Changelog

New activated users now see the founder setup offer reframed as "$60 off your first month" ($9 today, then $69/mo) right when they finish onboarding, and again in a one-time modal on the Articles page after they open their first article — the two highest-intent moments in the activation funnel.

Created a local-only Batch 012 challenger set to test whether Meta's free-calendar angle is selecting free-resource downloaders instead of content-led buyers with stronger product intent.

The Founder Setup bridge now sells the honest $60 first-month discount with founder setup included, adds a secondary onboarding-completion offer, and shows a one-time front-and-center modal after the first article is viewed.

Added concrete founder follow-up angles for the latest SEOTakeoff paid leads, allowed manually verified SITE_ANGLES contexts to avoid generic fallback copy on typo/redirect domains, and verified the recent no-touch queue is clear after sending the current clean T0 follow-ups.

SEOTakeoff's paid-lead system now separates guarded attempts from truly due work, blocks low-quality founder drafts with a hard copy validator, classifies paid-lead replies with SLA flags, runs a weekly cohort report, and has an approved live guarded founder autosend wrapper for clean known-site T0/T+3/T+7 follow-ups.

SEOTakeoff founder T0, T+3, and T+7 follow-up templates now end with a buyer-vs-resource-download question and a concrete next step to turn the prospect's sketch into a working SEOTakeoff calendar. The CRM now also marks replied threads as handled/waiting once a later founder reply is sent.

Restored Google Workspace auth, caught up missed clean SEOTakeoff founder follow-ups with Gmail duplicate guards, added a quiet pre-expiry auth watchdog, corrected a Dreamcave Studio site-personalization mismatch in the paid-lead CRM, and hardened outbound founder outreach against unresolved domains.

SEOTakeoff now has a generated internal ad library and competitor comparison layer that joins local creative specs, draft manifests, live experiment-ledger metrics, and the latest competitor sweep into tested/drafted/untested hook intelligence.

The weekly SEOTakeoff Meta competitor sweep now turns Ad Library intake into repo-backed markdown and JSONL artifacts with observed hooks, hook families, offers, formats, landing URLs, longevity proxy, tested-by-us inference, and suggested SEOTakeoff tests.

Documented the next SEOTakeoff paid-conversion experiment: a proof-backed $9 first-month subscription bridge with a founder setup call at the post-first-article / activated-user moment, including the existing activated backlog, renewal disclosure, matched founder follow-up, lifecycle support, spend guardrails, and measurement criteria.

SEOTakeoff now has a local paid-lead CRM that consolidates founder email state, product nurture state, reply review, LinkedIn/manual status, and site-aware founder outreach drafts behind an approval gate.

Acquisition nurture now extends beyond the initial three-email lead-magnet flow with proof, urgency, and a $9 intro-offer bridge, while founder-led sends have a shared state ledger for quiet-window guards.

Activated trial users now see a clearer $9 Starter upgrade path, and SEOTakeoff records billing-tab intent separately from Stripe checkout intent for the next paid-funnel diagnosis.

SEOTakeoff now gives new users a clearer first-article action and records more reliable diagnostics across generation, upgrade, auth, and keyword persistence failures.

Checkout-start conversion tracking now fires only after SEOTakeoff successfully creates a Stripe Checkout session, keeping Google Ads and marketing events tied to the real checkout-start action instead of signup or CTA intent.

The paid funnel now preserves Google click IDs and can fire Google Ads conversion events for the first paid-search test once account IDs and labels are configured.

Calendar-offer lead captures now preserve website and capture context through auth and onboarding so paid-lead follow-up and funnel debugging can distinguish signup, redirect, onboarding, and tenant creation states.

Calendar-page lead captures now use a calendar-specific lead-magnet label instead of inheriting the SEO Agent Playbook label.

The paid calendar landing page now captures a website URL alongside email so warm Meta visitors can request a site-specific 30-day SEO calendar before account creation.

