For small businesses now working to define expected outcomes for AI, the fix is not to cut AI adoption or planned uses, but to focus tightly on their why. When AI is layered on top of existing workflows as a general productivity enhancement, token spend has no anchor; however, when it's embedded in specific processes tied to defined and measurable outcomes, the value can be more closely connected down the line.
SMB leaders should look at AI's value (and the value of any technology) by measuring its return like any other capital allocation: Connect the dots through time saved, better customer and employee experiences or new revenue. If those markers aren't moved, it may not be worth the investment. Achieving savings within any of those areas, whether through economies of scale or by lowering the costs of specific budget lines, enhances the overall ROI and improves time to value. To that end, managing token costs when deploying AI is an essential piece the ROI puzzle.
DISCOVER: Microsoft 365 Copilot solutions can help you leverage the right technology for optimization.
AI Projects Must Be Tuned to SMBs' Purpose
Any business today can adopt a plug-and-play agent or off-the-shelf AI resources, but that doesn't mean they should. Fine-tuning AI models, agents and data to meet and advance the highly specific needs of a business's core value and mission will help SMBs to stand apart, and differentiate themselves against a crowded competitive landscape.
Whenever and wherever agents and AI models are deployed, they should work in lockstep with the company's purpose and culture, and augment the organization's people to advance product or service innovations, improve productivity and efficiency, and enhance value to customers. Has your team placed measurable targets on those outcomes? Start there to understand how AI has moved the needle, or where it can help teams gain ground faster as new AI initiatives come online.
79%
The share of executives who expect AI to drive significant revenue by 2030
Source: IBM, "The Enterprise in 2030," Jan. 16, 2026
How Copilot Cowork Helps SMBs Optimize AI Costs
In June, Microsoft announced metered Copilot Cowork pricing, which offers a huge opportunity for SMBs to optimize their AI costs while also unlocking real value. From its inception, Cowork offered a lower cost alternative for AI adoption, from a runtime that efficiently finds the right information and tools, model choice that matches the right model for each task, and billing that charges businesses only for what they use.
Under new pricing announced in June, Cowork 1 is expected to handle tasks at substantially lower costs than its competitors, and variable pricing models that meet different use cases or scenarios. Copilot Cowork requires a Microsoft 365 Copilot User Subscription License, billing customers through Copilot Credits on a usage-based model determined by tasks run. The price of each task is calculated from four inputs: model use, context retrieval, tool calls and runtime.
DIVE DEEPER: Learn what you'll need to build a foundation for scalable AI.
Microsoft has helpfully defined Cowork task types to help businesses improve AI budgeting given the variable model. Light, medium and heavy tasks have been assigned estimated Copilot credits, 100–300, 300–700 or more than 700, respectively. When combined with four typical user personas — corporate knowledge workers, management and senior leaders, customer-facing knowledge workers or technical workers — applying estimated prices per prompt helps SMBs achieve a flexible cost model against which they can estimate (and refine) expected AI costs over time.
As SMBs accelerate AI adoption, doing the important work now to understand precisely where and how it will add value is essential to delivering expected returns on any AI investment, while ensuring customers — the lifeblood of SMBs' value — won't be left behind.
