FigJam competitive analysis: fill the board from this week's competitor data

A FigJam competitive analysis is a whiteboard that compares your product with rivals, and the Figma connector is Figma's official MCP server at https://mcp.figma.com/mcp, which lets Claude read and draw FigJam boards. The pages that rank for this query are mostly templates. Figma's own FigJam competitor analysis template describes laying out competitors' strengths and weaknesses next to yours, then taking what works and leaving what does not. Figma's community has more: a design systems version that compares several systems by category, and a general one that suggests clustering competitors by characteristics, strengths and weaknesses and sorting findings with a Competitive Testing worksheet.

Panoramata's free Figma template goes furthest on facts. It covers acquisition channels (paid, organic, SEO, email marketing), plus pricing and reviews, and says each competitor takes about a minute to fill. That estimate is Panoramata's own, and it is the part that does not repeat. The grid is a one-time job. The data inside it goes stale as soon as a competitor ships something.

So the board is easy and the evidence is hard. A template cannot tell you that a competitor launched an assistant this week or published a security notice, and it cannot cite where it learned that.

Figma's help centre says its generate_diagram tool builds a FigJam diagram from a description or from Mermaid syntax, and lists the supported types: flowcharts, Gantt charts, state diagrams, sequence diagrams, architecture diagrams and entity relationship diagrams. It does not list a matrix or a table, so the classic strengths and weaknesses grid is not something this tool is documented to draw.

That shapes the plan. A Gantt chart suits a timeline of competitor launches by week. A flowchart suits a buyer's path from a trigger, such as a pricing change, to a switch. An architecture diagram suits a map of which competitor covers which part of your stack. Use the template for the grid and generate_diagram for the diagrams, and keep both fed from the same brief.

Figma also advises including the phrase using the Figma MCP generate_diagram tool in your prompt, because the agent usually picks it up on its own but the phrase makes sure. The Figma connector's listing shows 9 tools, 6 reads and 2 writes, so most of what it does is read the board back, which is how step five checks the result.

Compete produces one weekly brief across every competitor you track. The run above shows its shape: a window, a count of competitors monitored, signal counts by priority and then each signal with a category, a reason it matters and a source link. In that week the brief carried 19 signals for 5 competitors, 2 of them high priority.

Claude should put only those signals on the board and carry each source link into the node text or a sticky note. Then a teammate who doubts a node can click through to the changelog or blog post it came from.

Linkeddit does not write the analysis. It supplies the signals, and your own Claude decides what they mean for the board. The brief's why it matters lines are written for the tracked project's own team, so read them before you share the board with people outside it.

A board of competitor names is not the goal. A Claude user on Reddit put it this way: "Real competitive intelligence is not a list of names." The data from Linkeddit is what turns the board from a list into something with evidence behind each card.

The first problem is silent gaps. In the run above, plausible.io shows status pending with a count of 0, and the other four competitors show page_monitor:no_results. A board with five tidy columns would hide that. Make the prompt ask Claude to report each competitor's status and mark the thin ones on the board.

The second is drift between data and drawing. A diagram step can shorten text. That is why the prompt reads the board back with get_figjam and compares it with the brief.

The third is stale cells. A moves board is a snapshot of a window, so put the window dates on it, here 2026-10-26 to 2026-11-01, and rerun when the next brief lands rather than editing nodes by hand.