Should My Business Still Have a Blog?

I'm a solo consultant. Everyone says AI search is killing blogs, and I don't have time to write. Is a blog still worth it for a business like mine, or is it a waste of hours?

Yes, when each post answers a question a client asked. Then one post works three times: on the site, in a letter to your list and in social posts. AI tools read it too.

AI search changes who reads the page, not whether it's worth writing. A model reads the same post a prospect does, and it can only quote a clear answer.

The hours are the real problem. They shrink once the point and the typing are two separate jobs.

A Post Answers One Client Question

Clients still come through people. The post is what they read before they write to you.

Mine came that way: an intro from a client, a peer, a DM on X. When a client introduced me to someone new, that person read my site first.

That's the job a blog does well for a consultant: answering the person who was sent over, before the first call.

So a post earns its place by answering a client's question. One client asks it on a call. The next one types it into Google or ChatGPT, and the answer you gave once keeps answering.

My own Q&A collection runs on this rule: answers to real founder questions, grouped by topic.

The catch is still the hours.

AI Drafts, the Answer Stays Yours

AI can draft. The question, the answer and the examples have to be yours.

The expertise is there; the time to write isn't. So split the point from the typing. You bring a note, a voice memo or a question answered on a call, and a draft comes back for approval.

I learned where AI belongs on my own site. From March to June, a nightly pipeline turned notes from my work into posts, and output went up.

Then someone asked to read my writing, and I heard myself tell them not to take it seriously: it was AI slop.

I took the pipeline out in June.

Since then, each post passes one test: did I decide what this says, and have I read every line? Both yes, and it can ship, whoever typed it.

The test works for any business blog: AI drafts, and the ideas, the stories and the last read stay yours. Where Can AI Save My Business Time? covers the other jobs where that split pays.

Once a post is approved, it still has to reach people.

One Post, Published Everywhere

From one place: the site first, then the letter and the social posts cut from it.

Most small businesses run each channel in its own tool. A site builder, a newsletter tool, a social scheduler, a link-in-bio page: four logins, four bills, the same text pasted four times.

Consolidation fixes that with one place to publish. One approval sends the post to the site, the list and the social accounts.

Each channel still gets its own cut, written for its reader. A social post stands on its own, with the link as a footnote.

My own letter waits until there are three things to share, then links the posts since the last one. Each issue is also a page in Letters.

Over months, the posts pile up by date, and that's where a hub comes in.

A Hub Groups the Answers

One page per question your clients ask, linking the posts that answer it. That's a content hub.

A blog sorts posts by date, so last year's best answer sinks. A hub keeps it one click from the top. Search engines and AI tools find every answer from it.

One site I built runs on hubs like these: one for each question its clients ask, and one page that sells.

For a consultant, the hubs are the questions every prospect brings to the first call. That shape, clear pages that answer one question each, is most of what AI search asks for.

AI Search Reads the Same Page

AI search optimization starts with a fast page, clear headings and the answer early. llms.txt is optional.

GEO, AEO or AIO, the work is the same: make the site easy to read and to quote. Beyond the page itself, it has four parts, and I run all four on varstatt.com first:

1. Let AI crawlers in. My robots.txt admits every crawler, the AI search bots included, and names GPTBot, ClaudeBot, PerplexityBot and Google-Extended too. The robots.txt validator checks yours.
2. Serve markdown. In May, my homepage was 23,908 tokens as HTML and 2,130 as markdown: 11 times fewer for a model.
3. Publish an index, the llms.txt file.
4. Mark up the facts with schema for the business, its services, posts and FAQs.

The llms.txt file is markdown at the site root, following the llmstxt.org convention. It holds the site name, a short summary, then links with a short note each. Mine had 170 lines in May, generated from the same content as the site, so it can't drift.

It's cheap when it's generated. A hand-written one goes stale with the next new page.

What the Numbers Say About AI Search

The fear has a number behind it. Pew Research Center tracked the browsing of 900 American adults in March 2025. It published the result that July.

When Google showed an AI summary, people clicked a regular result in 8% of visits, against 15% without one. A link inside the summary got a click in 1%.

So search sends fewer clicks. The page keeps the job it had: answering the reader a client sent, and being clear enough to quote.

Google's own guide to AI features, updated in December 2025, says appearing in them needs no special files or markup. It also says indexing and serving aren't guaranteed.

AI tools decide what they quote; the site only decides whether they can read it. Whether yours can is what an audit checks.

How I Build the System

My business website week builds the site and the system to run it. It's the same system my own site runs on.

You can write yourself, or send notes and voice memos, and I draft by the week. AI helps me draft; the ideas and the stories stay yours, and nothing goes out without your approval. The price is on the homepage.

How I Check a Site

My technical SEO audit covers five areas, ranked by impact:

1. Crawl and index: robots.txt and the sitemap. Mine was wrong six times by hand, until I generated it from the content.
2. Speed: heavy images, scripts and fonts, fixed in the template instead of page by page.
3. Structure: schema for the business, its services and FAQs.
4. Share previews: the title, image and text a link shows on LinkedIn or X.
5. AI readability: the four parts above.

Some checks are free to rerun in my toolkit, like the sitemap validator and the share preview check. ChatGPT can list the usual issues on a page. It doesn't see your Search Console data or change the template.

The report is the cheap part, and the fixes are the work. Mine is one week per site, at the price on the homepage. It covers the check, the ranked list and, with access, the top fix.

The fixes after it run by the week. Agencies sell it under their own brand as technical SEO.

For a consultant's site, the check happens once. The posts are what keep the site worth crawling.

What the Owner Should Do Now

Write down the last five questions clients asked, and answer one a week. Once a week is a pace a small business can keep.

Two formats work best for a consultant or a small business. A case shows proof of work: a real problem and how you solved it. A Q&A teaches, explains your work in the open, and stays there to reuse the next time a client asks.

Each answer can start as a note or a voice memo after the call where the question came up. Publish it on the site first, then cut the letter and the social posts from it. When a few answers share a question, give them a hub.

The blog is worth the hours when it holds your answers, not when it chases traffic. The site and the tools behind it are in What Tech Does My One-Person Business Need?