Is blogging still profitable?
Yes for some models, harder for search-plus-ads blogs. AI-summary studies, a cost model, break-even sums and a neutral platform comparison.
Yes, blogging can still be profitable, but the old recipe (rank in Google, show display ads, collect the cheque) is harder than it used to be, and the reason is mostly arithmetic. Several studies from 2025 and 2026 find fewer clicks reaching websites when Google shows an AI summary, though they disagree about how many fewer, and Google says its total click volume is roughly flat. Meanwhile the cost side has collapsed: a blog can run on a few dollars a year in cash. So whether yours is profitable depends on whether you count your own time, which revenue model you pick, and whether readers find you through a search box or because they chose to follow you.
A disclosure before the comparisons: I work on VeloCMS, and it's one of the platforms priced below. It's a young hosted product with a smaller ecosystem than most of the others here, and I say where that matters.
Last verified: 2026-11-21.
How I checked: every price, limit and study figure below comes from the vendor's or researcher's own page, read on 2026-11-21. Where two pages from one vendor disagreed, I say so. I haven't quoted an "average blogger income", because I couldn't find one with a published method. The sums later on are arithmetic on published prices, using inputs I chose and labelled. They aren't forecasts. Prices are in USD unless stated, and they change, so check the pricing page before you pay.
What does "profitable" actually mean for a blog?
Profit is revenue minus cost, and the argument usually goes wrong because people count only the cash half of the cost. Write it as three lines.
- Revenue: money that arrives because of the blog, from ads, subscriptions, products, sponsors, or client work the blog brought in.
- Cash cost: domain, hosting or platform plan, email tool, analytics, payment fees.
- Time cost: the hours you spend writing, promoting and maintaining, multiplied by whatever they're worth to you.
Cash cost has fallen close to zero for a simple blog, so almost any blog is "profitable" on the first two lines. The third is where most lose. A blog earning $40 a month from ads at $1 of cash cost looks fine until you notice it took twelve hours to write the posts behind that traffic.
None of that is a reason to quit. Plenty of people blog for the work it attracts or the thinking it forces. It does mean you should decide which line you're optimising before you pick a platform.
What changed in search, and what do the studies actually show?
Several datasets show fewer clicks when an AI summary appears. They measure different things, and none is about developer blogs specifically.
Pew Research Center's July 2025 study is the cleanest browsing-based look. It took March 2025 browsing data from 900 U.S. adults on an online panel with a tracking app installed, then re-ran their searches in April to see what Google had shown. Of 68,879 unique searches, 12,593 (about 18%) produced an AI summary. Users clicked a traditional result in 8% of visits to pages with a summary and 15% of visits without one, and clicked a link inside the summary in 1%. Of visits with a summary, 26% ended the browsing session, against 16% without. Pew's caveats: summaries change over time, only Google was analysed, and user actions are inferred from the next URL visited. (Pew)
Ahrefs reports a bigger number by a different method. Its February 2026 update compares 300,000 keywords, half with an AI Overview and half informational keywords without one, using aggregated Search Console data. Position-one click-through on the AI Overview keywords went from 0.073 in December 2023 to 0.016 in December 2025, which Ahrefs puts at 58% lower than the roughly 0.037 it estimates they would have had otherwise. Ahrefs calls this a correlation, the counterfactual is a forecast, and the CTR figures are desktop only. (Ahrefs)
Seer Interactive's April 2026 update is the newest and moves in both directions. Across 53 brand accounts, 5,471,127 queries and 2.43 billion organic impressions, organic CTR on queries showing an AI Overview fell from 3.19% in January 2025 to 1.31% in December 2025, then rose to 2.36% in February 2026. Queries with no overview went from 2.75% to 3.16% to 3.82%. In a separate brand-level table for December 2025, CTR was 0.95% when the brand was cited, 0.55% when it wasn't, and 3.97% with no overview (Seer's all-query no-overview figure for that month is 3.16%, and it doesn't reconcile the two). Seer says it isn't back to pre-overview levels, and lists its own limits: no causation (stronger brands may simply get cited more), an AI Overview label taken from the latest SERP snapshot and applied to every earlier month, and one account making up 47% of one segment's impressions. (Seer)
One Seer detail matters for technical writers. AI Overviews appeared on 85.9% of question-shaped queries and 95.4% of comparison queries, against 27.3% of single-word ones. "How do I fix this error" is a question-shaped query. The sample is brand accounts, not developer blogs, so read it as a pointer to where the pressure lands.
