How I Brief AI to Write Industry-Specific Content That Does Not Sound Generic

This is the final post in the series. You can read the full strategy overview, keyword mapping, Sonar research process, cluster architecture, and publishing calendar strategy in the earlier posts.

The Problem With Generic AI Content

Ask any AI to write a blog post about AI for law firms without a proper prompt and you will get something that reads like a Wikipedia summary written by a consultant who has never worked with a law firm. Every sentence is technically accurate. Nothing is specific. The tone is the same whether the audience is a solo attorney or a healthcare CFO. It sounds like every other AI blog post because it was produced the same way every other AI blog post is produced.

The fix is not a better AI model. It is a better input. The model is only as specific as what you give it. A vague prompt produces vague output. A highly specific prompt with clear tone rules, structural requirements, and sourced research data produces something that actually reads like it came from someone who knows the industry.

For the Tiger Tail project, 110 posts across five distinct industries needed to sound different from each other. A home services contractor does not want to read the same prose style as a law firm partner. Healthcare administrators do not respond to the same tone as accounting firm owners. The writing framework had to produce genuinely different output for genuinely different audiences from the same underlying system.

The Master Prompt Structure

Every blog post across all 110 briefs was written through a single master prompt. The prompt has five core sections. Each one does a specific job.

The brand and audience context section is what most prompts skip entirely. Without it, the model has no reference point for who is reading. With it, every sentence is written toward a specific person in a specific situation rather than a generic professional in a vague industry.

The Tone Rules That Make the Biggest Difference

The tone-by-industry section of the prompt is what makes a legal post read differently from a home services post even though both go through the same system. Here is what each industry tone rule actually produces in practice:

The difference between a healthcare post and a home services post is not just vocabulary. It is the entire posture of the writing. Healthcare content that leads with money the way home services content does would feel wrong to a physician. Home services content that hedges and acknowledges complexity the way legal content does would lose a contractor in the first paragraph.

The Filler Openers List

One of the most specific sections of the prompt is a list of opening lines the model is explicitly told never to use. This matters more than it sounds. AI defaults to filler openers because they are the most statistically common way articles begin in its training data. Without a specific prohibition, the model will produce them every time.

Every one of those openers signals to a reader that the content was not written for them specifically. They have read that sentence a hundred times on a hundred different sites. Banning them forces the model to find a more direct entry point, which almost always produces a better opening than the default.

The Humanizer Pass

Even with a strong prompt, AI writing has patterns that accumulate across a long piece. Small habits that individually seem fine but collectively make the content feel generated rather than written. The humanizer pass catches these before anything gets published.

The humanizer pass is not a stylistic preference. It is a quality gate. Every one of those patterns makes content feel less trustworthy to the reader, even if they cannot articulate why. Removing them is what takes a competent draft and makes it feel like something a real expert wrote.

The Final Check Before Publishing

Before any post from the Tiger Tail calendar goes live, it passes a final checklist. This is the last gate between the draft and the published URL.

The last question in that checklist is the most important one. Not “does this read well” but “what would make a reader suspect this was generated.” That question forces a more honest review than general proofreading does because it is looking for the specific patterns that erode trust rather than just checking for errors.

What the Whole System Produces

Put together, the five layers covered across this series produce something most content strategies never achieve: a system where every post is intentional before it is written, researched before it is briefed, structured before it is published, and reviewed against the right standard before it goes live.

The keyword map tells you what to write and where it belongs. The Sonar research tells you what to say and how to back it up. The cluster architecture tells you how each post relates to everything around it. The calendar tells you when to publish it and removes every remaining decision. The writing framework and humanizer tell you how to make it sound like it was written by a person who genuinely knows what they are talking about.

I Built All of This. I Can Build It for Your Business.

This is the fifth post in the series. You can read the strategy overview, keyword mapping, research process, and cluster architecture in the earlier posts.

Why Most Content Calendars Get Abandoned

A content calendar that lives in a spreadsheet and gets ignored by week three is not a calendar. It is a guilt document.