Paid calendar-page captures now enter a calendar-specific acquisition email path instead of being skipped by the older playbook-only delivery logic.

The paid calendar landing hero now uses a lower-friction lead CTA and first-screen Search Console proof instead of cooling the page with disclaimer copy.

Paid campaign traffic can now land on a content-calendar offer that matches the new creative direction and routes captured leads toward the signup calendar flow.

CI migration checks are back in sync after aligning the tenant topic blocklist migration filename with the version already recorded in production.

CI now gives production deploys a short grace window before failing route smoke checks, reducing false alarms from Vercel deploy/cache lag while still catching persistent 404s and server errors.

Users without a tenant (pre-onboarding) no longer trigger a 404 error when visiting Settings. The API Keys tab now guards both rendering and data fetching until tenant data exists.

Seven lifecycle emails now nudge you through the first article experience: welcome, nudge to generate, first-article-ready, nudge to publish, first-publish celebration, nudge to second article, and a heads-up when free articles run out. All sends are signed with unsubscribe tokens and respect your preferences.

New two-layer validation pipeline catches bad hero images before they ship: heuristic checks first, then a GPT-4o vision pass (shadow mode) to flag irrelevant or off-brand images. Clear skip reasons logged for every validation so regressions are observable.

Transient backend hiccups (pool saturation, upstream timeouts) now surface as 503s instead of 500s so the frontend can retry with exponential backoff. React Query retries only on 502/503/504 (never on mutations), so you don't get the same failed action repeated.

New post-publication link auditor keeps your article links healthy. Tracks external, internal, and YouTube links with a three-state health model (healthy → suspect → broken) and auto-repairs broken links via SERP replacements or on-site title search.

Fixed request timeout errors when generating article headlines. The initial POST request now has a 60s timeout (up from 30s default) with automatic retry logic to handle Railway cold starts.

New SSR-safe component framework for automatic content quality upgrades on alternatives, comparison, and guide pages. All new sections render in initial HTML for crawler visibility.

Fixed remaining SEO gaps in tool page templates: removed animate-pulse skeleton markers from HydrateFallback (crawlers now see static content), added twitter:card meta tags for consistent social sharing.

Comprehensive fix for all SEO-related link issues across templates. Tool pages now have data-driven related tools sections. Compare pages show dynamic comparisons from CMS. Alternatives pages include contextual cross-links. Blog pages have robust fallback logic for related posts.

Hardened two SEO-critical routes to reduce intermittent crawler failures: /signup is now eagerly loaded, blog article pages gracefully return a noindex fallback when CMS data is missing/unavailable, and /api/health?seo=1 adds bot-style checks for /signup and /blog/ai-seo-vs-chatgpt.

Keywords now show real search volume and difficulty data instead of estimates. Enrichment happens automatically during onboarding and cluster generation, with cost controls and graceful fallbacks.

Long article lists (Upcoming, Completed) now collapse to show 10 items by default. Click the header or 'Show all' to expand. No more scrolling past 700 articles.

LSI Keyword Generator, Related Keywords Finder, and Long-Tail Keyword Generator now pull real suggestions from Google instead of static patterns. External Link Checker now actually verifies URLs server-side.

Image generation now produces more creative, editorial-quality visuals instead of literal stock-photo depictions. No more random animals or vehicles in unrelated articles.

Content gaps now adapt to your site's traffic profile. Instead of fixed thresholds, we use your median impressions and average CTR to find the real opportunities — works for sites of any size.

Lazy-loaded monitoring scripts (saves ~86KB upfront), compressed API responses with GZip, added browser caching on analytics endpoints, and lazy-loaded landing page images.

Fixed a bug where article performance cards showed 'No GSC data yet' for all articles. The query was only fetching 10 pages from Google Search Console — now fetches up to 5,000.

3 free articles, no credit card required — now prominently shown in the hero, pricing section, and bottom CTA. No more guessing whether there's a free tier.