Two news-specific datasets add context. Similarweb, as reported in July 2025, said the share of news searches ending without a click to a news site rose from 56% in May 2024 to nearly 69% in May 2025, with no method visible in the coverage and no Similarweb original I could find (TechCrunch). The Reuters Institute, using Chartbeat data from 2,576 sites, reported Google organic search traffic down 33% globally and 38% in the U.S. between November 2024 and November 2025, adding that "it is not clear how much of this is down to AI overviews" (Reuters Institute).
Google's side: in August 2025 Liz Reid wrote that "total organic click volume from Google Search to websites has been relatively stable year-over-year," called third-party reports of large drops "inaccurate," and published no underlying numbers (Google). Its documentation says AI feature traffic is counted under the "Web" search type in Search Console, with no separate report (Search Central).
What do I take from five sources that disagree? Not a number. The effect shows up in browsing and Search Console data for informational queries, it varies by query type and by whether you're the cited source, it may be partly reversing, and Google's aggregate claim and the independent studies measure different populations. Your own Search Console data is the only dataset about your blog, and it's free.
Did Google's quality updates hurt small independent blogs?
Google's stated aim is the opposite, and some publishers say the result hasn't matched. I have no verified number for "small blogs" as a class.
On 5 March 2024 Google announced a core update and three spam policies: expired domain abuse, scaled content abuse and site reputation abuse (enforced from 5 May 2024) (Search Central). A companion post projected a 40% cut in low-quality, unoriginal content, and an April update to it reported 45% once the rollout finished on 19 April (Google). On 15 August 2024 Google said its next core update aimed to surface "small or independent sites that are creating useful, original content" (Search Central). Those are intentions and Google's own measurements.
On the other side, HouseFresh, a product-review site, argues that Google's ranking favours large publishers' "best of" lists over sites that test products, and says its own traffic from those pages is falling. The author discloses affiliate income on the site, and the piece gives no traffic numbers in text, only charts and examples (HouseFresh). It's one site's account, not a statistic.
The practical lesson is smaller than the argument. If your income depends on one search engine's ranking decisions, you're exposed to changes you can't see coming or appeal, whichever side is right.
What does it actually cost to run a blog in 2026?
Less cash than ever, over a wide range. Each line is from the vendor's page, read today.
- Domain: a .com is $11.81 a year to register and renew on Porkbun's listing, about $0.98 a month. Cloudflare Registrar says it never marks domain prices up, but shows no dollar figure.
- Free static hosting: Cloudflare Pages' Free plan allows 500 builds a month, 20,000 files per site and 25 MiB per asset. GitHub Pages caps a site at 1 GB with a soft 100 GB monthly bandwidth limit, and says it isn't allowed "as a free web-hosting service to run your online business," which matters if you plan to sell from the site.
- A VPS: Hetzner's cloud page lists its lowest shared-resource plan (Regular Performance) at €5.49 a month excluding VAT, with no specs on the page; the cheaper-sounding cost-optimized line is marked unavailable. Ghost's install guide asks for at least 1 GB of memory, MySQL 8 and Nginx. Ghost's own comparison puts self-hosting from $10 a month and says self-hosters "pay for your own email delivery, analytics and CDN."
- Analytics: Cloudflare Web Analytics is free and says it uses no cookies or localStorage. GoatCounter's hosted service is "currently offered for free for reasonable public usage" and is open source. Plausible's cloud starts at $9 a month for 10,000 monthly pageviews, with a 30-day trial and no card.
- Email: Kit's Free plan covers up to 10,000 subscribers per its table and FAQ (one card on the page says 1,000); its Creator plan lists $39 a month for 1,000. Kit's sales fee is stated two ways, "only 0.6% of the total fees" in the text and 3.5% + 30c in the comparison table, so check it before you rely on either. Buttondown is free for the first 100 subscribers, with paid subscriptions a +$9 a month add-on. beehiiv's Free plan allows 2,500 subscribers without paid subscriptions; Lite starts at $49 a month on annual billing or $59 monthly.