Most content calendars fail for one of two reasons. Either they are built with titles and dates and nothing else, which means every publishing cycle starts with a blank page and a deadline. Or they are built with so much structure that maintaining the document takes more effort than writing the actual content.

The calendar I built for Tiger Tail was designed around one principle: every row should contain everything a writer needs to start immediately with no additional briefing required. The calendar is the brief. The moment a post moves to “in progress,” the writer already has the research data, the source URLs, the internal links, the intent classification, and the CTA. Nothing is left to figure out.

What Every Calendar Row Contains

Here is the exact column structure used for every one of the 110 posts in the Tiger Tail calendar:

Twelve columns per row. One hundred and ten rows. Every decision about every post made before the writing starts. A writer who picks up a brief from this calendar does not need to ask any questions. Everything is already there.

The Publishing Pace and Why It Was Set This Way

The publishing pace for a new domain is not just a volume decision. It is a trust-building decision. Google needs time to learn a new site. Publishing fifty posts in the first month on a brand new domain does not accelerate that process. It looks like a spam pattern to a domain with no history.

The ramp from one to two posts per week was deliberately delayed until week nine. Eight weeks of consistent single-post publishing gives the domain enough history that doubling the pace looks like organic growth rather than a sudden content dump. The distinction matters to how Google interprets the signal.

The Cluster Priority Order

The order in which clusters get their first posts published was not decided alphabetically or by which felt most important to the client. It was decided by competition level and by what would give the domain the fastest path to early ranking signals.

The first three clusters were chosen because they give a new domain the fastest path to real ranking signals. Lower competition keywords on a new domain rank faster. Those early rankings build the domain authority that makes it possible to compete for the higher-volume keywords in clusters four through seven later in the program.

Starting with the legal cluster, which targets “ai for law firms” at 1,300 monthly searches, on a brand new domain would mean months of sitting on page ten for a keyword that Forbes, HubSpot, and established legal tech publications are already competing for. Starting there after six months of authority building from clusters one through three changes that calculation significantly.

On-Page Requirements Built Into Every Row

The calendar also carries on-page SEO requirements for every post so nothing gets published with missing elements. These are not suggestions. They are publishing gates.

The meta title and meta description are written when the calendar row is built, not when the post is about to go live. This matters because writing SEO metadata under deadline pressure produces generic titles that do not perform. Writing them as part of the planning process, when there is no urgency, produces titles that are actually designed to be clicked.

The Domain Authority Building Work That Runs Alongside Content

Content is the primary organic acquisition channel but it does not operate in a vacuum on a new domain. The calendar strategy included a set of parallel activities designed to accelerate the authority-building process from day one.

Content without any off-page authority signals takes longer to move. These parallel activities do not replace the content work. They compress the timeline by giving Google additional trust signals while the cluster authority is still building.

What a 24-Month Calendar Actually Delivers

By the end of month 24, the Tiger Tail content program will have published 110 posts across 11 clusters, each one mapped to a commercial page, each one backed by real research and source citations, and each one part of an interconnected architecture that compounds in value every month it runs.

That is not a blog. That is an organic acquisition system that runs on a schedule, requires no paid media, and gets more valuable over time rather than less.

I Built This for a Client. I Can Build It for You.

The last post in this series covers how I brief AI to write industry-specific content that actually sounds like it was written by someone who knows the subject: how I brief AI to write content that does not sound generic.

This is the second post in a series about building a 110-post SEO content strategy from scratch. If you missed the first one, start here for the full overview.

The Problem With Keyword Research Done in Isolation

Most businesses approach keyword research the same way. They find a tool, type in their industry, get a list of terms with search volumes, pick the ones that look promising, and hand them to a writer. The writer produces content. The content gets published. Nothing ranks.

The missing step is not better keywords. It is understanding which page on the website each keyword belongs to and why. A keyword does not exist in a vacuum. It needs a home. And that home needs to be the right type of page for the intent behind the search.

Without that mapping, you end up in one of two bad situations. Either you create blog posts competing against your own service pages for the same keywords, or you create service pages targeting keywords that should be blog content. Both confuse Google and split your ranking potential instead of concentrating it.

