Becoming evidence-guided | Itamar Gilad (Gmail, YouTube, Microsoft)

Gilad's GIST model distinguishes what the team wants to achieve, possible ways to achieve it, steps that both build and validate, and delivery tasks. The separation helps avoid treating an idea as an outcome. It combines existing methods rather than claiming to replace every part of product work.

He says the process is the part most likely to need adjustment to a company. Strategic context sits outside GIST rather than being solved by it. Adopt the underlying questions without assuming one organization's routines can be copied unchanged.

Gilad recommends ICE: impact, confidence and ease. Impact relates to a clear goal, while ease considers how difficult the work is. Both begin as estimates. Making those estimates visible can improve a discussion that would otherwise be a competition between opinions.

He warns against giving an idea high confidence simply because someone believes in it. His confidence meter makes the supporting evidence explicit. An attractive deck, a competitor feature or a strong opinion should not silently become proof that the idea will work.

Gilad starts with cheaper learning and invests more when evidence supports it. He also notes that some low-risk changes do not need an elaborate test. The amount of validation should fit the decision, rather than becoming a ritual applied equally to every task.

He contrasts speed of putting code into production with speed of creating the needed impact. Learning and building can happen together. This makes discovery part of delivery rather than a separate delay, while leaving room to evaluate whether the work actually helps.

Gilad advises against adopting the entire system at once if that creates too much process and fatigue. A team can begin with one tool and learn from it. Choose a small practice that makes uncertainty clearer instead of launching a large transformation solely for its own sake.