support predictive maintenance, quality prediction, anomaly detection, production scheduling and digital work instructions. The value is measurable through reduced downtime, lower scrap rates, improved throughput, fewer warranty claims and better labor utilization. The earliest gains can come from preventing avoidable disruptions and improving production consistency.
7. Trusted Data And Executive Visibility: AI cannot produce reliable recommendations from duplicated, incomplete or disconnected data. A practical program should strengthen dashboards, data quality, metric definitions and cross-functional reporting. Executives cannot optimize what they cannot measure, and AI cannot reason from data the business does not trust.
8. Knowledge Access And Workforce Productivity: Midsize businesses often rely on informal knowledge held by senior employees, engineers, service managers or finance specialists. AI knowledge assistants can search policies, SOPs, product documents, tickets and project histories to find accurate answers faster. I’ve found that this can lead to less dependency on scarce experts, faster onboarding and fewer delays.
9. Technology Modernization And IT Productivity: Modernization should start where legacy systems are limiting margins: manual reconciliations, slow reporting, unsupported platforms, duplicated work, integration gaps or security exposure. AI can assist with code documentation, test generation, incident triage, log analysis and migration planning. This can reduce support burden and spur faster delivery.
10. Compliance, Cybersecurity And Governance: Compliance, cybersecurity and governance are important because the cost of failure often exceeds the cost of prevention. AI can classify sensitive data, monitor anomalies, support audits, check policy adherence and summarize regulatory obligations. For agentic AI, this must also include guardrails: defined identities, clear approvals and human oversight.
Conclusion
Midsize businesses do not need to copy global enterprises. They can win by being more disciplined: choosing fewer AI use cases, tying them to measurable outcomes and expanding only after value is proven.
In most cases, your biggest problems should already be visible in your organization’s daily operating reality. The real work is deciding which of those problems are ready for AI, which require process or data preparation first and which can produce measurable results fastest.
AI prioritization is not a technology roadmap; it is a management discipline. The companies that capture returns first must deploy AI where the economics are already clear and the work already matters, and they can’t hesitate to bring in specialized expertise when it can reduce risk, accelerate execution and improve outcomes.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
