Publish less.
Rank more.
Rankow turns competitor data into a page-level plan, writes every article to the same quality floor, publishes it, and proves what worked with Search Console.
Automatic publishing live on Shopify · 28-day feedback loop
Today's run
Cycle 04 · 120 page units
Discovery
4 sources · 12,000 candidates
Planning
120 page units · 8 hubs
Writing
Draft 03 / 10 · 3,000+ words
Quality gate
4 / 4 checks passed
Publish
Shopify · scheduled 14:00
Feedback
Search Console · 28-day read
Pages tracked
14
Index rate
57%
Words drafted
3,000+
Shopify Admin API
Automatic publishing is live today
Search Console
28-day performance feedback
SERP & keyword data
Official API, metered per request
What the engine produces
Four outputs. Every claim points at the file that implements it.
engine/dfs/planning/plan.py
A 12,000-keyword candidate pool becomes this month's page list.
120 page units per cycle: hubs first, supporting pages after, pillars gated.
engine/dfs/writing/gate.py
Every draft clears 3,000 words and an information-gain check before it can publish.
Length is a floor, not a finish line. A draft below the floor or without a gain signal is blocked.
Google Search Console
Day 28: Search Console tells you what happened to each page.
Index status, average position, impressions, and CTR are read per URL, not per account.
engine/dfs/feedback/rules.py
Published pages keep getting rewritten until position moves — or get rolled back.
14-day refresh eligibility, 7-day thin-content rule, position 5–20 with CTR under 2%, rollback on a drop over 3 positions.
Publishing channels
How publishing worksCMS
Framework
Webhook
Automatic publishing is live on Shopify. Every other channel takes publish-ready drafts or structured payloads.
The gap
Three gaps that volume cannot close.
Most content programs fail before writing starts. The rest fail because nobody reads the data after publishing.
01
Keywords are not a plan.
A list of 12,000 keywords tells you what people search. It does not tell you which pages to build, which pages compete with each other, or which one comes first.
02
Word count is not quality.
A 3,000-word draft can still say nothing new. Length is a floor, not a finish line. Without information gain, volume becomes filler at scale.
03
Publishing is not proof.
Without post-publish feedback, every article is a guess. No index rate, no position movement, no decision about what to do next.
The loop
One loop. Four decisions.
Each step has an input and a deliverable. Nothing runs on vibes.
Discover → Plan → Write → Prove
Every cycle, measured at 28 days
Discover
Input
Competitor domains + Search Console
Output
12,000 keyword candidates
Plan
Input
Entities × intent + hub rules
Output
120 page units per cycle
Write
Input
Page brief + 3,000-word floor
Output
A draft that passes the gate
Prove
Input
Search Console at 28 days
Output
Keep, refresh, retitle, or roll back
Decision rules
Our decision rules, in the open.
Most tools hide how they choose what to write. We publish the weights. Every keyword is scored by the same five-factor linear formula — no machine-learning black box, no vibes, no silent reranking.
Scoring weights
Log-normalised against a 5,000-search reference, so one huge keyword cannot dominate.
Lower difficulty scores higher. Harder keywords have to earn their place.
Intent class weighted by detection confidence, not a flat label.
CPC against a $2.00 reference — a proxy for what advertisers already pay for.
The ad-competition signal, inverted: less contested, more room to rank.
Linear weighting on real market metrics. Deterministic, auditable, and identical on every run.
Try the scoring formula
Move the inputs. The bar shows how much each factor contributes to the final 0–1 score. Runs entirely in your browser.
1,200searches / mo
30
$1.2$ / click
0.40
Simulated score
0.719
0–1 scale
The same weights our engine runs in production.
The quality gate
A draft is checked before it can be scheduled. Two rules decide the outcome.
3,000-word floor
Length is a floor, not a finish line. Below it, a draft fails regardless of who approves it.
Information-gain check
Comparison tables
The page has to organise facts, not just narrate them.
Specification numbers
Concrete figures a reader can act on.
Verifiable sources
Claims that trace back to a checkable origin.
Craft or process detail
First-hand specifics generic AI text cannot fake.
Search Console decision rules
The only feedback that counts is what the data shows. Four rules turn it into action.
