Generative AI ROI in SMBs: How to Measure Real Gains
Between spectacular promises and scepticism, how can you concretely evaluate what AI brings to your SMB? A simple method and realistic orders of magnitude.
Studies on generative AI productivity announce gains ranging from 10% to 40% depending on the profession. These aggregated figures are of little use to an SMB leader: what matters is the measurable gain in your own processes, with your teams. Good news: the calculation is simpler than it seems, provided you proceed methodically.
Start by identifying "high-leverage" tasks: frequent, time-consuming, and language-based. In most SMBs, this includes drafting emails and proposals, summarising documents and meetings, preparing marketing content, answering customer support tickets, and searching for information in internal documentation. List these tasks and estimate the weekly time spent on them by each role.
Then measure a before-and-after sample. Take five volunteer employees, three typical tasks, and track time for one week without AI, then one week with AI. Feedback from the field converges: drafting a first outline typically goes from 45 to 15 minutes, summarising a 20-page document from 40 to 10 minutes, and translating a page from 30 to 5 minutes with proofreading. The gain does not come from eliminating the task, but from shifting the effort towards proofreading and decision-making.
Translate these gains into value. An employee who saves 4 hours per week at an employer cost of 80 CHF per hour represents about 1,250 CHF of freed-up capacity per month. For a team of 20 people with a conservative average gain of 2 hours per week, the freed capacity exceeds 12,000 CHF per month, compared to the cost of an subscription of a few hundred francs.
Beware, however, of ghost gains: freed time is only valuable if it is reinvested. SMBs that get the most out of AI set explicit objectives: more proposals sent per week, reduced customer response times, or documentation finally kept up to date. Tracking usage statistics also helps identify teams that have not yet adopted the tool and target training.
The final element of the calculation, often forgotten, is the cost of risk. Using personal accounts on public services exposes the company to data leaks whose potential cost (loss of clients, reputational damage, regulatory fines) is orders of magnitude greater than the savings made. The ROI of an enterprise solution is therefore also calculated in risks avoided, not just hours saved.