If you publish a blog post twice a year and nothing happens, the frequency is almost certainly the problem. The honest answer to how often should a small business blog is this: at least weekly if you want measurable results within a year, and at whatever pace you can sustain without stopping. A cadence you keep beats a cadence you abandon.
That answer needs unpacking, because the reason frequency matters is not what most people assume. Google does not reward you for being busy. It rewards you for covering your subject properly, and covering a subject properly takes a lot of articles. Frequency is simply how fast you get there.
A blog with six posts written over three years is not really a blog. It is six isolated pages, each competing alone against sites that have answered the same question and forty related ones.
Three things work against the occasional publisher:
This is also why a burst of activity followed by silence does not work. Ten posts published in January and nothing after that gives you a brief crawl bump, then decay. The signals that compound, coverage, freshness, internal links between related articles, all need a steady drip rather than a flood.
Rather than a single magic number, it helps to look at what each level of effort actually buys you. The examples below use an IT support company, but the pattern holds for accountants, trades and online shops alike.
Twelve articles a year is the minimum that counts as an active blog. It keeps your site from looking abandoned, gives you something to share with clients and on LinkedIn, and slowly builds a small library.
What to expect: modest. After a year you have twelve pages, each targeting one question. A few will rank if they hit low-competition queries, and you may see a trickle of enquiries. What you will not see is momentum. Twelve pages is rarely enough to establish topical coverage in any subject, so each article fights on its own.
This cadence suits businesses where the blog is a supporting act, not a growth channel. If that is you, one good post a month is a perfectly respectable choice. Just set your expectations to match.
Weekly publishing is where most small businesses should aim if they want the blog to generate leads. Fifty articles a year is enough to cover a subject area properly. An IT support firm publishing weekly can, within a year, answer most of the questions a prospective client types into Google before they ever pick up the phone.
What to expect: the first three months usually feel like shouting into a void. Somewhere between month four and month eight, if the articles genuinely answer specific questions, individual posts start ranking and the site as a whole starts being crawled more often. Traffic tends to arrive unevenly, a few articles doing most of the work, which is normal.
The catch is discipline. Weekly publishing done by the owner in spare evenings is the cadence most likely to collapse by March. If you choose this level, decide up front who writes, who approves, and what happens in a busy month.
Rule of thumb: pick the fastest cadence you could still keep during your busiest month of the year. If a quiet January pace would collapse in June, it is the wrong pace.
Daily publishing sounds excessive, and for a human writer it usually is. But the logic is straightforward: if your buyers ask three hundred distinct questions across a purchase journey, daily publishing answers all of them within a year instead of within six.
What to expect: faster crawling, faster coverage of the long tail, and a much better chance of being the site that AI assistants quote, because tools like ChatGPT and Perplexity pull from pages that state answers plainly, and you simply have more of them. We covered that mechanism separately in \"How to Get Mentioned in ChatGPT Answers: A Practical Guide\".
The caveats matter. Thirty poor articles a month will do nothing except clutter your site; volume only helps when each piece targets a real question and answers it honestly. Daily publishing at agency rates is also financially out of reach for most small firms, which is why it has historically been the preserve of large content teams. Whether you use software, a writer or your own keyboard, the quality bar does not move.
Whichever cadence you choose, the steady version outperforms the bursty version of the same total volume. A post a week for a year serves you better than fifty-two posts in a fortnight followed by silence, for three reasons:
A blog is not a campaign. It is a habit with a compounding return.
Work through this honestly:
Key takeaway: results scale with total useful articles published, not with enthusiasm per article. Twelve months at any steady cadence beats a heroic fortnight every time.
Helio SEO exists for businesses that want the weekly or daily end of this scale without hiring anyone: it plans a month of buyer questions, drafts one article a day in your voice, and publishes each one after you approve it. If a post a month written by you is the right cadence for your business, you do not need us for this one. If the maths above points to a pace you cannot sustain by hand, that is the gap we fill.
"},{"slug":"get-mentioned-in-chatgpt-checklist","title":"How to Get Mentioned in ChatGPT When a Competitor Already Is","meta_description":"How to get mentioned in ChatGPT: how assistants choose sources, why plain specific pages get quoted, and a checklist to apply this week.","published_at":"2026-08-12T18:33:58.317001","body_html":"A customer telling you \"ChatGPT recommended you\" is a pleasant surprise. A customer telling you it recommended a competitor is a problem you want to fix quickly. If you are wondering how to get mentioned in ChatGPT, the honest answer is less mysterious than most of the advice circulating: assistants recommend businesses whose websites state, in plain words, exactly what they do, for whom, where, and on what terms. Your competitor probably got the mention because their site says those things and yours implies them.
