Outrank teardown: how it works, and what Fimo SEO + GEO do differently

A 5-step wizard runs after the $1 trial checkout: Business → Audience & Competitors → Blog → Articles → Integration. Outrank scrapes the site and pre-fills every step. In all three videos the user mostly clicks Continue.

Two inputs drive everything that follows. The competitor list (up to 7 domains) seeds keyword discovery: "bigger competitors provide more valuable insights". The 3 example article URLs set the writing style, which then shows as a greyed "Informative" or "Professional" label. Search Console is offered here only to "avoid suggesting keywords you already rank for", and every video skips it.

The Articles step is where the output gets its shape: auto-publish, internal links per article, global instructions, brand colour, image style (Brand & Text, Watercolor, Cinematic, Illustration, Sketch), title-based featured images, and toggles for YouTube embeds, a table of contents, CTAs, infographics, emojis, "mention similar products" and first-person writing.

After onboarding a modal says "Creating a Content Strategy for You · 4–5 minutes". In the April 2026 review it took 16–17 minutes, and the browser console filled with failed requests. Then "Your 1 month Content Strategy is ready" offers one button: Create & Publish First Article.

The Content Planner is the home screen. Each day holds one card: an article type (Guide: Explainer, Guide: How-to, List: Resources, List: Round-up), the keyword, Vol and Diff. Cards can be dragged. The Automatic Keyword Generation setting keeps 25–30 keywords queued, so the calendar never runs empty. The 2026 product adds a keyword-research chat that pulls live volume, KD and CPC, and lets you drag the results onto the calendar.

The article page shows the text, a featured image in the chosen style, the target keyword, the article type, the status, the slug and a meta description with a counter (157/160). Edit Article opens a rich-text editor. An AI Rewriter works on highlighted text, with presets (Shorten, Simplify, More detailed, More informative) or a free prompt. Articles export as Markdown.

Internal links come from Linking Configuration, which reads the sitemap (534 to 678 pages detected in the videos) with no filter, so Terms and Cookie pages count as link targets. The 2026 editor (marketing image) replaces the modal with a side chat and a Rewritten / Reject diff for each section.

Quality is the main complaint. The 3-month reviewer found no sources, no FAQ, an unattributed quote and an off-topic YouTube embed, and GPTZero rated the text 90% AI. A 60-article blogger test found that 12 of 60 articles needed heavy rewrites. The founder says the engine was rebuilt after "AI slop" complaints.

Articles go out at 7–9 AM UTC to the connected integration. Notion needs a field mapping (title, slug, meta, publish date, ready-to-publish flag). Without an integration Outrank offers a free hosted blog, whose default domain is noindex until you add your own domain.

Volume is the pricing lever: 30 articles a month for $99, 60 for +$85, 90 for +$160. A low-volume 8-article plan swaps standard articles for premium images and 100 bonus backlink credits, and also unlocks Publishing Days. Each day on the calendar can be pre-created up to 7 days ahead for review. Otherwise it auto-publishes.

Turn on Network Participation and Outrank does two things. It puts links to other customers into the articles it writes for you, which earns credits ("higher DR hosts earn more per link"). It spends your credits to put links to you into their articles. The screen shows Domain Rating (from Ahrefs), credits ("1 Credit = 1 Domain Rating"), total backlinks, unique sources and an Earned Backlinks table (date, status, credits, DR, source site, source article).

Paid tiers sell more credits: Premium $199/mo (400 credits, about 10 links) and Enterprise $899/mo (1,000 credits, about 25 links). A separate marketplace pays outside DR 40+ sites about $1 per DR per post. The founder says 92% of applicants are rejected.

Reviewers see DR 5–15 sources, with a few at 25 or higher, and the same sites repeat after about 14 links. Users report DR gains of +4 to +12 in a few months, and little effect above DR 25–30. The Terms (as quoted by reviewers) state "no manual preview or approval step" and put the search-penalty risk on the customer.

Improvements needs three things: the feature switched on, Search Console connected, and a CMS that supports republishing. It sorts candidates into five types: almost-there rankings (just outside the top 5), click-through gaps (good position, low CTR → title and meta), declining articles, content gaps (questions not answered yet) and thin content. The panel shows these as filters such as "Edge of 1st page 8" and "Click recovery 6".

It runs at most two publishing days a week, one improvement per day. Each task appears on the same calendar as new articles. By default it waits in an Improvements inbox, where you compare and approve each section and can edit it inline. Auto-push is a setting. Before review, automatic checks confirm that no headings, links or images were lost, run an AI quality review, and strip "AI fluff".

