Google Gemini Advanced

Use 4-part prompt structure: Role and goal, Context, Constraints and format, Numbered tasks

Be explicit not clever: specify word counts, tone, audience, and output format upfront to avoid revision loops

Use Flash as default for everyday tasks, Pro only for the hardest strategic 5-10% requiring deep reasoning

For images and video, specify what to focus on: hook, visual hierarchy, call-to-action, specific patterns

Use thinking mode for complex reasoning: ask for visible steps, plan-then-act, or self-critique sections

Create custom Gems for recurring roles (copywriter, data assistant, designer) encoding your preferred formats

For long documents give jobs not just summarize: extract risks, score items, identify patterns in tables