Most “AI content marketing” advice still assumes you will run the project: choose topics, brief, prompt, edit, fact-check, optimise, format, publish, measure, repeat.
If you are a team of one, that is the problem.
This post gives you an operating model for shipping quality posts for 90 days with minimal coordination. It also gives you a test for tools: do they remove steps, or create new ones.
Define the job in 90 days: one outcome, one proxy
If you start with “we need more posts”, you will get busier and stay stuck. Start with an outcome you can measure in a quarter.
Step 1: pick one primary goal
Choose one goal for the next 90 days:
- Qualified pipeline created
- Demos booked
- Trials started
- Email sign-ups
Be specific about ownership. “Pipeline” only works if you can see content-assisted conversions in your CRM or analytics.
Step 2: pick one proxy metric you can check weekly
Pick one proxy metric that moves faster than revenue:
- Organic sessions (Google Analytics 4)
- Search impressions and clicks (Google Search Console)
- Top 20 rankings for a defined query set (Ahrefs, Semrush)
Write the constraint as one sentence and put it at the top of your backlog:
In the next 90 days, content exists to increase trials. We will track Search Console impressions for our target query set as the proxy metric.
That sentence prevents a common failure mode: shipping articles that read well but do not map to a goal.
Set a publishing promise you can keep
Cadence is an operations decision. Pick a promise that survives sick days, launches, and end-of-quarter chaos.
Sensible defaults
- 1 post per week if content is your main growth lever and your domain already has some authority
- 3 posts per month if you also own lifecycle, paid, partnerships, events, or product marketing
Make review bandwidth explicit
If your CEO “approves everything” but takes 10 days to reply, you do not have weekly publishing. Build a workflow that can publish without them for low-risk posts.
Pick one content focus for the quarter
A team of one wins by repetition. Pick one focus and stick to it for 90 days.
Choose one lane
- Problem-led: pain points and “how do I…” queries
- Product-led: use cases, workflows, implementation guides
- Category-led: comparisons, point of view, why the category exists
A simple rule of thumb
- If buyers do not understand the problem, go problem-led.
- If buyers understand the problem but not your approach, go product-led.
- If you are trying to name the thing, go category-led.
Mixing all three is what big teams do when they have editors and SMEs on tap.
Audit your workflow: where AI usually adds work
Most AI setups speed up drafting but add overhead in the slow parts: tone, accuracy, SEO, formatting, and publishing.
Step 1: trace your last three posts backwards
For each post, list every step from idea to publish:
- Topic selection
- Brief
- Outline
- Draft
- Rewrite to match voice
- Fact checking and sourcing
- On-page SEO (title, headings, internal links, metadata)
- Images
- Formatting in the CMS
- Review, comments, rewrites
- Publish, sitemap, distribution
- Performance check and updates
Now mark what took the longest.
For most solo marketers, the time sinks are not “writing words”. They are coordination and clean-up.
Step 2: list the hidden AI overhead
If your current approach looks like this, you have not reduced work:
- Prompting and re-prompting to get the structure right
- Editing because the tool does not hold your voice between sessions
- Verifying claims because the draft is confidently vague
- Retrofitting SEO after the draft exists
- Copying and pasting into WordPress, Webflow, or a headless CMS, then fixing formatting
Put your blog on autopilot
Highway researches, writes, and publishes SEO content for you. Get early access.
No spam, unsubscribe anytime.
One rule for tools: they must remove steps
Put this at the top of your process document:
If a tool does not remove steps, we do not add it.
What step removal looks like
Good (fewer steps):
- No per-article prompting
- No separate SEO checker for basic on-page work
- No separate doc tool for reviews
- Publishing happens in the same system that drafts and schedules
Bad (faster broken process):
- Drafts in Google Docs, approvals in Slack, tasks in Notion, SEO in another tool, then manual CMS work
- You still manage freelancers, chase edits, and reconcile versions
If the system does not cover strategy to publish, it is an assistant, not an operating model.
A realistic solo marketer pipeline (brief to publish in one loop)
If you cannot point to one backlog and one status for each article, you do not have a pipeline.
Step 1: use one backlog with status gates
Use one place (Notion, Airtable, Linear, Trello, or your content platform). Add status gates that match reality:
- Idea
- Planned
- Drafted
- Ready for review
- Approved
- Scheduled
- Published
- Updating
Required fields:
- Target query (or cluster)
- Primary goal alignment (pipeline, demos, trials, sign-ups)
This prevents random acts of content.
Step 2: define minimum viable inputs
Do not build a briefing process you cannot maintain.
Minimum viable inputs that make quality repeatable:
- Target customer: role, company type, constraints
- Positioning notes: what you do differently, what you do not do
- Compliance constraints: forbidden claims, required disclaimers
- Voice examples: 3 to 5 posts that match, plus 1 that does not
Voice is not a paragraph in a brand doc. It is examples plus constraints.
Step 3: automate the boring middle
This is the part that kills consistency:
- Finding topics from gaps (what you should rank for but do not)
- Aligning outlines to search intent
- Drafting in your voice
- On-page SEO (headings, FAQs when relevant, metadata)
- Internal links (new to old, old to new)
- Formatting and publishing
If your workflow still depends on a chatbot and a copy-paste into the CMS, you are still doing the job.
Governance for small teams: guardrails, not bureaucracy
Governance is what lets you publish without worrying that every post creates legal or brand risk.
