River
Log InStart Free

Desktop Starter Pack/seo-growth-engine

AI SEO Audit & Growth Workspace

Audits your whole site and Search Console, then gives you clear actions and a live dashboard.

Your data in, one ranked queue out

Most SEO tooling starts advising within seconds, which is exactly how long it takes to know nothing about a site. The first read here is deliberately slow, and you get a line as each phase finishes, so you can see what every recommendation afterwards is resting on.

Your data

  • Your site
  • Search Console
  • Live Google results
  • Volume & difficulty
  • Competitor sites
  • Google Analyticsoptional
  • Your CMSoptional
  • Your repooptional
River analyzing…~95 min to ready

Agent Toolbelt — 13 skills, run at your command or on schedule

  • Full Site Assessment
  • Opportunity Sweep
  • Technical Sweep
  • Content Decay Check
  • Cannibalization Resolver
  • Internal Link Pass
  • Keyword Cluster Builder
  • SERP Teardown
  • Competitor Watch
  • Content Brief Builder
  • Content Refresh
  • Editorial Review
  • Weekly Scorecard

What you get

  • SEO Homelive
  • Opportunity Backloglive
  • Site Modellive
  • ReportsDocs
  • DraftsDocs
README.md

SEO Growth Engine

A working SEO desk for one site. It reads the site and Search Console, builds a ranked backlog of work worth doing, and gives you skills to dig into specific problems — striking-distance pages, technical regressions, cannibalization, decay, competitors, briefs.

Nothing publishes to your site from here. Drafts and proposed edits stay in this space until you take them.

What happens when you open it

Type your domain in the box at the top of this page. The read begins as soon as your workspace finishes opening.

It asks for three things, one at a time: your domain, a Search Console connection, and the competitors you care about. Then it runs a full read:

  • every indexable URL on the site
  • ninety days of Search Console (pages, queries, and page×query)
  • the live top ten for the queries that matter
  • a sample of your own pages, so it learns how you write

That takes a while. You will see progress as each phase finishes. You can close the tab.

When it is done you get:

  1. An initial assessment in Reports/ — where traffic comes from, what is close to ranking, what is broken, what looks dead.
  2. Opportunity Backlog — one ranked queue of actionable work. Everything you do in this space eventually lands as a row here.
  3. SEO Home — a dashboard built from that queue and the site model. Ask for a refresh any time.

How you use it day to day

  • Work the Opportunity Backlog from the top. Each row has a type (technical fix, snippet rewrite, new content, refresh, etc.), impact, confidence, effort, and the numbers behind it. Mark things done or delete bad rows — the next sweep respects your edits.
  • Run a skill when you need a specific answer. Open it from the Skills folder, or ask for it in chat. Nothing here is on a schedule; you run what you want, when you want it.
  • Keep the Site Profile honest. Especially the brand-term list. Brand queries get pulled out of striking-distance and click-through analysis, because sitelinks make ordinary pages look like position-one disasters. A wrong list quietly ruins the queue.

Skills

  • Full Site Assessment — full re-read of the site and Search Console. Writes a new dated assessment
  • Opportunity Sweep — refills the backlog from Search Console — striking distance and click-through gaps
  • Technical Sweep — finds indexation, canonical, redirect, schema, and soft-404 problems
  • Content Decay Check — finds declining pages and diagnoses why before queueing a refresh
  • Cannibalization Resolver — which of your URLs should own a query, and what to do with the rest
  • Internal Link Pass — suggests contextual internal links
  • Keyword Cluster Builder — groups keywords by intent, not string similarity
  • SERP Teardown — whether a query is winnable before you spend a week on it
  • Competitor Watch — what moved in the top ten since the last snapshot, including new AI Overviews
  • Content Brief Builder — a brief a writer can work from without a follow-up
  • Content Refresh — diagnoses why a page fell, then proposes edits
  • Editorial Review — thin content, stale stats, off-voice lines — as margin comments
  • Weekly Scorecard — did last week’s shipped work move anything

Practical tips

  • Clear critical technical rows before writing new content. A bad canonical or inherited noindex costs more than any article you ship this month.
  • Ask what a backlog row rests on if it looks wrong. The answer should be a real number from Search Console or the crawl — or it is a bug.
  • Use chat for the thing you are doing right now. Use the backlog for what to do next.
Contents21 files · 5 folders
SEO Growth Engine/The space that lands in your account
Start hereHow the space works, start to finishDoc
SEO HomeThe live picture — rebuilt from the queue, never a stale copy of a reportfirst run
Opportunity BacklogThe one queue. Seven types, ranked, each row carrying the evidence it rests onSheet
Site Model/The living model of the siteFolder
Content InventoryOne row per URL, with a verdict on eachSheet
Keyword MapQuery, cluster, intent shape, and the URL that should own itSheet
Skills/Open one and run it, or just ask in chatFolder
Full Site AssessmentThe deep read, again — produces a new dated reportSkill
Opportunity SweepStriking distance and click-through gaps, against a curve fitted from your dataSkill
Technical SweepIndexation, canonicals, soft 404s, schema, hreflang, redirectsSkill
Content Decay CheckDiagnoses why a page is falling before queueing a refresh that cannot workSkill
Cannibalization ResolverWhich of your URLs should own a query, and what to do with the restSkill
Internal Link PassContextual links only — site chrome is counted separatelySkill
Keyword Cluster BuilderGrouped by intent shape, not by string similaritySkill
SERP TeardownWhether a query is winnable, before anyone spends a week on itSkill
Competitor WatchReal movement since the last snapshot, including new AI OverviewsSkill
Content Brief BuilderA brief a writer could work from without a follow-up questionSkill
Content RefreshDiagnoses why a page fell before proposing a single editSkill
Editorial ReviewThin content, stale statistics, off-voice lines — as margin commentsSkill
Weekly ScorecardDid last week’s work move anything — losses reported firstSkill
Drafts/Work in progress, each piece tied to a row in the queueFolder
Briefs and draftsWritten as you work the queue — nothing leaves here without your approvalfirst run
Reports/Dated, frozen, and allowed to go out of date — that is what makes them evidenceFolder
Initial AssessmentWritten by the first read, then never quietly editedfirst run
Settings & Sources/Domain, connections, competitors, voice — read before every runFolder
Site ProfileDomain, property, page classes, and the brand terms every analysis excludesDoc
Editorial StandardsHow you write, observed from your own best pagesDoc
Competitor SetWho actually holds your top ten, and which of them are institutionsDoc
SEO Growth Engine Space RuleEvidence or silence — read by every task in the spaceRule