The Two Types of Pages That Need Keywords

For the Tiger Tail project, the website had two distinct types of pages before a single blog post was written. Service pages and industry pages. Each type needs its own keyword logic.

Service pages target keywords where the searcher is looking for a solution or a provider. Someone searching “ai strategy consultant” or “workflow automation services” has commercial intent. They are not looking for an explanation. They are looking for someone to hire. These keywords belong on service pages, not blogs.

Industry pages target keywords where the searcher is a specific type of business looking for AI solutions relevant to their sector. Someone searching “ai for law firms” or “ai for real estate agents” has commercial intent too, but with an industry-specific lens. These keywords belong on the industry pages, not the blog either.

Blog posts serve a different purpose. They capture informational searches from people who are not ready to buy yet but are researching the problem. The blog content feeds authority to the service and industry pages. The pages convert. The blog attracts.

The Actual Mapping: Real Data From the Project

Here is what the keyword-to-page mapping looked like for the Tiger Tail service pages. Every page got its primary keywords and monthly search volumes confirmed before any content was briefed.

Page URL                              Primary Keyword                    Monthly Searches

/services/ai-audit-strategy            ai strategy consultant                      880
/services/ai-audit-strategy            ai readiness assessment                     720
/services/ai-audit-strategy            ai implementation consultant                390
/services/ai-audit-strategy            automation consultant                       480
/services/workflow-automation          business process automation services        320
/services/custom-ai-development        custom ai development company               480
/services/custom-ai-development        ai integration services                     590
/services/growth-engineering           ai marketing automation                     720
/services/growth-engineering           ai lead generation agency                   110
/services/ai-training-enablement       corporate ai training                        40

And here is the same mapping for the industry pages:

Page URL                              Primary Keyword                    Monthly Searches

/ai-for-legal                          ai for law firms                          1,300
/ai-for-real-estate                    ai real estate agent                        590
/ai-for-real-estate                    ai for real estate agents                   480
/ai-for-healthcare                     healthcare workflow automation              170
/ai-for-finance-accounting             ai for accounting firms                      70
/ai-for-home-services                  ai for contractors                          110
/ai-for-legal                          legal document automation                   170
/ai-for-healthcare                     ai for medical billing                       90

Looking at this data together, the legal page stands out immediately. “Ai for law firms” at 1,300 searches per month is the single highest-volume keyword across all pages on the site. That tells you the legal cluster needs serious depth in the blog to give that page the authority it needs to compete.

The corporate AI training page, on the other hand, targets “corporate ai training” at just 40 searches per month. That is a low-volume keyword but the commercial intent behind it is very high. Someone searching that phrase is almost certainly a business ready to spend money on training. Low volume does not mean low value.

How Search Volume Shapes Priority, Not Just Selection

This is the part most keyword guides miss. Search volume is not just a filter for deciding which keywords to target. It is an input for prioritising which content to build first and how much of it you need.

A page targeting a keyword with 1,300 monthly searches needs more supporting blog content around it than a page targeting 40 monthly searches. Not because the second page matters less, but because Google needs to see more topical depth before it will trust a new domain with a high-volume, competitive keyword.

Volume Range      What It Means                          Content Priority

1,000+            High demand. High competition.             Deep cluster needed.
                  Big brands likely dominating page 1.       10+ supporting posts.
                  New domain needs time and authority.

300 to 999        Solid demand. Beatable competition         Strong cluster needed.
                  with quality content and good structure.   8 to 10 supporting posts.

100 to 299        Moderate demand. Often less competitive.   Medium cluster.
                  Good early target for a new domain.        6 to 8 supporting posts.

10 to 99          Low volume. Often high commercial intent.  Focused cluster.
                  Worth targeting if buyer intent is clear.  5 to 6 supporting posts.

Under 10          Very niche. May still be worth it          Evaluate carefully.
                  if the buyer value per conversion is high. Single post may be enough.

This framework shaped the entire cluster structure for the project. The legal cluster targeting 1,300 searches got ten posts. The AI training cluster targeting 40 searches also got ten posts, but those posts are written differently. More specific, more technical, more conversion-oriented, because the person reading them is further along in their decision.

