The Publisher Hex: Running Multiple Sites With Six Tools
You would think that running five websites requires five times the tools. That is what every SaaS pricing page is designed to make you believe — and what most publishers end up building toward, one subscription at a time, until the tool stack costs more than the hosting and the management overhead eats the hours that were supposed to go to writing. One publisher runs five Ghost sites, publishes 15 to 20 articles per week across all of them, and does it with six tools total. Not six per site. Six.
This is how the hex constraint works when the job is publishing at scale on a solo budget.
The Profile
The publisher operates five niche sites on Ghost, each targeting a different vertical. Combined email list is around 9,000 subscribers across the five properties. Revenue comes from a mix of membership tiers, affiliate links, and one sponsored newsletter. No employees. No contractors for writing — though there is a VA who handles image uploads and scheduling for roughly ten hours a week.
Before the hex, the tool count was somewhere north of twenty. Every site had accumulated its own set of subscriptions — partially because different sites had different needs, partially because the publisher kept trying new tools and never canceled the old ones, and partially because the SaaS model is specifically designed to make cancellation feel like loss. Two SEO platforms. Three different AI writing tools across the sites — Claude for one, ChatGPT for two others, Jasper for the remaining two — because each had been adopted at a different time and "worked fine" in isolation. Image generation split between Midjourney and two other tools. Project management lived in Notion for three sites and Trello for two. Email marketing used Ghost native on two sites and Kit on the other three because those had been started before the publisher moved to Ghost.
Monthly subscription cost across all sites: $480 [VERIFY]. The publisher could not name all the active subscriptions without checking the credit card statement. That fact alone was diagnostic.
The Hex Applied to Multi-Site Publishing
The hex question for a multi-site publisher has a wrinkle that single-site operators don't face: do the tools need to be the same across all sites, or can each site have its own hex? The answer — arrived at through a month of experimentation — was that shared tools are the whole point. The constraint isn't just about limiting the number. It's about limiting the cognitive switching cost.
Running five sites with five different tool configurations means five different muscle memories, five different prompt libraries, five different workflows. The mental overhead of remembering "which site uses which tool" was eating time the publisher didn't even notice was missing — until the hex forced standardization and the time came back.
Six tools survived. One LLM — Claude — for all drafting, editing, and repurposing across every site. One image generation tool for all visual assets. Ghost for publishing and native email on all five sites, which required migrating the three Kit-based newsletters into Ghost's email system. One analytics dashboard — Plausible, replacing Google Analytics on the sites that had it and two other analytics tools on the sites that didn't. One bulk scheduling tool for social promotion. And a single shared Notion workspace — stripped down from an elaborate five-board system to one database with a site column.
That's it. Five sites. Six tools. The same six for every site.
What Got Cut and Why
The hardest cut was the second and third LLMs. The publisher had genuinely different prompting approaches for different sites — the health site used a more clinical voice, the lifestyle site was conversational, the finance site was data-heavy. The assumption was that different voices required different tools. That assumption was wrong. Different voices require different system prompts in the same tool. Claude with a clinical system prompt writes clinically. Claude with a conversational system prompt writes conversationally. The voice lives in the prompt, not the platform.
Jasper and ChatGPT were canceled in the same week. The publisher spent one afternoon recreating the three best prompts from each tool as Claude Projects — custom instructions, example outputs, and reference materials loaded into persistent project contexts. The migration took four hours. The monthly savings were $85.
The SEO tools went next. Two platforms — one for keyword research, one for on-page optimization — that together cost $170/month and produced recommendations the publisher followed roughly 30% of the time. The publisher was paying for data and then mostly ignoring it, which is the subscription model's ideal customer. Google Search Console — free — provided the actual performance data that mattered. The keyword research was replaced by a Claude prompt that analyzed Search Console data and suggested related terms. Not as comprehensive as a dedicated SEO platform. Comprehensive enough for five niche sites where the competitive landscape is thin.
The multiple image tools collapsed into one. The publisher had been using Midjourney for "quality" images and two faster tools for quick blog headers, reasoning that different content types justified different tools. After the hex, one tool handled everything. The quality difference between tools was real but irrelevant — readers of niche Ghost blogs are not evaluating header images against a Midjourney portfolio. They are scanning past them to the content. The image needs to be relevant and not ugly. One tool clears that bar.