The measurement is the important part. About two weeks after a republish, Outrank compares the article’s clicks and impressions with the baseline from before the rewrite and shows the gain as a win on the Home dashboard. That is the outcome report our agents do not have.

Across the four videos, the in-product metrics are: Vol and Diff per planned keyword, the plan quota (30 articles/mo), Domain Rating "powered by Ahrefs", backlink credits, total backlinks and unique sources, the Earned Backlinks table, the count of linkable pages, and the meta-description character count. No video shows a traffic, rankings or AI-citation dashboard in the product.

The documentation adds a Home dashboard that lists improvement wins (extra clicks against baseline). The AI-visibility marketing page shows citations for each engine (Google AI Mode, Perplexity, AI Overviews, Gemini, ChatGPT, Copilot) for three customer sites, with "Source: Ahrefs, September 2026". It sells results. It is not a product view. Product Hunt reviewers asked for "deeper built-in analytics".

The core plan is "All in One", $99/mo, shown against a struck-through $199. Upsells in the sidebar: Free Tools Builder $199/mo (5 programmatic tools a month on your domain), Human Curated Service $1,399/mo (human editors, 3 clients a month), Directory Submission $500 one-time (350+ directories), and the backlink credit tiers. There is also a REST API, an MCP server and a CLI, so other agents can plan, write and publish through Outrank.

Founder-reported MRR (vendor claims): about $300 as "ContentPie", then $60k in Jun 2025, $200k in Oct 2025 and $300k+ in Apr 2026 across "2,500+ sites". About $10k a month goes to affiliates. The founder’s own chart shows an unexplained drop from about $76k to $20k in Jul 2025. Trustpilot is 3.8 from 98 reviews: 67% are 5-star, and 28% are 1-star, mostly about cancelling, charges and refunds.

Vendor: "median site 2.0× clicks about 7 months in" across 2,119 sites, and selected Search Console charts (15.7K clicks / 6.42M impressions on one). Independent: the April 2026 reviewer paid for 3 months and got 38 clicks, 2.85K impressions, 1.3% CTR and average position 75.6, about $8 per click. He sells a competing tool, so weigh his verdict with that in mind. A blogger’s 60-day test on 3 sites got 57 of 60 articles indexed and DR +2 to +4, with no penalties.

No case of a confirmed Google penalty or deindexing was found. The nearest signal is a user who saw impressions drop after 3 auto-published articles and does not blame Outrank.

Each claim on this page names its source layer. Video frames were catalogued by subagents with exact timestamps, and the parent spot-checked them against the frames. Social and review claims carry URLs in the research notes. Vendor numbers are labelled as vendor claims.

Outrank’s only write path is "create a post in the CMS". It cannot fix a template, a canonical, a redirect, the sitemap or JSON-LD, and its internal links point to whatever the sitemap lists. Fimo’s agents run with Studio tools in the project sandbox and change the source. One template fix repairs 69 pages.

In return, Outrank brings external data on day one (keyword volume and difficulty, Domain Rating, and Search Console for Improvements) and a writing engine. Fimo has neither yet. Its SEO template even says "once Search Console is connected".

"Fimo today" means origin/develop at 19a435a5: the SEO and GEO templates in packages/agents/src/fimo/templates.ts and their built-in skills.

The result is two products that solve different halves of SEO. Fimo is ahead wherever the fix lives inside the site (technical SEO, structured data, AI readiness, safety). Outrank is ahead wherever the input comes from outside (keyword data, Search Console, content volume, outcome reporting). The gaps on our side are integrations and reporting. None of them needs a new architecture.

The SEO template runs 6 daily routines (from 04:00 in the project timezone, 20 minutes apart) and 1 weekly routine. The GEO template runs 2 daily and 1 weekly. Each routine’s instructions say "Load the fimo-seo skill and check…". The skill decides what a run may change on its own.

Each item below already has a home in our plan (mockups/chats-agents/seo-geo-plan.md). Outrank shows that the ordering works commercially.

The consequence for the roadmap: Search Console moves from "phase 2" to the next step, because it unlocks three routines and the win measurement at once. A Content agent comes after it and starts with refreshes, so its first output improves pages that already rank instead of adding more pages.

Outrank’s buyer wants traffic without work and accepts AI posts and link-network risk. Fimo’s buyer owns a real site and wants it correct, fast and citable, with changes they can preview and undo. When Fimo adds Search Console, measured wins and a refresh-first Content agent, it covers Outrank’s best feature (Improvements) inside a product that can also fix the template under the article.