Intent Is More Important Than Volume

Search volume tells you how many people are searching. Search intent tells you why. Getting the intent wrong is worse than targeting a low-volume keyword because it means you are attracting the wrong people even when you do rank.

Every keyword in the Tiger Tail mapping got an intent classification before it was assigned to a page. The classification is simple but it matters every time.

Intent Type       What the Searcher Wants                 Right Page Type

Informational     Learning about a topic.                     Blog post.
                  Not ready to buy yet.
                  Example: "what is ai readiness assessment"

How-To            Looking for a process or steps.             Blog post or guide.
                  Example: "how to automate workflow"

Commercial        Researching providers or solutions.         Service or industry page.
                  Getting close to a decision.
                  Example: "ai strategy consultant"

Comparison        Evaluating options.                         Blog post or landing page.
                  Example: "make vs zapier vs custom automation"

Transactional     Ready to buy or contact.                    Service page with clear CTA.
                  Example: "hire ai implementation consultant"

A keyword like “what is an ai readiness assessment” is informational. It belongs in the blog as a post that educates the reader and links to the service page at the end. A keyword like “ai readiness assessment” with no qualifier is commercial. Someone typing that is likely comparing providers. It belongs on the service page itself.

Those two keywords look similar. They would land on completely different pages in a well-structured site. Getting that distinction right is what separates a site that converts from one that attracts traffic that never does anything.

The Before and After of Keyword Mapping

Here is what the approach looks like without mapping versus with it:

WITHOUT KEYWORD MAPPING

"Let's write a blog about AI for law firms."
"Let's write about what an AI consultant does."
"Let's cover AI pricing."

Result: Random posts. No page authority built.
        Service pages get no support.
        Blog competes with its own pages.
        Nothing ranks for anything meaningful.


WITH KEYWORD MAPPING

"ai for law firms" (1,300/mo, commercial) → /ai-for-legal service page
"how small law firms use ai" (informational) → blog post in legal cluster
"ai contract review" (informational/how-to) → blog post in legal cluster
"legal document automation" (170/mo, commercial) → /ai-for-legal page
"ai and billing ethics law firms" (informational) → blog post in legal cluster

Result: Service page targets commercial keywords.
        Blog cluster builds topical authority around it.
        Every post links back to the parent page.
        Google sees depth and relevance. Rankings follow.

The difference is not subtle. In the first approach, a business is just publishing. In the second, every piece of content has a specific job to do and a specific place in the architecture.

What Good Keyword Mapping Produces

By the time the keyword mapping was done for the Tiger Tail project, every page on the site had a clear primary keyword, a confirmed search volume, an intent classification, and a list of supporting blog topics that would feed it authority over time.

That groundwork meant every brief written after it had a reason to exist. Not just “here is a topic someone might find interesting” but “here is a keyword a real person searches for, here is the page it supports, here is how it fits into the cluster that will eventually rank the parent page.”

What Comes Next

With the keyword map in place, the next step was research. Not the generic kind where you read a few articles and summarise them. Proper data-backed research using Perplexity Sonar that produced real statistics, named sources, and proof points for every single post across all 110 briefs.

That process is what I cover in the next post: how I use Perplexity Sonar to research blog topics with real data.

If you want to talk through what keyword mapping would look like for your own website, book a call. I can usually tell within the first conversation whether a site’s content architecture is working for it or against it.

See how I approach SEO strategy →

Book a free 30-minute call →

Dhruv is an SEO consultant working with business owners, founders, and agencies. If organic search is not delivering for your business, this is where to start.

dhruv-seo.online

If you have not read the earlier posts in this series, start here to understand why most blogs fail and here for the competitor research approach.

Two Problems That Are Actually the Same Problem

The first problem is not knowing what to write about when competitor data is not an option. Either nobody in the niche is blogging with measurable results, the industry is too specific for competitor keywords to be meaningful, or the business simply wants to create content on its own terms rather than chasing what others are ranking for.

