Blogging statistic: effort, formats and results

Most blogging figures in circulation trace back to a handful of surveys, and few say who was asked, when, or what the number measures. This report lists the figures that hold up, each with its population, year and evidence type. Where the evidence cannot support a number, it says so.

The short version: the most recent effort figure that could be verified is a 2023 self-reported average of 3 hours 51 minutes to write a post. Publishing cadence is a 2025 self-reported distribution. The only measured result on cadence comes from two in-house tests in 2015. None of the sources reviewed measures how long a post takes from idea to distribution, and none separates what people report about results from what was measured.

Key blogging statistics

Writing time: Bloggers surveyed by Orbit Media in 2023 reported an average of 3 hours 51 minutes to write a blog post. Population: 1,051 respondents to the 10th Annual Blogger Survey. Evidence type: self-reported average. Source: Andy Crestodina’s republication of the survey findings.
Writing time, older benchmark: In HubSpot’s 2015 question, most of 4,000 marketers and salespeople worldwide said they spend 1 to 2 hours writing a 500-word post. Evidence type: self-reported modal band. Source: HubSpot’s blogging time benchmark.
Publishing cadence: Among marketers at businesses that maintain blogs, 22% said their business publishes daily, 37% two to three times per week and 30% weekly. Population: 500+ marketers overall; the size of the blog-maintaining subgroup is not disclosed. Period: 2025 report. Evidence type: self-reported. Source: HubSpot’s 2025 State of Blogging.
Effort and results: In the 2023 Orbit survey, a third of bloggers who spend 6 or more hours on a typical article reported “strong results”, against one in four bloggers overall. Evidence type: correlation between two self-reports. Source: Crestodina’s republication.
Archive effect: In HubSpot’s 2015 cadence test, 92% of its leads and 75% of its traffic in a given month came from previously published posts. Population: HubSpot’s own blog. Evidence type: measured, single publisher, reported second-hand. Source: Contently’s report on the HubSpot and Moz tests.
Reported ROI direction: 50% of marketers at businesses that maintain blogs said blogging ROI was higher in 2024 than in 2023. Evidence type: self-reported judgment with no disclosed calculation. Source: HubSpot’s 2025 State of Blogging.
AI-classified new pages: Ahrefs classified 2.5% of 900,000 newly detected English-language pages as “pure AI” and 71.7% as a mix of AI and human content, April 2025. Evidence type: vendor model classification. Source: Ahrefs’ study of 900k pages.
AI attribution on Medium: A detector assigned 1.77% of sampled Medium posts to AI in January 2022 and 37.03% in October 2024. Evidence type: model classification, peer-reviewed. Source: ACL 2025 paper.
Median writer pay: The US Bureau of Labor Statistics reports a median of $36.98 per hour, or $76,910 per year, for writers and authors in May 2025. Evidence type: government wage estimate. Source: BLS Occupational Outlook Handbook.

How long a post takes, and why the published answers differ

The two effort figures above are not in conflict, because they do not measure the same thing. If you reuse either, state its word count and its method.

The Orbit figure comes from the survey’s author, in a September 2023 republication. Respondents averaged 3 hours 51 minutes to write a post, and the same survey reports an average post length of 1,427 words. The republication does not publish the question wording, so you cannot check exactly how respondents were asked.

HubSpot’s 2015 question asked about a 500-word post, and the most common answer was 1 to 2 hours. That is a modal band, not an average, for a shorter post, collected a decade ago. It is not an earlier version of the Orbit number.

This article uses no figure attributed to Orbit’s 2026 edition, because the survey’s own pages were not available to read. If you see a current effort number cited to that survey, check that the cited page can be opened and says what is claimed.

Neither source publishes the spread around its number. A mean of about 3.85 hours is compatible with a team where most posts take 90 minutes and a few take ten hours, and with a team where most posts take close to four. Nothing obtained distinguishes them.

The same Orbit survey is the closest thing here to an effort-to-outcome link: a third of bloggers who spend 6+ hours per article reported “strong results”, compared with one in four overall. Both answers come from the same respondents, so this shows that two self-reports move together. It does not show that extra hours cause results.

Pricing an hour of writing. BLS reports a median of $36.98 per hour ($76,910 per year) for writers and authors in May 2025, with 140,300 jobs in 2025, 65% of them self-employed. Read it as a labour-cost reference, with three limits. The hourly figure is an annualized estimate, with annual wages calculated over 2,080 hours, and it is not a freelance rate or a fully loaded employer cost. BLS’s wage estimates exclude self-employed workers, while its job count includes them, so the two numbers do not describe the same population. And it tells you what an hour costs, not how many hours a post takes. BLS also projects 0% change, or −500 jobs, from 2025 to 2035 and says AI use for writing is projected to dampen demand. That is a forecast by BLS, not a measurement.