Common questions

What does it need access to?

Your domain, and Google Search Console. The domain is enough to crawl the site, read your pages and check the live results; Search Console is what turns that into an audit, because it is the only source that says what you actually earn rather than what you publish. Connecting it takes a minute and River only reads. Without it you still get the crawl, the technical findings and the SERP analysis — every section that depends on query data is labelled as missing rather than quietly estimated. Analytics, a CMS and a repository are optional, and they change what the queue can be ranked by rather than whether it works.

Will it change my live site?

No. Nothing publishes, pushes or edits a live page without your explicit approval, and that is a rule shipped inside the space rather than a setting you have to find. Rewritten titles, patched schema, new internal links and finished drafts all land in the workspace as proposals with the current and proposed version side by side. If you later connect a CMS or a repository, the same rule still applies to every change.

Does anything run on a schedule?

No, and that is deliberate. A sweep that fires every Tuesday whether or not anything changed produces a report you learn to ignore, and it spends your credits to do it. Instead the space ships thirteen skills you run when you want the answer — open one and hit run, or ask for it in chat. What stays current between runs is the queue and the dashboard, because both are re-derived from whatever the last run found rather than being a snapshot somebody has to remember to refresh.

How is this different from Ahrefs or Semrush?

Those are excellent instruments, and they hand you a dashboard. This reads your specific site once, in depth, and then produces the artifacts the work actually happens in — one ranked queue of opportunities, a live model of your site, briefs, drafts and margin comments. The difference in practice is that a dashboard tells you 340 pages have an issue, and this tells you which four are worth your Tuesday, why, and what to change on each...and can even make the changes!

Why is there only one queue?

Because a technical defect, a decayed page, a title worth rewriting and an article that should exist are all the same thing to the person deciding what to do next: work, with an impact and a cost. Splitting them across a fix queue, a content calendar and a backlog means ranking them against each other by hand, every week. Here every skill files into the same Opportunity Backlog under one of seven types, and River deduplicates and re-ranks as new evidence arrives — so the top of one list is genuinely the next thing worth doing.

How long does the first read take?

Usually somewhere between forty minutes and a couple of hours, depending on how many pages you have and how much history Search Console holds. You can close the tab — it keeps working and posts a line as each phase finishes, so you can see what the assessment is actually resting on. Very large sites take longer, and River samples rather than crawling everything, and tells you what it sampled.

Does it make up numbers when a source is unavailable?

No, and this is the part most worth checking in anything you use. Every impression count, position, volume and competitor rank has to come from a real API call in that run; a figure it could not retrieve reads "not retrieved" rather than becoming a plausible-looking estimate. Modelled figures — projected gain, expected click-through — are labelled as modelled where they appear. A failed source is reported as a coverage gap, never as a finding, so a Search Console outage never becomes "your traffic is zero."

What does it do about AI Overviews?

It checks for one on every query before recommending anything, and reports it as a first-class fact rather than a footnote. An AI Overview sitting above the first organic result takes most of the click, so a query that has acquired one is worth a fraction of its apparent volume — River ranks those opportunities down and says why, instead of sending you to fight harder for a result that no longer pays. The Competitor Watch skill also reports queries that newly gained one, which is a change most rank tracking cannot see because nobody’s position moved.

Does it replace an SEO hire?

Not for strategy, and not for anything involving relationships, digital PR, or arguing a technical fix through an engineering backlog. What it does replace is the part of the job that is reading — the corpus audit nobody has time to do properly, the weekly sweep that gets skipped, and the diagnosis that takes an afternoon per page. An SEO with this is considerably faster. A team without one gets a competent read where they previously had none.

What does it cost?

A free account gets you the workspace and the first read. Ongoing use runs on River credits like everything else in the product, and because nothing here runs on a schedule, you only spend them on the runs you actually asked for. The first read is the single most expensive thing the space does; a skill run afterwards is a fraction of it.

Find out what your site is actually doing

Not a score out of a hundred. Which pages carry the site, which are one good week away from ranking, and which will never work no matter what you write.

River is an independent research and drafting workspace. It is not affiliated with, endorsed by, or connected to Google, and Google Search Console and Google Search are trademarks of Google LLC. Search rankings depend on factors outside any tool's control and no ranking or traffic outcome is guaranteed. Search volume and keyword difficulty figures come from third-party providers and are estimates. Verify any change against your own data before acting on it. By using River you agree to our Terms of Service and Privacy Policy.