Card 7: "ChatGPT or Claude, and nothing else" · 09 Battlecards

Their pitch: cheap, already open, and it writes a decent post. One of the two alternatives we expect to meet most often. Do not argue about writing quality in the abstract: show it.
Our edge: everything after the draft: each person's voice, rules checked, approvals, one calendar, publishing, results per person, and Holly doing the work.
Ask: "Who checked the last AI-drafted post against your policy before it went out?"
Where it wins: cheap, flexible, already on screen, good at a first draft, no procurement.
Where we win (GREEN): voices per person, kept distinct for twenty people; rules checked, approvals, one calendar, publishing with an exact preview, testing, results per person in the app (reports sent outside Maktura are AMBER until D1); ideas, not a blank box (news feeds, morning list, meeting ideas).
Evidence you may quote (say the source): LinkedIn launched a "seems like AI slop" report button on 30 July 2026; over a million members used it within about three weeks, and flagged content sees roughly 40% fewer views (press reports of LinkedIn's figures: Moneywise, TechCrunch, Social Media Today). LinkedIn says refining your own words with AI is fine. For "we already pay for tools", the pilot testimonial from a freelance marketer (the only usable testimonial): "You end up with logins for so many pieces of software and you don't end up using them. This replaces the lot. And it does my least favourite job, end-of-month analytics, in a click." For you, not quotable yet: in September 2026 LinkedIn replaced its own write-from-scratch tool with a proofreader for drafts you have written (Metricool, 2 Oct 2026) [C].
The demonstration: ask for a topic. Ask them to paste a post they drafted with a general assistant, or generate one live. "That's a good draft. The trouble is it could be anyone's." Show the same topic drafted in a demo voice for someone in their role (their own voice only from posts they paste in on the call, with their consent, marked test first, Alex or John demoing; never connect their LinkedIn). Then: "Everybody has the writing now. The difference is that the second one is you, and the first one has no idea who read it, or whether it broke your policy."
Landmine: "If a client complained about a post, could you show who wrote and approved it?"
If they say:
"We'll just use ChatGPT." Say: "Lots of firms do, and for one person it's fine. Can I show you something on a topic you'd post about? Then judge for yourself."
"Isn't Maktura just Claude with a wrapper?" Say: "The writing does run on Anthropic's Claude, yes. The model is the engine. You're paying for everything around it: each person's voice, your rules checked, approvals, the calendar, and Holly doing the work."
"AI posts get punished on LinkedIn now." Say: "Generic ones can. Flagged posts see roughly 40% fewer views, according to press reports of LinkedIn's figures, and LinkedIn says refining your own words with AI is fine. That's the point of voices and approvals: your thinking, your words, and a person signs it off."
Never say: "Invisible to AI detection"; that we make posts sound "human"; that output "beats the algorithm". Do not lead with the model or put down the tools by name.
Switch story: "Keep using it for everything else." The standard switch kit, starting from their best recent posts. Regulated firms start at Observe so they see Holly's drafts first.
Reference: ChatGPT Plus $20, Business $25; Claude Pro $20, Team $25 a user a month; no native LinkedIn connection, scheduling or analytics (PostEverywhere, Oct 2026, not verified at the vendors) [C].