CASE STUDY · HARBOR.IO

From 0 to 10,000 monthly organic visitors in 6 months.

The exact playbook we ran on our own site — founder's account, written by Hamish, who also runs the Income Stream Surfers YouTube channel. No fabricated numbers, no client NDAs to hide behind. This is the site you're reading.

By Hamish · Founder, Harbor
Monthly visitors (month 6)
10,000+
Articles published
~200
Avg. time per article
~12 min
Articles ranking in 30 days
~80%
CONTEXT

The problem (month 0).

When we launched Harbor, we had a product built to generate AI-written SEO content at scale. We also had the classic founder problem: the product was good, nobody knew it existed, and we couldn't credibly tell people to use AI-generated content for SEO if our own site relied on paid ads to acquire users.

The obvious answer: run Harbor on Harbor. Eat our own cooking. If the product couldn't rank our own site, we had no business selling it to anyone else.

Month zero: blank domain, zero organic traffic, zero content. Six months in: over 10,000 monthly organic visitors and a handful of citations across Google AI Overviews, Perplexity, and ChatGPT. No paid boosts, no link buying, no PR stunts. Just the workflow we built the product to enable.

What follows is the actual playbook — the strategy, the execution, what we'd do differently, and the Harbor features we leaned on. Specific hard numbers are reported as ranges or round abstractions where the exact figure would be noisy; everything directional is what we actually saw.

MONTH 1–2 · THE STRATEGY

Four pillars we committed to on day one.

Everything that followed was downstream of these four decisions. Every time we were tempted to deviate, growth slowed.

01

One domain, one topic

We didn't split across microsites. Every article lived on harborseo.ai under a cluster that mapped to a primary topic we could credibly own.

02

Cluster-first, keyword-second

We designed the clusters before writing anything. Each cluster had a pillar page and 20–40 supporting articles, all interlinked. Keywords slotted into the cluster map, not the other way round.

03

8 articles a week

A sustainable cadence: two writers' worth of output, produced by Harbor in a few hours per day. Enough to compound; not so much that editorial quality slipped.

04

Publish, measure, iterate

Every article got a 30-day check. If it didn't rank in the top 30 on its target query, it got a rework pass. Most of the rework was answer front-loading and FAQ expansion.

MONTH 3–4 · THE EXECUTION

The six-step loop we ran, every week.

Briefs → drafts → editorial pass → publish → measure → rework. Every step lived inside Harbor, and the whole loop closed itself.

01

Keyword and cluster design

We started by mapping every query a Harbor user might have across their SEO journey. We grouped queries into clusters, assigned each cluster a pillar page, and drafted a spoke list of 20–40 supporting articles per pillar.

02

Brief generation

Harbor generated a full brief for each article: target query, search intent, top 10 SERP analysis, H2/H3 outline, internal link suggestions, and FAQ questions. A human editor approved or tweaked the brief in under 2 minutes.

03

Draft writing

Harbor wrote the full 3,500-word draft from the approved brief — with answer front-loading, FAQ schema, comparison tables, internal links already in place, and every stat cited. Average draft time: ~8 minutes.

04

Editorial pass

A human read every article before publish. Tone adjustments, local context, adding anecdotes from our own product use. 10–15 minutes of human time per article. This step is non-negotiable — it's what keeps the site from feeling like a content farm.

05

Publish to WordPress

Harbor pushed the final article to WordPress via the REST API — featured image, meta description, schema, category, tags, all populated. Zero copy-paste. Publishing was literally one click.

06

30-day review

Every article hit a review queue 30 days after publish. If it was ranking top-30, we left it alone. If not, Harbor's rework agent analyzed the current top 10 SERP and regenerated the sections that needed strengthening.

MONTH 5–6 · THE INFLECTION

Where the curve bent.

In our testing, the first two months look slow and the last two look magical. Nothing changed between them except compounding.

Month 4

First AI Overview citation

A mid-tail query in our primary cluster started showing our article as a cited source in Google's AI Overview box. Referral traffic from AI Overviews appeared in GA4 as direct/organic the same week.