The second problem is that even when topic ideas exist, they never become a consistent publishing schedule. A blog calendar gets created in a meeting, lives in a Google doc for two weeks, and then quietly disappears. Publishing becomes irregular. Months go by. The blog never builds the compounding value it was supposed to.

These two problems look different on the surface but they come from the same place: there is no system underneath the content. The persona approach solves both at once. It gives you a method for generating months of relevant topics and a calendar that is specific enough to actually use.

Why Persona-Driven Content Works Differently

Keyword research tells you what people are searching for. Persona research tells you why they are searching for it and what they actually need when they get there.

Both matter. But for building long-term authority and genuine trust with your audience, persona-driven content wins. It speaks directly to the person behind the search rather than just matching the query. Readers feel understood. That is what makes them come back, share the content, and eventually reach out.

Content written without persona thinking tends to feel generic even when it is technically accurate. It covers the topic but it does not resonate with anyone in particular. It gets read and forgotten. It builds no relationship and no trust.

What a Buyer Persona Actually Is

A buyer persona is a detailed profile of an ideal customer. Not a demographic summary. A real picture of the person: their job role, their industry, what their day looks like, what keeps them stuck, what they are trying to achieve, what they search for when they have a problem, and what kind of content actually helps them make decisions.

Most businesses either have no defined personas or have ones that are too vague to be useful. Something like “marketing manager, 30 to 45, works at a mid-sized company” is not a persona. It is a demographic filter. A useful persona includes the specific frustrations, the exact questions they type into Google, and the outcomes they are trying to reach.

The good news is that you do not need a formal persona document to start. A rough description from someone who knows the customers well is enough to build on.

Why Blogs Without Persona Thinking Fail to Build Authority

The content is technically correct but feels like it could have been written for anyone. There is no consistent point of view. The topics jump around instead of building a coherent body of knowledge in one area. Readers do not feel like the brand actually understands their situation. They read, get the information they needed, and leave without ever considering the business behind the content.

Trust does not come from being informative. It comes from being specifically relevant to the person reading. That only happens when the content was built around a real understanding of who that person is.

The Full Process: From Personas to Published Calendar

I am building a blog content strategy for [Brand Name].
The website is [URL].
The brand [describe what it does and who it serves].

The buyer personas are:
[List each persona with job title or description]

Target location: [country or region]

Please use deep research to give me a detailed breakdown
of each persona including:
- Who they are
- Their biggest pain points and daily challenges
- The questions they commonly search for online
- The type of information they look for before making decisions
- What content would genuinely help them

After completing the research, generate [number] blog topic
ideas directly based on the pain points and questions you found.

Topics should be educational and informational, not promotional.
Format the topics as a numbered list.

Using the blog topics listed above, please create a blog
calendar for [Brand Name].

Starting month: [month and year]
Blogs per month: [number]
Total duration: [number of months]

For each blog topic include:
- The topic title
- A brief content outline covering the key points
- The target buyer persona this post is written for
- A suggested publish date

Format this as a table with four columns:
Topic Title | Content Outline | Persona | Publish Date

So I can copy it directly into a spreadsheet.

What the Calendar Actually Gives You

The obvious output is a publishing plan. But the less obvious output is the removal of decision fatigue. One of the main reasons blogs become inconsistent is that every publishing cycle starts with the question of what to write next. That question never fully gets answered, the deadline passes, and the blog goes quiet for another month.

With a calendar in place, that question is already answered for the next six months. The only job left is execution. That shift from deciding to doing is what makes consistent publishing actually happen in practice rather than just in plans.

For consultants and agencies, the calendar also works as a client deliverable. It demonstrates strategic thinking beyond just writing. It shows that the content has a reason to exist, a defined audience, and a structure that builds toward something over time.

Why Consistency Is the Most Underrated Factor in Blog SEO

One blog post almost never produces meaningful results on its own. SEO from blogging is a compounding activity. The value builds as more posts are published, more keywords get covered, and Google increasingly recognises the website as a trustworthy source on a specific set of topics.