- Payments: beehiiv's FAQ quotes Stripe's standard fee as 2.9% + $0.30 per transaction. Stripe's UK page, which is what loaded for me, lists 1.5% + 20p for standard UK cards, 3.15% + 20p for international cards and £20 per dispute received.
- Time: whatever you decide it's worth. This is the one line that can outweigh the rest combined.
A bare-bones static blog costs roughly $12 a year. A managed platform plus an email tool runs to a few hundred. Neither is much next to an evening's wages.
What does the break-even arithmetic look like?
Three worked examples, all arithmetic on published prices with inputs I picked. They show how the models behave, not what you'll earn. One assumption throughout: twelve hours of work a month, valued at $40 an hour, so $480 of time. Change both to fit your life.
Example 1, display ads. Carbon starts at 10,000 pageviews a month with a typical CPM of $1.50 to $2.30; EthicalAds asks for 50,000 or more and estimates about $2.50 per thousand (Carbon, EthicalAds). At those floors that's $15 to $23 a month on Carbon and about $125 on EthicalAds. Run it backwards: covering the roughly $0.98 a month of domain takes about 400 to 650 pageviews. Covering the $480 of time takes 192,000 pageviews a month at $2.50, or roughly 209,000 to 320,000 at Carbon's range. Ads pay your hosting almost immediately and your time only far above either network's entry threshold.
Example 2, paid subscriptions. Assume readers pay $7 a month by card and Stripe takes 2.9% + $0.30, which is $0.50. I applied that rate to every option, and it ignores any Stripe Billing fee. Substack's page says writers keep 90% of revenue "minus credit card fees" without giving a card rate, so I assumed the same one: about $5.80 a subscriber. Platforms listing a 0% fee leave about $6.50. WordPress.com's Business plan takes 2% and Personal 8%, leaving about $6.36 and $5.94.
Subscribers needed to cover the plan price plus $480 of time: Substack, which states no monthly fee, about 83. Ghost Publisher at $29 (billed yearly), the cheapest tier listing paid subscriptions, about 79. VeloCMS Pro at $9, about 76. WordPress.com Business at $25 (yearly billing), about 80. beehiiv Lite at $49, about 82. Once your time is in the sum, they all land between roughly 75 and 83. Fee structure matters at the margin, not at the order of magnitude.
On cash cost alone, a 10% share equals a flat fee F when monthly revenue hits 10 × F: $290 (about 41 subscribers at $7) for Ghost Publisher, $90 (about 13) for VeloCMS Pro, $490 (70) for beehiiv Lite. Below that the percentage is cheaper, above it the flat plan is.
Example 3, indirect value. Say the blog brings in clients or a job offer and the platform is a portfolio. Yearly cash cost with a domain: a static site on Cloudflare Pages is about $11.81. Hashnode Pro is $50 a seat a year and lists the custom domain under Pro, so about $61.81. VeloCMS Pro is $86.40 billed yearly, about $98.21. Ghost Starter at $18 a month billed yearly is $216, about $227.81. WordPress.com Personal at $4 a month on yearly billing is $48, and the page says it includes a free domain for the first year. Any one paid engagement larger than that covers the year. I have no sourced figure for how often a post leads to paid work, so I won't offer one. The break-even is low and the variance is enormous.
Together: ads need traffic most small blogs won't reach to pay for your time, subscriptions need a few dozen to a hundred loyal readers, and indirect value needs one right reader. Those thresholds, not platform fees, decide which model suits you.
How do you measure whether your blog is profitable?
Keep a ledger, because a feeling of profitability is the thing most likely to be wrong. A monthly sheet with revenue by stream, cash costs by item and hours by activity takes ten minutes. After a quarter you'll know which posts earn and which just consume evenings.