Month 5

First ChatGPT citation

A prompt we tracked weekly started citing our pillar page as one of four sources. Small traffic volume on its own — but a leading indicator that the on-page structure (answer front-loading, FAQ schema, clean headings) was doing its job.

Month 6

MoM growth decouples from new content

The site started growing ~20% month-over-month even in weeks where we published less. Internal linking density, backlink accumulation, and cluster maturity had become compounding inputs in their own right.

Why month 4 is the hinge
The first three months look like a flat line and the fourth month looks like a hockey stick. Most teams quit between month 2 and month 3. Our biggest unfair advantage wasn't the content — it was not quitting.
THE NUMBERS

Traffic by month.

Month-by-month visitors, with the context of what was actually happening that month. Figures are rounded — directional, not audited.

MonthOrganic visitorsWhat was happening
Month 00Blank domain. No content. No rankings.
Month 1< 100First cluster published. Crawl starts, nothing ranks yet.
Month 2~400First long-tail rankings appear. Pages 3–5 mostly.
Month 3~1,400Cluster maturing. Internal linking starts compounding.
Month 4~3,200First page rankings on mid-tail queries. First AI Overview citation.
Month 5~6,500Compound growth kicks in. First ChatGPT citation tracked.
Month 610,000+Steady state. Traffic now grows ~20% MoM without new clusters.
Articles published
~200
Avg. time per article
~12 min
From brief to publish, including editorial pass.
Articles ranking in 30 days
~80%
Top 30 position on the target query.
TOP RANKING KEYWORDS (CURRENT)
how to rank in perplexity ai
Top 10
Pillar page, high-intent
ai blog writer seo
Top 5
Product-adjacent, converts well
topical authority
Top 10
Informational, feeds cluster
seo content brief
Top 5
Product-adjacent, converts
programmatic seo
Top 10
High-volume, feeds cluster
LESSONS

What we'd do differently.

Four things we got wrong the first time. If you're starting now, skip our mistakes.

01

Build the pillar pages first

We wrote pillar pages and spokes in parallel. In hindsight, pillars first — let them mature for two weeks — then release the spokes underneath. Pillars anchor the cluster; spokes without a ranked pillar float.

02

Start Bing submission on day one

We added Bing Webmaster Tools in month 3. That was ~8 weeks of SearchGPT/ChatGPT invisibility we didn't need to eat. If you care about AI search citation, Bing submission is a month-zero task.

03

More original data earlier

Our most-cited pages are the ones with our own data — survey results, usage benchmarks from the product, before/after metrics. We added these in month 4. They should have been in from month 1.

04

Don't over-edit early drafts

We spent too long in the first month manually rewriting Harbor's drafts. By month 3 we had learned to trust the output and edit lightly. The output quality hadn't changed — our over-editing had.

THE EXACT HARBOR WORKFLOW

Every Harbor feature we used, in the order we used it.

If you want to run the same playbook, these are the tools — links take you to the feature page for each.

01

Keyword → cluster map

Harbor's keyword research surfaces clusters and maps intent to H2/H3 structure. The cluster becomes the content plan.

See feature
02

Brief generator

Full briefs with SERP analysis, outline, FAQ suggestions, and internal link candidates — generated in under a minute per article.

See feature
03

AI blog writer

3,500-word drafts in ~8 minutes. Answer front-loading, FAQ schema, internal links, comparison tables — all baked in.

See feature
04

WordPress publishing

Direct REST API publish with featured images, meta, schema, and categories. No copy-paste, no 'staging in Google Docs'.

See feature
05

30-day rework loop

Articles that don't rank get an automated rework pass against the current top 10 SERP. The loop closes itself.

See feature
YOUR TURN

Run this playbook on your site.

Harbor generates the briefs, writes the drafts, publishes to WordPress, and reworks the articles that don't rank — on a workflow we built by running it on ourselves.

From 0 to 10,000 Monthly Visitors — Harbor SEO Case Study