28d
Performance read
Impressions, clicks, and position are read at 28 days, with a lag buffer so partial data never drives a decision.
>3
Rollback trigger
If a page drops more than three positions after a change, it can be rolled back to the pre-change snapshot. URLs never change.
14d
Refresh eligibility
A page becomes eligible for a refresh 14 days after publish, once it has real data.
90d
Edit cooldown
Ninety days between edits on the same page, so measurement windows stay clean.
Fit check
Is Rankow for you?
Four questions. If most answers are yes, a pilot cycle will tell you more than any conversation.
You run a Shopify storefront
Publishing is live through the Shopify Admin API, with metafields, JSON-LD, and rollback.
You budget for content every month
The engine plans and ships on a steady cadence. It needs scope, not bursts.
You want scale without filler
Every draft clears the same 3,000-word floor and information-gain check.
You want content tied to Search Console
Publishing is only half the loop. The 28-day read decides what happens next.
Rankow is not for you if…
You need a one-off batch of posts next week with no plan behind them, or you expect guaranteed rankings. We do not promise positions; we report what the data shows.
Capabilities
Six layers, one engine.
Open any layer to see what goes in, what comes out, and what stops a bad run.
Evidence-based discovery
A ranked keyword pool with a source on every term.
How
Competitor rankings and Search Console queries certify the seeds; keyword ideas, suggestions, and related terms expand them into a ranked, source-tagged pool.
Evidence · 4 sources · 12,000 candidates, capped by volume
Page-level planning
120 page units per cycle, ordered hub-first.
How
Entities and intents split into page units. Hubs come first, supporting articles follow, and pillars are gated.
Evidence · Pillar gate: over 50k volume or KD above 40
Brief-driven writing
A draft that either clears the gate or does not ship.
How
Every draft starts from a brief and has to clear a 3,000-word floor plus an information-gain check.
Evidence · 3,000-word floor · 1 of 4 gain signals required
Shopify publishing
An article live on your site, with a snapshot stored first.
How
Live publishing through the Shopify Admin API, with SEO metafields, JSON-LD, and idempotent writes.
Evidence · Idempotent write · URL never changes
Search Console feedback
Refresh, optimize, hold, or roll back — decided by data.
How
At 28 days, index status and position decide whether a page is kept, refreshed, retitled, or rolled back.
Evidence · Day 28 read · position, impressions, CTR per URL
Guardrails
A bad cycle stops before it becomes a bad month.
How
Budget gates, offline drills, monthly caches, cost ledgers, and snapshots keep a bad cycle from becoming a bad month.
Evidence · Budget gate · offline drill · cost ledger · snapshot
Deep dive · 01
Every competitor. Every query. One plan.
The pool is always bigger than the page count, on purpose. Competitor organic coverage and your Search Console queries supply the demand evidence; keyword ideas, suggestions, and related terms expand that into the whole keyword landscape; the planner then collapses it into page units, one intent each.
- Hub-first. Category and hub pages are planned before supporting articles.
- Pillar gate. A pillar page needs more than 50k monthly volume or keyword difficulty above 40.
- One intent, one page. Entities and intents are split so pages do not compete with each other.
Candidate pool → plan
Cycle 04Set at onboarding
Capped and volume-ranked
Planned, not dumped
Page units
Ceramic cookware
Cast iron care
Seasoning
Rust removal
Sample plan · illustrative data
Deep dive · 02
The gate before publish.
Every draft has a 3,000-word floor. Before it can be published it also has to pass an information-gain check: at least one of four signals that the page adds something the ranking pages do not.
- Comparison table
- Specification numbers
- Verifiable source
- Craft detail
We audited 37 published articles on our customer-zero site. The floor held: nothing under 3,000 words was indexed.
Read the writing gateQuality gate
PassUnder 3,000 words
0 / 7
indexed
3,000 words or more
17 / 30
indexed (57%)
Audit of 37 articles · one site · not a ranking guarantee
The loop continues
Publish is the midpoint.
Most tools stop at publish. Rankow treats a live article as the start of the measurement window: 28 days of Search Console data decide whether the page is refreshed, optimized, held, or rolled back.
Publish
The article goes live with the URL untouched and the previous version stored as a snapshot.
Threshold
Refresh eligibility opens at day 14; pages under 3,000 words at day 7.
