Competitor analysis presentation in Canva, built by Claude from this week's brief

Competely's guide to doing this analysis in Claude gives the structure that works: a fixed set of headings, so every competitor lines up side by side. Its eight are marketing and messaging, product and features, pricing and sales, audience and customers, customer sentiment, company profile, market and recent moves, and a SWOT. It also insists on two rules that carry over to slides. Put the source URL in the same sentence as the claim, and write that no public source was found rather than guess.

Its checklist adds that features should be split into live today versus announced, and that recent moves should each carry a date. A deck that follows this reads as evidence. A deck that skips it reads as opinion.

Slides add one constraint the guides do not cover. A slide holds a claim and a source, not a paragraph of caveats. So decide up front which claims earn a slide: moves rated high or watch priority, a pricing change, a launch, a message shift. Leave low priority items in an appendix or out. The brief's own priority grading does that sorting for you.

Competely says pricing is the first thing to verify, because pricing pages change without notice, and that the output is a snapshot. A deck built from a one-off research prompt is true on the day it is made. Rivals keep moving after you file it.

That is the gap the Compete brief fills. On the test project, the brief for 26 October to 1 November covered 5 competitors and graded 19 moves: 2 high, 9 worth watching and 8 low. The two high priority moves were a Fathom launch of an AI assistant called Luna and Matomo's critical PHP security guidance. Each comes with a category, why it matters and a source link, so a slide can carry its evidence.

Linkeddit supplies the data and writes no deck. Claude reads the brief and chooses what to put on a slide, which keeps the facts traceable to a source and the judgement with you.

Canva's announcement says its Claude connector lets users apply their Brand Kit when generating designs, so presentations follow brand colours, fonts and voice from the first draft. Claude was the first assistant to support these on-brand capabilities, according to Canva. The connector listing shows 17 tools, 12 reads and 5 writes, including generate-design, export-design, get-design-content and add-comment-thread-to-design.

Write the outline in text first and design second. That way you correct a wrong claim in one place, before it is spread across fifteen slides. When the design exists, ask Claude to use get-design-content to read it back and check that every number and source link survived the layout.

Read the result before you share it. One viewer of a Claude and Canva automation video commented that great content cannot simply be automated, and for a deck that a sales team will quote, that caution is right.

A battlecard is the one-page version of the deck for one competitor. Use three blocks. The first is what they just did, taken from the brief's high and watch moves with the source link. The second is what their customers say, taken from the reviews Linkeddit has stored for that competitor. The third is how you answer, which is the part you and Claude write together.

Linkeddit stores reviews from G2, Capterra, TrustRadius and Trustpilot as the sites show them and does not summarise or score them. Ask Claude to call compete_signals_get for the reviews view, and to quote a line word for word with the review site named, never a paraphrase. Quote only text you have read in the stored review.

Keep each card to one competitor and one week. When the next brief lands, ask Claude to refresh the first block only. The battlecard stays short enough to use in a call, and it stays dated.