A business that publishes four well-targeted posts per month for six months has 24 pages competing for organic traffic. A business that publishes randomly has gaps, inconsistency, and a much weaker topical authority signal. Google notices the difference.

The calendar is not just a content planning document. It is the system that makes compounding SEO possible by turning irregular publishing into a predictable habit.

Competitor Approach vs Persona Approach — Which One Is Right

The competitor approach works best when there is proven search demand in the niche, multiple competitors are already getting blog traffic, and the primary goal is capturing a share of existing organic traffic as efficiently as possible.

The persona approach works best when the industry is niche or specialist, competitors are not actively blogging, the business wants to build a distinct voice, or the goal is long-term audience trust rather than short-term traffic volume.

The strongest content strategies use both. The competitor approach fills the calendar with high-demand topics that have a direct path to organic rankings. The persona approach fills the gaps with audience-first content that builds deeper relevance and trust over time. Together they cover both the traffic goal and the authority goal that I wrote about in the first post in this series.

Want Help Building This for Your Business?

A blog calendar built on real persona research gives you months of direction in a single session. But the research is only as good as the understanding of the audience behind it. If you want to build a content strategy that is actually tailored to your customers and your business goals, this is something I work through with clients directly.

Whether you need a full content strategy, help with SEO, or a conversation about what your blog should actually be doing for your business, book a call and we can get into the specifics.

See how I approach content and SEO strategy →

Explore done-for-you SEO →

Book a free 30-minute call →

Dhruv is an SEO consultant working with business owners, founders, and agencies. If you want a blog that actually builds something, this is where to start.

dhruv-seo.online

Published by Dhruv — SEO Consultant for Agencies & Businesses

TL;DR

I built an AI workflow to write long-form blog content at scale. It took four rounds of iteration, broke in four different ways, and never fully hit the original word count target. But it now produces a clean, on-brand 1,300 to 1,450-word draft in about 3 minutes, costs $0.20 per post, and needs only 5 to 10 minutes of human editing. Here is exactly what happened, what failed, and what I actually ended up with.

I want to start with something most AI content posts will not tell you: the first version was pretty bad.

Not unusable. Not embarrassing. But nowhere near what I needed. And the gap between “it kind of works” and “I would actually publish this” took a lot longer to close than I expected.

This is the honest version of that story.

What I Was Trying to Build

The goal was a repeatable system for long-form blog content. Each post needed to land between 1,950 and 2,100 words, follow a specific structure, carry 4 to 5 internal links placed naturally, stay within short readable paragraphs, and sound like it was written by someone who actually knows the industry. No filler. No generic advice dressed up as insight.

Before this system existed, one blog post took 4 to 5 hours of combined writing and editing time, and cost anywhere from $50 to $150 depending on who was writing it. Across 52 posts a year, that is 208 to 260 hours and up to $7,800. Those numbers made it very easy to justify building something better.

Round One: It Just Stopped Halfway

The first approach was simple. Write one prompt, get one blog post. It seemed reasonable.

What actually happened was that the model would write a decent introduction and a solid first section, then quietly start losing structure. By the time it reached the middle of the article, paragraphs were getting longer, sections were getting thinner, and the whole thing just stopped around 950 to 1,200 words. About half of what I needed.

The links were almost never there. When they were, they appeared once, usually in the wrong section. Paragraph length crept up to 90 to 150 words regularly. The brand voice held for the first third and then drifted.

The lesson from round one was simple: a single prompt cannot hold the shape of a long article. The model does not forget what you asked for, but it does lose discipline over distance.

Round Two: Split It Into Two Passes

The fix seemed obvious. Break the article into two prompts. First prompt handles the introduction and the first three sections. Second prompt handles the remaining sections, the checklist, and the close.

This was better. The structure got more consistent and the output felt more controlled. But two new problems appeared almost immediately.

First, the total word count only reached 1,250 to 1,450 words. Still short. Second, the second prompt kept repeating the article title at the top of its output, even when I told it not to. Every single time. It was one of those things that seemed easy to fix with a clearer instruction, and then kept happening anyway.

Two-pass generation was clearly the right direction. But prompting alone was not going to solve everything.