For attribution, tag the links you control. Put UTM parameters on newsletter and social links, and give each product its own landing path or discount code. Plausible can record revenue against a goal: its docs say revenue tracking is a Business plan feature, takes an ISO currency code and an amount in the main currency unit (not cents), and silently records the event without revenue if the code is invalid. A spreadsheet does the same job for free.
Pair that with traffic data you can trust. Search Console shows your real impressions, clicks and CTR, AI feature traffic included. A cookieless analytics tool from the cost list covers on-site behaviour. Then watch three ratios: revenue per post, revenue per thousand pageviews, and subscribers per post. If one post brings in nearly all your subscribers, you've learned something no pageview total would show. And log where each reader came from: if most traffic is search and your CTR on informational queries is sliding, you're in the exposed group described earlier.
Which models hold up when search traffic falls?
The ones where readers arrive on purpose. An email list, a membership, a community, a product with its own audience: all depend less on a ranking you don't control. A list is the clearest case, since you hold the contact details with no search engine in between, and the sums above show paid newsletters need dozens of subscribers, not tens of thousands of visits.
What about AI assistants and crawlers sending visitors? The evidence is thin. As reported, Similarweb's figures show ChatGPT referrals to news sites growing from just under 1 million in January to May 2024 to more than 25 million in 2025, still small next to the search losses it describes. That's news sites only, with no method I could read and only second-hand through TechCrunch. On llms.txt, Google's documentation says you don't need "new machine readable files, AI text files, or markup" to appear in AI Overviews or AI Mode. Search Engine Roundtable reports that on 17 June 2025 Google's John Mueller wrote on Bluesky "FWIW no AI system currently uses llms.txt," which is one person's statement from last year, not a measurement. Clear, crawlable, indexed content is sensible. Betting income on a file format isn't.
My guess, which I can't prove: the blogs that keep earning treat search as a way to meet readers, not as the place they live.
Does your platform choice change the economics?
It changes the fixed cost, the percentage you give up, and how much work you do yourself. Nine ways to run a blog, compared on running cost and revenue share only, with prices from each vendor's page read today.
- Static site on Cloudflare Pages: $0 hosting within the Free limits, plus a domain. No revenue share, since nothing takes a cut. You add your own email and payment tools.
- Self-hosted WordPress: the WordPress.org page lists "Free" as a feature, and you pay for the host (Hetzner from €5.49 a month excluding VAT is one example). No share from WordPress itself; plugins and gateways have their own terms. You do updates, backups and security.
- WordPress.com: Free at $0 (10% fee on paid content, ads shown to visitors), Personal $4 yearly-billed ($9 monthly), Premium $8 ($18), Business $25 ($40), Commerce $45 ($70). The fee on paid content falls from 10% to 8%, 4%, 2% and 0% across those plans, on top of Stripe's fees. Plugins need a paid plan.
- Self-hosted Ghost: base self-hosting from $10 a month per Ghost, with email delivery, analytics and CDN paid separately. The page states no Ghost-side revenue share for self-hosting, so check it.
- Ghost(Pro): Starter $18 a month billed yearly, with 1,000 members and no paid subscriptions; Publisher $29 with paid subscriptions; Business $199 (the pricing page shows only yearly-billed prices). It lists 0% Ghost transaction fees on Publisher and above, with payment processor fees still applying. Ghost's hosting docs say Ghost(Pro) starts "From $15/mo", so its own pages disagree. Confirm at checkout.
- Substack: the page states no monthly fee, says writers keep 90% of revenue minus credit card fees, and says you own your subscriber list.
- beehiiv: Free for up to 2,500 subscribers without paid subscriptions, Lite from $49 a month yearly-billed, with a stated 0% take rate on paid subscriptions and Stripe's fee on top.
- Hashnode: free publications, plus Pro at $5 a seat a month or $50 a seat a year, listing a custom domain, GitHub backup and scheduled publishing. The page lists no paid subscriptions, ads or sponsorship tooling, so income happens elsewhere.
- VeloCMS: Pro $9 a month ($86.40 billed yearly), Business $29, Agency $69. Pro lists 5,000 subscribers, 5 GB of storage and one site. Every plan carries a 0% VeloCMS platform fee on your Stripe sales, and you pay Stripe's standard processing fee. There's no free plan: a 14-day trial needs a card unless you hold a cardless trial code, and the first charge lands on day 15.