How often teams publish

HubSpot’s 2025 survey asked businesses that maintain blogs how often they publish: 22% daily, 37% two to three times per week, 30% weekly, 7% bi-weekly and 5% monthly. The buckets are independently rounded and add to 101%, so quote them as separate shares rather than summing them into an “at least weekly” figure. The most common answer is two to three times per week, not daily. HubSpot’s page does not define “bi-weekly”, so it is unclear whether that means twice a week or every two weeks. It also publishes no average cadence, so any “average publishing frequency” attributed to HubSpot is not its figure. The fieldwork dates are not disclosed either.

Orbit’s survey series, which ran annually from 2014 to 2023, describes “a steady decline in publishing frequency”, with fewer bloggers publishing more than weekly and more publishing less than weekly. That is a self-reported trend and it ends in 2023.

The only measured cadence result is from 2015, as reported by Contently. HubSpot tested three cadences over six weeks in two-week phases: a benchmark of 23 posts per workweek, a low of 11.5 and a high of 34.5. The low-volume phase received over 30% less traffic than the benchmark and high-volume phases, which received almost the same. Moz ran a separate test around a once-per-weekday schedule: halving frequency produced a 2.9% drop in sessions, and doubling it raised traffic.

HubSpot counted only traffic from posts published within the test window, because 92% of its leads and 75% of its traffic in a given month came from earlier posts. That is the most decision-relevant number in this report. If most of a month’s results come from the archive, this week’s publishing rate is a weak lever on this week’s results, and what you have built up matters more. The sources do not test whether that holds for your site.

Treat both tests with care. Each is a single-publisher internal test from 2015, with no control group or significance testing, reported second-hand, and both predate generative AI. They are the best measured evidence available on cadence, and they are not strong evidence. If your own archive is the lever, the useful next step is deciding which existing pages to improve; see how to choose between a refresh and a rewrite.

Which formats teams say perform

HubSpot asked businesses that maintain blogs which formats perform well. The shares were 51% how-to guides, 45% industry news, 42% case studies, 34% encyclopedic explainers and 23% product listicles. Population: businesses that maintain blogs, 2025 report. Source: HubSpot’s 2025 State of Blogging.

This is a preference ranking, not a performance benchmark. The survey does not define “perform well” and attaches no traffic, conversion or revenue data to any format. No source reviewed here measures format performance. “Case studies get 42% more results” is not something the data says.

Two adjacent findings describe the work rather than its outcome. In the 2023 Orbit survey, almost all bloggers add images, about half add supporting data and statistics, and one in four use video. In HubSpot’s 2025 data, 39% of blog teams increased citations of primary sources and 36% added original research in response to E-E-A-T guidance (Google’s experience, expertise, authoritativeness and trust criteria). Those are reported adaptations by one respondent group, not a measured industry trend.

What “ROI from blogging” can and cannot mean

Before pricing an hour of writing, check how much publishing effort typically goes into a post, using the labour-cost reference to ground assumptions in reported figures rather than guesses.

HubSpot reports that 50% of marketers at businesses that maintain blogs said they saw a higher ROI from blogging in 2024 than in 2023, 31% said it was roughly the same and 15% said it was lower. The source discloses no calculation, attribution method or denominator for “ROI”. It supports one sentence: half of these respondents believe blogging paid back better in 2024 than in 2023. It does not show that ROI rose 50%, or that blogging does or does not pay back.

The Content Marketing Institute’s 2026 research, fielded 24 June to 14 August 2025 among 1,015 B2B marketers, asked for top-three content marketing challenges. The leaders were creating content that prompts a desired action (40%), resource constraints of time, people and budget (39%) and measuring content effectiveness (33%). These are self-reported and cover content marketing broadly, but the second and third are the effort and measurement problems this report keeps meeting. Source: CMI 2026 research.

CMI’s 2027 research announcement, published 4 November 2026, reports that 78% of B2B marketers say speed is AI’s most significant contribution, while 17% say it contributed to quality. It also reports that 65% faced increased pressure to produce content faster in the past 12 months, and 48% of individual contributors report burnout as a result. The announcement publishes neither fielding dates nor full methodology. These are perceptions. “Speed” being the most-cited benefit says what respondents value, not that posts got faster. Source: CMI’s 2027 announcement.

On spending, the 2026 edition reports that 45% of B2B marketers planned to increase 2026 spend on AI-powered marketing tools, 32% on owned media and 9% on human resources such as salaries and training. Those are shares choosing a planned increase, not dollars or hires, so read the contrast as stated intent. HubSpot’s blog survey adds that 56% of blog teams expect investment in blogging to expand and 66% name brand awareness as their top goal.