We covered the broader mechanics in How to Get Mentioned in ChatGPT Answers: A Practical Guide. This piece is narrower: it explains why the competitor is winning right now, and gives you a checklist you can apply to your own pages this week.
There are two routes into an assistant's answer. The first is training data, the enormous snapshot of the web the model learned from. You cannot influence that on any useful timescale, so set it aside.
The second route matters far more. When someone asks a recommendation question, something like \"best IT support company for a small accountancy firm in Leeds\", modern assistants usually run a live web search, read a handful of pages, and summarise what they find. ChatGPT does this with web search enabled. Perplexity searches on almost every query. Google's AI Overviews are built directly on Google's own index.
That changes the question entirely. You are not trying to get inside a model's head. You are trying to be one of the handful of pages a search retrieves and an assistant can confidently quote. That is a writing and website problem, and it is fixable.
Open the competitor's site next to yours and the difference is usually obvious once you know what to look for. Their service page says something like \"we provide IT support for firms of 10 to 50 staff across West Yorkshire, with a one-hour response target and pricing from a stated monthly figure per user\". Yours says \"award-winning technology solutions tailored to your business\".
An assistant can quote the first sentence. It cannot do anything with the second. Assistants do not read between the lines, infer your pricing, or guess your service area from your phone number's dialling code. If a fact is not written on the page, then as far as the assistant is concerned it does not exist.
Rule of thumb: if a stranger could not answer \"what do they do, for whom, where, and roughly at what price\" after sixty seconds on your site, an AI assistant cannot either.
Marketing fluff fails for a second reason too. Retrieval works by matching the searcher's question to text on a page. Nobody asks an assistant for \"tailored solutions\". They ask specific, long-tail questions: how much does managed IT support cost, who covers Microsoft 365 migrations near me, what is a reasonable response time. Pages written in the language of those questions get retrieved. Pages written in the language of a brand brochure do not.
A single page that plainly answers \"what does outsourced IT support cost for a ten-person firm?\" will outperform a general services page for that question every time. Give the direct answer near the top, then the detail: what drives the price up or down, what is included, what is not. One question, one page, answered properly.
Somewhere easy to find, in plain HTML text rather than an image or a PDF, your site should state: exactly what you offer, who it suits (company size, sector, situation), the geographic area you serve, indicative pricing or at least a pricing model, and anything measurable you commit to, such as response times. If you cannot publish exact prices, publish a range or a \"from\" figure. Vagueness here is what hands the mention to someone else.
Check your robots.txt file, the small text file at yoursite.com/robots.txt that tells crawlers what they may fetch. Some sites, often on a developer's well-meaning advice, block GPTBot, OAI-SearchBot or PerplexityBot outright. Blocking them removes you from consideration. Also confirm your key pages appear in Google's index, since AI Overviews and several assistants lean on it, and that your important text is served as real HTML rather than loaded only by JavaScript, which some crawlers handle poorly.
Your business name, location, services and contact details should match across your site, your Google Business Profile and any directories you appear in. Assistants cross-reference. Inconsistency does not disqualify you, but it lowers confidence, and confidence is what gets a business named rather than described vaguely.
Worked example: an IT support firm replaces \"trusted technology partner for growing businesses\" with \"managed IT support for firms of 5 to 50 staff in Manchester and Cheshire, priced per user per month, with remote response inside one hour\". Nothing else changes. That single sentence gives an assistant everything it needs to match the page to a real question and quote it.
Skip anything sold as a shortcut. Hidden text addressed to AI models, paid \"get listed in AI search\" schemes, and keyword-stuffed pages either do nothing or actively hurt you. There is no submission form for ChatGPT, and no one can guarantee a mention. The only durable lever is the boring one: clear, specific, honest pages that answer real questions, on a site crawlers can read.
It is also worth saying that one rewritten page will not transform things overnight. Retrieval favours sites with depth on a topic, which is why the businesses that get mentioned consistently tend to be the ones publishing steadily, not the ones that fixed a single page once.
Helio SEO exists for the last item on that checklist and everything after it: it scans your site, works out the specific questions your buyers ask, and writes and publishes one plainly worded article a day in your voice, with every draft approved by you first. If you have the time to write those pages yourself, the checklist above is all you need. If you do not, that is the gap we fill.
"}]}