← Chat Archive

I'll fetch signalpost.net to review its content.

The initial fetch only surfaced a bare-bones headline/tagline — likely because signalpost.net is a JS-rendered single-page app and the fetcher only saw the initial HTML shell. Let me try pulling more of the actual page content and check common subpages.

The fetch tool is only seeing the `<title>` tag — the actual content is likely client-side rendered (JS) and not present in the raw HTML. Let me check the raw HTML directly via curl.

That confirms it — SignalPost is a client-rendered SPA (React/Vite), so the raw HTML is empty except for `#root`. I got useful meta tags already; let me check if a headless browser is available to render the real page, and also mine the JS bundle for extra copy.

Playwright is available. Let me use it to render the actual page and capture the full text content.

Chromium installed. Now let's write a script to render the page and capture text, screenshots, and any linked routes.

Now I have the full page content. Let me check the screenshot and look at what other pages exist (About, Careers, Contact, etc.).

Let me quickly check the /pricing and /about subpages for additional detail beyond what's on the homepage.

I rendered the full site (it's a client-side React SPA, so a plain fetch only sees the empty shell). Here's the detailed breakdown. ## What SignalPost Is A LinkedIn writing tool aimed at **founders, PMs, and senior practitioners** — not general creators. Positioning: "Write posts that sound like *you*, not AI." The workflow is Voice Profile → Signal Search → Draft (7-dimension strength map) → Publish/copy. Pricing: $0 trial (3 generations, 10 signal searches) vs. $29/mo Pro (unlimited + performance insights + teaching graphics). ## Strengths - **Sharp, specific ICP.** "Founders & PMs" repeated in the eyebrow tag on every page — good discipline, not trying to be everything to everyone. - **Concrete mechanism, not vague AI-magic.** "7-dimension strength map," "V1/V2 versions," named steps (voice profile → signal search → write hub → editor) make the product feel real and demoable rather than a wrapper. - **Good anti-positioning copy.** The "Generic AI draft vs. SignalPost draft" comparison on the homepage is the strongest piece of content on the site — it shows rather than tells, and the fake-emoji-thought-leadership example is instantly recognizable as the thing everyone's tired of. - **Trust signals up front:** "No fake testimonials," "No credit card," "You own your content," "No auto-publish" — smart, since AI ghostwriting tools face real skepticism. - **Consistent voice across pages** (About, How it Works, Pricing) — same tone, no drift. - **Honest pricing framing:** "$2.42 cost per post at 12/month" is a nice anchor that reframes $29 as cheap relative to output. ## Issues & Gaps 1. **Broken Google Fonts load — real bug.** The page's own Content-Security-Policy (`style-src 'self' 'unsafe-inline'`) blocks its own `<link>` to `fonts.googleapis.com`, so IBM Plex Sans / Source Serif 4 silently fail to load and the whole site falls back to system fonts. This is worth fixing — either add `style-src-elem https://fonts.googleapis.com` to the CSP, or self-host the fonts (recommended anyway for performance). 2. **No social proof at all.** The site explicitly brags about having "no fake testimonials," which is a fine differentiator, but it leaves zero third-party validation — no logos, no user count, no "used by X posts generated," nothing. For a $29/mo B2B-ish tool, some real proof (even "127 professionals in trial," a founder's own LinkedIn post using the tool) would help conversion without contradicting the "no fake testimonials" stance. 3. **"Illustrative examples" disclaimer undercuts its best asset.** The signal-vs-noise comparison is compelling, but the caption "Illustrative examples — not user testimonials" reminds the visitor it's not real output from a real account, weakening the very proof point it's trying to make. 4. **Product screenshot is a mockup, not a real screenshot.** Worth confirming — if it *is* a live product screenshot, label it that way for credibility; if not, same issue as above. 5. **Pricing has only one paid tier.** Fine for simplicity, but there's no team/agency tier and no annual discount — likely leaves money on the table if any users are ghostwriters managing multiple executives' voices (a natural upsell for this exact product). 6. **FAQ section on pricing page is a list of questions with no visible answers in the rendered text** ("How does the trial work?", "Can I upgrade anytime?" etc. appear as bare question strings with no answer copy following them) — worth checking whether these are supposed to be expandable accordions that just weren't triggered, or whether answers are actually missing. 7. **No blog/content marketing surface.** For a product literally about LinkedIn thought leadership, there's no visible content hub demonstrating the founders using their own tool — a missed "eat your own dog food" opportunity. 8. **Nav has no dedicated "Product" page** — clicking "Product" in the nav just goes to the homepage (`/`), which is a bit of an odd affordance; consider either removing it as a nav item or pointing it somewhere more specific (e.g., an anchor to the workflow section). ## Bottom line Clear positioning, disciplined copy, and a genuinely differentiated mechanism (voice profile + ranked signals + scored draft) instead of a bare prompt box. The main gaps are trust/proof (no real testimonials or usage numbers) and one concrete technical bug (CSP blocking its own web fonts). Fixing the font CSP is a quick win; adding even minimal real social proof would meaningfully strengthen the pitch given how proof-focused the rest of the copy already is.