What is being published now

Three datasets describe AI in published content. They measure different things on different populations, and none pairs them, so do not stack them into one trend.

A vendor crawl (one month, one language). Ahrefs ran its detector over 900,000 English-language pages newly detected in April 2025, one per domain: 2.5% “pure AI”, 25.8% “pure human”, 71.7% a mix. Ahrefs says no AI content detector is perfect, so these are model classifications, not verified authorship. Its finer breakdown of the mixed category does not reconcile with the 71.7% base, so it is not used here.

A peer-reviewed series (three platforms, 2022 to 2024). A 2025 ACL paper analysed about 2.4 million posts from Medium, Quora and Reddit with a detector benchmarked against text from 12 language models. Its “AI Attribution Rate” for Medium rose from 1.77% in January 2022 to 37.03% in October 2024, and for Quora from 2.06% to 38.95%. It is the only trend series in the sources reviewed. It measures detector-assigned share, not quality, intent or outcomes.

A process survey (self-reported). In HubSpot’s 2025 survey, 4% of bloggers said they never use AI to support content creation. The rest: 21% under a quarter of content, 37% for 25% to 50%, 19% for 51% to 75% and 16% for more than 75%. After adopting AI, 19% said production volume significantly increased and 48% moderately increased; 17% reported no change, 9% a moderate decrease and 3% a significant decrease.

Platform volume. WordPress.com states that users produce about 70 million new posts a month. This is the only count rather than a survey in this report, but the page is undated and publishes no method. Its denominator is WordPress.com-hosted blogs plus externally hosted blogs that use Jetpack and are in its network, so it is not a count of all blogs.

What the reviewed sources leave unanswered

The sources in this report leave four gaps. Each has a clear way to be settled, and none of those studies has been run here.

1. The spread of time per post. Every effort figure obtained is a central value of a self-reported estimate. None publishes quartiles or a distribution. A survey publishing the full distribution, or time tracking on a defined content operation, would settle it.

2. Cycle time by stage. The Orbit and HubSpot figures measure writing. None of the sources reviewed breaks effort into research, drafting, editing, design, publishing and distribution. A consented sample tracked stage by stage, reported as medians with ranges, would show where the hours go.

3. Inputs paired with outcomes. The Ahrefs crawl has no performance data, the cadence tests predate generative AI, and the ACL study has no traffic or revenue measures. One dataset recording effort, process and results for the same posts would close this.

4. Reported versus measured results. No source obtained separates what respondents say about results from what anyone measured. HubSpot’s ROI figure is a belief, the cadence tests are measured but old and single-publisher, and the ACL study measures output, not outcomes.

These gaps also work as a checklist if you are weighing in-house against outsourced SEO. Ask any provider, The Blog Writer included, which stage-level times it tracks, what distribution sits behind its averages and how it connects publishing decisions to measured results. If you are mapping those stages for your own team, a research-to-publication workflow gives you the records to track.

Methodology and source notes

This is a source-led report. It is not based on a new survey or experiment by The Blog Writer and contains no first-party performance data from The Blog Writer or its clients.

Sources were fetched on 6 November 2026 (UTC). Pages used: HubSpot’s 2025 State of Blogging (updated 14 July 2026) and its blogging time benchmark (2015 data, updated 12 June 2025); CMI’s 2026 B2B research page and 4 November 2026 press announcement; Ahrefs’ study, published 19 May 2025; BLS Occupational Outlook Handbook for writers and authors; WordPress.com’s live activity page; Crestodina’s republication of the 2023 Orbit survey, published 28 September 2023; Contently’s 4 August 2015 report on the HubSpot and Moz tests; and the ACL 2025 paper (DOI 10.18653/v1/2025.acl-long.1120).

How to label each figure when you reuse it:

Measured, historical: the 2015 HubSpot and Moz cadence tests, reported second-hand by Contently. The underlying data belong to HubSpot and Moz.
Self-reported: Orbit’s 2023 effort, cadence and format figures; HubSpot’s 2025 cadence, format, ROI and process figures; CMI’s 2026 and 2027 figures.
Model classification: the Ahrefs crawl and the ACL attribution rates.
Forecast: BLS’s 2025–35 projection of 0% change (−500 jobs). Use BLS’s published figure, not one you calculate from its tables.
Wage estimate: BLS’s May 2025 median, annualized over 2,080 hours.
Platform count: WordPress.com’s undated statement, with the Jetpack-inclusive denominator above.

Figures attributed to Orbit’s 2026 edition and WordPress market-share figures from W3Techs are excluded because they could not be verified for this report. W3Techs measures websites rather than bloggers or posts, so it would not answer a blogging question.