The highest cash cost isn't automatically the worst. If a managed platform saves four hours a month of server upkeep, that's $160 at the $40 an hour I assumed, more than most of these plans cost. Self-hosting is cheaper only if your hours are free.
Who is each option best for?
- Static site on Cloudflare Pages: developers happy with Git workflows who earn indirectly and want the lowest cash cost.
- Self-hosted WordPress: people who need plugin breadth or already know it, and accept the maintenance.
- WordPress.com: people who want WordPress without a server, from a free hobby site to a managed Business plan.
- Self-hosted Ghost: writers comfortable managing a server who want Ghost without the monthly plan.
- Ghost(Pro): writers whose blog and paid newsletter are the business and who prefer a flat plan to a percentage.
- Substack: writers who want the quickest start on paid email and accept a 10% share.
- beehiiv: newsletter-first writers who want a free tier for a free newsletter and paid growth tooling later.
- Hashnode: developers who want an audience and portfolio with almost no operations work, earning through jobs, clients or products elsewhere.
- VeloCMS: people who want a blog, newsletter and memberships under one hosted subscription with no platform fee on their own Stripe sales, and can accept a young product.
When is VeloCMS the wrong choice?
- You want to try the idea for nothing. There's no free plan.
- You want to self-host or own the code. It's hosted only.
- You need a deep plugin marketplace or a large community. WordPress has far more of both.
- You earn only from ads on a static site. A subscription stack is cost you don't need.
- You want the lowest possible cash bill and have Git skills. A static site is cheaper.
FAQ
Is blogging still worth it in 2026?
For some models, yes. Cash cost is low enough that the real questions are your time and whether readers come to you directly. Display ads on search traffic are the weakest case, given the click studies above.
How much do AI summaries reduce clicks?
It depends on the study. Pew found 8% of visits with a summary clicked a traditional result against 15% without. Ahrefs estimated a 58% lower position-one CTR on AI Overview keywords. Seer saw a drop through 2025 and a partial recovery in early 2026. Google says total click volume is relatively stable. None measures developer blogs specifically.
How many readers does a blog need to pay for itself?
For cash costs alone, very few: a few hundred pageviews at Carbon's CPM range covers a domain, and a handful of subscribers covers a $29 plan. Covering your time too, the sums above need roughly 75 to 83 paying subscribers at $7 a month, or about 200,000 monthly ad pageviews at $2.50 per thousand. Those rest on my stated assumptions.
Do I need llms.txt to get cited by AI assistants?
Google's documentation says you don't need new machine-readable files, AI text files or markup to appear in AI Overviews or AI Mode. The evidence that other AI systems use the file is thin.
Does VeloCMS have a free plan?
No. The pricing page says so in as many words. What you get is a 14-day trial on each plan, started with a card unless you hold a cardless trial code.
Which fee matters more, the platform's or the processor's?
Both, differently. The processor's fee is similar everywhere. The platform's is the one you can choose, and flat versus percentage crosses over at 10 times the flat fee.
Sources
All accessed 2026-11-21.
Pew Research Center · Ahrefs · Seer Interactive · TechCrunch on Similarweb · Reuters Institute · Google on clicks · Search Central AI features · Google March 2024 · Search Central March 2024 · Search Central August 2024 · HouseFresh · Search Engine Roundtable on llms.txt · Porkbun · Cloudflare Registrar · Cloudflare Pages limits · Cloudflare Web Analytics · GitHub Pages limits · Hetzner Cloud · Ghost install guide · Ghost hosting · Ghost pricing · WordPress.org download · WordPress.com pricing · Substack · beehiiv · Kit · Buttondown · Plausible pricing · Plausible revenue tracking · GoatCounter · Carbon · EthicalAds · Stripe · Hashnode Pro · VeloCMS pricing
About the author
- Building VeloCMS.org , A fast multilingual CMS for blogs and online stores. I write about headless CMS, Next.js, SEO and moving off WordPress and Shopify.
Comments
Join the discussionNo comments yet. Be the first to comment.