Step 1: write a one-page content policy
Keep it to one page and make it enforceable.
Include:
- Claims that require sources (stats, market sizes, legal assertions, performance guarantees)
- Claims you cannot make (“guaranteed results”, unverified competitor claims)
- Naming conventions (product names, feature capitalisation)
- Style rules (British English, contractions or not, stance on jargon)
- Linking rules (when to link out, nofollow policy, competitor links)
Step 2: use quality bars that take 5 minutes to check
A checklist that takes 45 minutes will be ignored.
Use five fast bars:
- Structure: clear intro, logical headings, skimmable sections
- Intent match: answers the query plainly, not a product brochure
- Unique insight: one point that is clearly yours (framework, process, numbers, lesson)
- Proof points: examples, screenshots, quotes, or sourced stats where needed
- CTA alignment: matches intent (template for how-to, demo for comparisons)
Step 3: permissions and audit trails
If approvals live in Google Docs comments and Slack threads, publishing slows down.
Minimum roles:
- Owner: backlog, policy, schedule
- Reviewer (SME): accuracy, missing steps, edge cases
- Approver: brand and risk
- Viewer: visibility without drive-by edits
Audit trails matter when someone asks, six months later, why you said something.
Put your blog on autopilot
Highway researches, writes, and publishes SEO content for you. Get early access.
No spam, unsubscribe anytime.
Approval flows that do not block publishing
Approvals should reduce risk, not stop output.
Step 1: choose a default model
Two models cover most teams of one:
- Default auto-publish: posts publish unless flagged as sensitive
- Review window: posts publish automatically after 48 hours unless vetoed
Use reviews for:
- Pricing, guarantees, performance claims
- Regulated topics (finance, health, security claims)
- Competitive comparisons
- Customer stories and logos
Step 2: split reviews by responsibility
Stop asking one person to “review the post”. Split it:
- SME review: technical accuracy, missing steps, unsafe simplifications
- Brand and risk review: tone, forbidden claims, positioning
Each reviewer gets a short checklist.
Step 3: standardise reviewer actions
Reviewers get three actions:
- Approve
- Request specific edits (must be concrete)
- Block with reason (must cite policy or factual error)
No open-ended comment threads.
Cadence planning that survives real life
Consistency is an operations problem.
Step 1: plan monthly, publish weekly
Use a simple rhythm:
- Week 1: plan topics, groom backlog, confirm priorities
- Weeks 2 to 4: publish and refresh
Maintain a buffer of two ready-to-publish posts.
Step 2: mix effort levels
A practical portfolio for a solo marketer:
- 1 pillar post per month (1,800 to 2,500 words)
- 2 to 3 supporting posts per month (800 to 1,400 words)
Supporting posts come from:
- Sales call questions
- Comparison queries (“X vs Y”, “alternatives to…”, “best tools for…”)
- Implementation guides (“how to set up…”, “checklist for…”, “template for…”)
Step 3: reserve one slot per month for refreshes
Once you have 20 to 50 posts, refreshes are often higher ROI than net-new content.
Use data to pick targets:
- Search Console: high impressions, low CTR (title and intent mismatch)
- Search Console: high impressions, low average position (needs depth and links)
- GA4: posts with traffic but no conversions (CTA mismatch)
What to automate vs what to keep human
Automate repeatable work. Keep decisions and accountability human.
Automate mechanical tasks
- Site crawl and content gap detection
- SERP and competitor analysis (common headings, angles, missing subtopics)
- Draft generation in your voice
- Internal linking suggestions
- Scheduling and publishing
Keep human work where it matters
- Positioning: what you stand for, what you disagree with
- Product nuance: edge cases, trade-offs, implementation reality
- Customer proof: case studies, quotes, numbers
- Final sign-off on regulated or high-risk claims
AI can draft. It cannot be accountable.
What “self-driving content” means in practice
Autonomy is not “we use ChatGPT”. Autonomy is an end-to-end pipeline that runs on a schedule without per-article steering.
A self-driving system does this:
- Crawls your site and maps existing content
- Finds gaps and prioritises opportunities
- Researches SERPs and competitors
- Writes in your voice from calibration, not prompting
- Publishes on schedule (metadata, internal links, formatting done)
No prompts. No copy-paste into your CMS. No project management.
Guardrails that make autonomy safe
- Voice calibration once (using your examples and terminology)
- Policy guardrails (sources required, forbidden claims, naming conventions)
- Approval workflows (auto-publish for low-risk, review windows for sensitive)
Success criteria a team of one should care about
Measure leverage and compounding results, not “posts generated”:
- One backlog, one system
- Fewer tools (no separate drafts, SEO, publishing, and review layers for basic output)
- Fewer meetings (monthly planning, lightweight reviews)
- Cadence holds for 90 days
- Performance improves over time (refreshes happen, winners get expanded)
If you hit those, content stops being a guilt pile and becomes a reliable channel.
Where Highway fits
Highway is a self-driving content platform: it crawls your site, finds gaps, researches competitors and trends, writes in your voice, and publishes on a schedule.
No prompts. No project management. No writers to hire or manage.
If you want to see whether a self-driving model would work for your site, the only useful test is operational: can you get from “nothing planned” to “three posts published this month” without adding meetings, tools, or headcount.
Put your blog on autopilot
Highway researches, writes, and publishes SEO content for you. Get early access.
No spam, unsubscribe anytime.