Estimating project impact
Impact estimation offers two perspectives: evaluating impact of launched products and forecasting the potential of future projects. This post focuses on the latter – addressing the critical question, "Will it be worth it?".
Begin with a simple, back-of-the-envelope calculation to assess whether a project is worth pursuing. This preliminary assessment is sufficient for building a prototype. However a full-scale project requires a more thorough analysis which includes:
Data-driven decisions. Base your estimates on robust data. Collaborate with your analytics team or, if necessary, run the SQL queries on internal databases yourself. If internal data is insufficient, augment your research with secondary sources such as online reports or reputable agencies. AI chatbots like ChatGPT and Gemini can also provide valuable insights, especially when asked good questions.
Question assumptions. Explicitly state your assumptions – whether they concern market penetration rates, user psychographics, or product preferences. – and critically evaluate them with various experts. Ask probing questions like, “How risky is making this assumption?” and “What have I overlooked?” Remember Richard Feynman’s advice: "You must not fool yourself—and you are the easiest person to fool."
Comparative analysis. Examine comparable products, their monetization strategies, pricing model, and customer demographics. Ensure your comparisons are appropriate; avoid benchmarking against giants like Apple unless you are in direct competition. Use this analysis as both a guide and a check on your estimates.
Think in ranges. Present your estimate as a range, or better yet, multiple conditional estimates. For instance, “first year’s annualized revenue is forecasted between $4M and $6M, depending on focus demographic segments” highlights a condition, a range, a specific timeframe, and a metric.
Avoid linear thinking. Recognize that progress is not linear. The effort to acquire the first 100 customers is often greater than securing the next 100. There are multiple factors contributing to this kind of non-linear behavior – network effects, economies of scale, market saturation, technological advancements, and regulatory changes. When making multi-year projections, avoid simple linear extrapolations.
In the next discussion, I will explore strategies to get impact from ML projects earlier.