Build your AI strategy in four decisions

Define the business result you want to achieve. “Improve productivity” may describe the ambition, but it isn’t specific enough to guide investment or measure progress.

Be specific about what should change, who should benefit, and when you expect to see results. For example, rather than setting a broad productivity goal, you might aim to reduce the time it takes the enterprise sales team to produce a customer proposal from nine days to four by the end of the quarter.

As you define the outcome, consider three questions:

• What changes? Use a metric your organization already recognizes, such as cycle time, cost, revenue, retention, or quality.
• For whom? Name a specific team or function, not the organization as a whole.
• Who sponsors it? Name an executive who can remove barriers and keep the work connected to business priorities.

Next, determine where to focus. Look for the intersection of business impact and readiness, using the outcome you identified.

Which departments are closest to the work that needs to change? If the goal is to reduce proposal cycle time, for example, sales may be a natural candidate.

Next, assess readiness. Look for teams with leadership support, capacity to participate, and access to the information and tools they need to get started. A department that’s navigating a major change—like a reorganization—probably isn’t the best candidate for a first wave, even if it offers significant potential.

Impact and readiness won’t always point to the same place. In that case, decide whether to move forward with a team that’s ready now but may not have the greatest impact, or invest in getting a higher-impact team ready.

Select one department for your first wave so you have a practical starting point to learn from.

With a department in mind, identify the specific workflows where AI could make a difference.

Focus on recurring workflows that people can clearly outline from start to finish and where Copilot has the potential to improve business results. Rather than trying to change everything at once, prioritize a few workflows as candidates for your first wave.

For inspiration, explore the Microsoft Scenario Library to see AI scenarios by function and industry.

Evaluate each candidate against four criteria:

• Business impact: Improving the workflow contributes to the business outcome you identified.
• A named owner: A person close to the work is willing to own the workflow and its results.
• Measurability: You can establish a baseline and set a clear target for improvement.
• Readiness: The team has what it needs to begin testing the workflow within the next few weeks.

If a workflow doesn’t meet those criteria, save it for a later wave. Aim for two or three workflows so you can gather meaningful evidence about what works before expanding the scope.

Usage can tell you whether people are adopting Copilot, but it doesn’t tell you whether that adoption is improving business outcomes.

Before you begin the pilot, select a metric for each workflow that reflects the result you want to improve, such as cycle time, cost, completion rate, or quality. Then define four things for each:

• Baseline: Where the metric stands today
• Target: Where you want the metric to be in 90 days
• Owner: Who will track and report progress
• Review date: When you’ll evaluate the results

Also decide in advance what you’ll do if a workflow doesn’t meet its target. You might refine the approach and try again or stop and prioritize another workflow. Setting those expectations up front gives you a clearer basis for deciding what to do next.