The Migration Pain
Moving from Kit to Ghost native email on three sites was the most technically painful part of the hex. Kit had automations — welcome sequences, tag-based segmentation, conditional content blocks — that Ghost's email doesn't replicate one-to-one. The publisher had to simplify the email strategy for those three sites.
This turned out to be a benefit disguised as a loss. The Kit automations had been built during a "marketing optimization" phase that produced elaborate funnels for sites with 1,200 to 2,400 subscribers each. The welcome sequences had five emails. The segmentation used twelve tags. The conditional content blocks served different versions based on subscriber behavior. All of this infrastructure existed for audiences small enough that the publisher could have — and probably should have — written personal emails.
Ghost native email does one thing: sends a newsletter when you publish a post or write an email. No automation. No segmentation beyond basic tiers. For a solo publisher running five sites, that simplicity is the feature. The publisher writes the email, the email goes out, the publisher moves on to the next site. The open rates across all five sites held steady within two percentage points of where they'd been under Kit. The conversion rates on membership CTAs didn't change. The infrastructure that had taken hours to maintain was producing zero measurable lift.
The Workflow After the Hex
A publishing day now follows the same pattern regardless of which site is getting content. Open Claude with the site's project context loaded. Draft or outline the article. Edit the Claude output into the publisher's voice. Generate a header image. Paste into Ghost. Schedule. Move to the next article.
The uniformity is the productivity gain. There is no "let me remember how the finance site workflow works" step. There is no context switch between tools. The publisher's fingers do the same thing in the same order for every piece of content on every site. The muscle memory compounds.
The Notion workspace — the only project management tool that survived — tracks all five sites in a single database. One view per site, filtered by a site property. One "this week" view across all sites. The publisher checks one place to know what needs writing, what's in draft, and what's scheduled. Before the hex, this information lived across five different boards in two different tools, and the "what's the status" question required opening multiple tabs and reconciling conflicting information.
The VA's workflow simplified in parallel. Instead of learning the idiosyncrasies of different tools for different sites, the VA has one image tool, one CMS, one project database. Training time for new tasks dropped from "let me write you a doc explaining how this site's workflow differs" to "same as the other sites."
The Numbers After Three Months
Before the hex: 20 tools, $480/month in subscriptions, 15 articles per week across five sites, roughly 40 hours per week of publisher time plus 10 hours VA time.
After the hex: 6 tools, $140/month in subscriptions [VERIFY], 18 to 20 articles per week across five sites, roughly 32 hours per week of publisher time plus 8 hours VA time.
The subscription savings alone — roughly $340/month — paid for the VA's additional hours. The publisher's time savings — roughly 8 hours per week — went to writing more articles and to a new site that launched two months after the hex was implemented. The sixth site runs on the same six tools. No new subscriptions required.
The hex didn't make the publisher more creative or more talented or more strategic. It removed the overhead that was sitting between the publisher's existing skills and the published output. The tools that survived are the tools that directly participate in the act of turning ideas into articles that readers see. Everything else was infrastructure theater — real tools doing real things that didn't connect to the only metric that matters for a publisher: pieces published per week that people actually read.
The Ongoing Discipline
Three months in, the publisher reports that the hardest part is not the initial cut. The initial cut is painful but finite — you make the decisions, you cancel the subscriptions, you migrate the workflows, and then it's done. The hard part is the ongoing discipline of not adding tools back.
Every week brings a new tool launch, a new Twitter thread showing a workflow that "saves 10 hours," a new free trial that promises to solve a problem the publisher is currently solving manually. The impulse to try it hasn't gone away. What's changed is the response. Before the hex, the response was "let me try it." After the hex, the response is "which of my six tools does this replace, and is the replacement worth the migration cost?" The answer, so far, has always been no — not because the new tools aren't good, but because the switching cost exceeds the marginal improvement.
The hex for a multi-site publisher isn't about finding the perfect six tools. It's about finding six tools that are good enough across all your sites, standardizing on them, and then redirecting the time you used to spend evaluating tools toward the work the tools are supposed to serve.
This is part of CustomClanker's Hex in the Wild series — real setups from real people. Start with The Hex Explained if you haven't downloaded the constraint PDF yet.