28-day read
Search Console reports index status, impressions, average position, and CTR for that URL.
Threshold
The read happens at day 28, per URL — never averaged across the account.
Decide
The page is refreshed, optimized, or held, and the rule that fired is recorded with the action.
Threshold
Optimize when position is 5–20 with impressions ≥ 5 and CTR under 2%; one edit per URL every 7 days.
Rewrite & republish
The page is rewritten, republished at the same URL, and measured again in the next window.
Threshold
90-day cooldown before the next edit; a drop over 3 positions rolls back to the snapshot.
One run, end to end · Illustrative run · sample data, not a client result
See the full decision rulesPipeline
4 sources merged → quality gate → publish → 28-day read
01
4 sources merged
12,000
Competitor organic · Search Console · ideas · related terms
02
Quality gate
4 / 4
3,000-word floor plus one information-gain signal
03
Publish
Shopify · live
SEO metafields, JSON-LD, WebP, snapshot stored first
04
28-day read
GSC · per URL
Index status, impressions, position, and CTR
Deep dive · 03
Search Console closes the loop.
After publish, every page is measured at 28 days. Index rate, position, and clicks decide what happens next. No opinion required.
Day 0
Publish with a stored snapshot
Day 14
Refresh eligibility opens
Day 28
Position and index read
Day 90
Cooldown ends before another edit
Index rate below 70%
The queue throttles until indexation recovers.
Position drops more than 3
The previous snapshot can be restored.
Search Console · 28 days
SampleTracked page
Positions 5–20
Sample chart · index rate 57% from the 37-article audit
The difference
Built to answer one question.
Why buy planning, a quality floor, and feedback instead of another monthly batch of articles?
| Capability | Typical stack | Rankow |
|---|---|---|
| Keyword & topic selection | Manual, gut feel | Difficulty + intent scored, auto-clustered into hubs |
| Research depth | Skimmed stats, no citations | Source-grounded briefs with verified claims |
| Brand voice | Generic AI tone | Trained on your voice, enforced on every draft |
| Internal linking | Manual or missing | Mapped to money pages automatically |
| Images & screenshots | Stock photos or none | On-brand images and real UI captures |
| Publishing cadence | Sporadic, slips | Steady weekly drumbeat |
| Attribution | Vanity metrics PDF | Per-post rankings, clicks, and signups |
Every Rankow claim links to the product section that implements it.
Plans
Scoped by work, not seats.
Every plan runs the same engine. Starter is the default: one site, the full loop, $49 a month.
Starter
A first cycle on a first content section.
Billed monthly · one site
Start free trialStart with one site. Add more only when you need them.
What's included
- 1 site, 1 content section
- 8 articles per month
- 4 refreshes per month
- 5 competitor domains
- 20,000 data requests per month
- 2M LLM tokens per month
- 16 images per month
- Search Console feedback, 28-day reads
Growth and Scale are quoted after a demo.
See what every plan includesQuestions
Answers, without a slide deck.
If your question is not here, ask it on a 20-minute call. We will answer with the implementation, not a promise.
Ask on a callA writing tool starts with a prompt. Rankow starts with competitor data, plans the page set first, and refuses to publish below a 3,000-word floor without an information-gain signal. The draft is the middle of the loop, not the product.
Yes on Shopify, through the Admin API, with SEO metafields, JSON-LD, and WebP images. On other platforms Rankow produces publish-ready drafts and structured data for your team to ship. Automatic publishing is live for Shopify today.
You do. Every cycle passes through your review before publishing. Rankow handles discovery, planning, drafting, quality checks, and measurement; your team owns brand voice and exceptions.
Performance is read at 28 days using Search Console data. Ranking movement depends on your site, competition, and indexation. We report what the data shows rather than promise a position.
If a tracked page drops more than three positions at the 28-day check, Rankow can roll back to the snapshot taken before the last change. URLs never change in the process.
Not necessarily. Rankow plans new page units from competitor coverage, and existing posts are audited and refreshed instead of duplicated. A content section makes the first cycle easier to ship, but it is not required.
Start here
See the plan before you commit.
Book a 20-minute session. We will walk through a page-level plan built from your competitors, not a deck.
7 days · 1 site · No credit card