New fields in Settings let you list your product's key features and important URLs. These get injected into article generation prompts so your content naturally references what you actually offer.

Site analysis previously only read the top-level sitemap index. It now recursively fetches sub-sitemaps, so all your URLs are discovered during onboarding.

Fixed an issue where AI-generated articles sometimes included bare URLs instead of properly formatted markdown links. Links are now wrapped with descriptive anchor text.

See which pages have high impressions but low click-through rates. Track AI citation appearances. View traffic breakdown by device type and country.

Export your analytics data as CSV or PDF. Quick-select date ranges. Smoother loading experience throughout.

Paste URLs to your existing articles and SEOTakeoff analyzes your writing style. Future articles automatically match your voice, tone, and patterns.

You can now sign in with your Google account — one click, no password needed. Works alongside email/password for anyone who prefers it.

Previously, your GSC connection would silently expire every week and require manual reconnection. That's fixed — your connection now persists indefinitely.

You can now rate and leave comments on individual sections of any article, as well as give overall article-level feedback. This helps us track what's working in generated content and improve quality over time.

We've been shipping a lot and not saying much about it. That changes today. This page is a running record of everything we ship — features, improvements, and fixes — going back to launch.

When importing headlines via CSV, SEOTakeoff now:

• Automatically creates any keyword clusters that don't exist yet
• Skips headlines you've already added (no more duplicate errors)
• Enriches cluster keywords with real volume and difficulty data on import

Added higher-frequency autopilot options for high-volume plans — 50/week and 100/week. Articles are distributed evenly across your publishing schedule.

Error messages across headlines, clusters, settings, and publishing now show the actual reason something failed (e.g., "Duplicate headline detected") instead of a generic "Failed" toast.

Headline generation now runs in parallel batches, cutting wait time significantly on larger keyword clusters.

Added a Manage Billing button in Settings that opens your Stripe billing portal directly — update your payment method, download invoices, or cancel from there.

Translation jobs now appear in the Articles list with progress indicators. Translated articles auto-publish when complete and get native-language titles and slugs (not translated English ones).

Also added a keyword research preview before translation kicks off, so you can see what keywords will be targeted in the target language.

You can now delete articles from the article editor. All related records (keywords, queue entries, revision history) are cleaned up automatically.

The onboarding flow now searches the web to find your actual competitors instead of relying on guesses. Competitors are validated before being shown to you, so you only see relevant results.

When publishing to your CMS, you can now choose whether articles go out as drafts (for review) or are published immediately. Unified across all supported CMS platforms.

Added a theme toggle to the app header. Defaults to your system preference.

SEOTakeoff is now open for self-serve signups — no waitlist, no invite needed. Cloudflare Turnstile bot protection is in place to keep spam accounts out. Password auth also added as an alternative to magic links.

New Quick Wins section on the Strategy page highlights the easiest fixes for a traffic boost:

• Articles with outdated years in the title (auto-update available)
• Keyword cannibalization — multiple articles competing for the same keyword
• Dynamic competitor tracking updates as your landscape changes

Several improvements to how autopilot schedules articles:

• Skip weekends toggle — keeps your publishing cadence to business days
• 14 articles/week frequency option added
• Randomized publish times — articles no longer always go out at midnight
• Auto-rescheduling — when you change your frequency, future articles are redistributed automatically

Keyword cards now show real monthly search volume and keyword difficulty scores powered by DataForSEO — replacing the placeholder estimates that were there before. You can also manually refresh metrics on any cluster.

The SEOTakeoff blog is now live with articles on SEO strategy, content marketing, AI content, and how to use the platform effectively.

Autopilot can now automatically publish completed articles to your connected CMS — no manual intervention needed. Articles move from generated → published on their scheduled date without you touching anything.

A floating toolbar now appears when you select text in the article editor, with quick access to bold, italic, heading levels, and link insertion. Keyboard shortcuts also supported.