Round Three: Stop Trying to Fix Everything in the Prompt

This was the turning point, and it came from accepting something uncomfortable: some problems are easier to fix with code than with words.

The internal linking issue was a good example. Asking the model to place links naturally and distribute them across the article produced inconsistent results. Sometimes it worked. Sometimes all four links landed in the same paragraph. Trying to write a prompt that reliably fixed this was a diminishing returns exercise.

So instead, I introduced ALL CAPS placeholders inside the generated content. The model would write something like “working with a NYC EVENT PRODUCTION team means…” and a post-processing script would replace that placeholder with the actual hyperlink after generation. Clean output during writing, correct links in the final version.

The duplicate title problem got solved the same way. A function scanned the combined output for any repeated H1 and removed it automatically. Two lines of code that worked every time, compared to prompt instructions that worked most of the time.

Round Four: Tighten Everything and Accept the Tradeoff

The final version of the system pulled everything together. Two-pass GPT-4o generation with lower temperature for consistency, maximum token allowance to reduce early stopping, strict paragraph length instructions, the duplicate title remover, and the link post-processing step.

Here is what stable output looked like:

• Word count: 1,320 to 1,450 words per post
• Generation time: around 3 minutes
• Human editing time: 5 to 10 minutes
• Internal links: 4 to 5, placed naturally
• Paragraph length: almost always under 70 words
• Cost per post: approximately $0.20
• Brand voice: consistently strong

The one thing that did not get solved was the word count target. The original goal was 1,950 to 2,100 words. The system never reliably got there. After a while, I stopped trying to force it and started asking a different question: is a consistent 1,380-word post that ships every time actually worse than a 2,000-word post that requires constant intervention?

For this workflow, the answer was no. The shorter, cleaner draft was more useful.

What the Numbers Actually Look Like

Compared to the manual baseline, the impact was significant across every metric that mattered.

• Annual writing time dropped from 208 to 260 hours down to roughly 26 to 39 hours including editing
• Cost per post dropped from $50 to $150 down to about $0.20 in generation cost
• Annual content cost dropped from up to $7,800 to just over $10 in API spend
• Time saved across 52 posts: roughly 180+ hours per year

The word count shortfall meant each post needed a bit more from the human editor to expand key sections. But that tradeoff was worth it given everything else the system got right.

The Four Things That Kept Breaking

If you are building something similar, these are the failure modes worth knowing about before you hit them yourself.

Word count collapse. Single-pass prompts consistently fell short. The model does not run out of knowledge, it just loses structural discipline over long outputs. Two-pass generation is the practical minimum for anything over 1,200 words.

Link clustering. Left to itself, the model tends to place multiple links close together or in a single section. Post-processing is far more reliable than prompt instructions for solving this.

Title duplication. The second prompt often repeated the article title even with explicit instructions to skip it. A simple programmatic check fixed this permanently.

Paragraph bloat. Explanation-heavy sections regularly ballooned past 140 words. Explicit limits help, but a human pass is still the most reliable way to catch this.

The Actual Takeaway

The system did not produce the perfect 2,000-word automated blog post I originally planned for. What it produced was something more practical: a dependable, brand-consistent draft that is ready for a light human edit in under 10 minutes, at a cost that makes weekly publishing genuinely viable for any business.

The biggest mindset shift was treating the human editor as part of the system rather than as evidence that the system failed. The AI handles the structural heavy lifting. The editor catches what slipped through. Together, they produce something neither would get to as quickly alone.

If you are trying to build the same thing, start with two-pass generation, move formatting fixes into post-processing early, and pick a word count you can hit consistently rather than one you can hit occasionally.

Want to Talk Content Systems or SEO Strategy?

This kind of workflow thinking sits at the intersection of content operations and SEO. If you are trying to build something similar for your agency or business, or if you just want to talk through what a content system could look like for your specific situation, I am happy to get into it.

Read more on the blog →

Book a free 30-minute call →

Dhruv is an SEO consultant working with agencies, founders, and business owners. 500+ projects. 6+ years. No fluff.

dhruv-seo.online