Last updated: 20 August 2026. Written by Adrian Tan, Singapore Digital Marketing (SDM).
Two things are true at the same time, which is why this question produces such bad answers. Blogging as most Singapore businesses practise it — a few hundred words on a generic topic, published because the calendar said so — has stopped working, and the data on that is now unambiguous. Blogging as a way to be the page that answers a specific commercial question is working roughly as well as it ever did, and in some categories better, because so many competitors have given up.
The useful question is therefore not “does blogging still work” but “which posts still earn their cost, and what do I do with the ones that don’t.” This piece answers both, with the 2026 evidence, the Singapore specifics, and a triage process for an existing blog that has flatlined.
What the data actually shows
Start with the number everyone is arguing about. Ahrefs compared 150,000 keywords with an AI Overview present against 150,000 informational keywords without one, using aggregated Google Search Console click-through data, and set December 2023 against December 2025. Two findings matter.
First, click-through rates fell across the board, AI Overview or not. Position-one CTR on keywords with no AI Overview went from 0.076 to 0.039 — roughly halved — over the same period. Something broader than AI Overviews is compressing clicks: more ads, more SERP features, more answers rendered on the results page itself.
Second, on top of that general decline, keywords with an AI Overview lost substantially more. Position-one CTR there fell to 0.016 against a forecast of 0.037 had the general trend applied — a 58% shortfall, up from 34.5% in Ahrefs’ April 2025 read. The effect weakens down the page: position two showed a 50.8% reduction, falling to 19.4% by position ten.
Pew Research Center measured the same phenomenon from the user side rather than the site side: across about 70,000 Google searches by around 900 US adults in March 2025, the organic click rate was 15% without an AI Overview and 8% with one, links inside the AI Overview were clicked about 1% of the time, and the share of sessions that simply ended at the search page rose from 16% to 26%.
So: fewer clicks per ranking, and the loss concentrated exactly where a generic explainer post was designed to win. That is the case against blogging, and it is a real one. Now the other half.
What did not change
Three things survived intact, and they are the basis of everything that still works.
Google is still the market. StatCounter put Google at 92.46% of Singapore search in July 2026, Bing at 3.37%, everything else in fractions. Whatever happens to search, it happens on Google, and the mechanics of being found there are unchanged.
The rules for appearing in AI answers are the ordinary SEO rules. Google’s own documentation on AI features is blunt about this: “There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary,” and “You don’t need to create new machine readable files, AI text files, or markup to appear in these features. There’s also no special schema.org structured data that you need to add.” Eligibility is being indexed, being eligible for a snippet, meeting the technical requirements and following Search policies. Anyone selling a separate “AI optimisation” discipline at a separate fee is selling you the thing you already bought.
Intent still determines the value of a click. A person searching “what is content marketing” was never a lead. A person searching “content marketing agency singapore pricing” is. The AI Overview took the first one; it did not take the second.
Which posts still earn their cost
This is the working answer. Sort every post you are considering along two axes: how much of the query can be satisfied by a paragraph of generated text, and how much of the answer is proprietary to you.
In practice that produces a short list of post types that still reliably pay in Singapore:
| Post type | Why it survives | Typical SG search volume |
|---|---|---|
| “How much does X cost in Singapore” | Commercial intent; needs local prices a generated summary will hedge on | Tens to a few hundred a month, very high value per visit |
| “X vs Y” comparisons | Decision intent; the reader wants a recommendation with reasons | Low but qualified; also heavily used as source material by AI answers |
| Local and regulatory specifics (PDPA, IMDA, grants, licensing) | Wrong answers are costly, so readers verify at the source | Small, but almost no competition and high trust value |
| Case studies with real numbers | Cannot be synthesised; nobody else has your client’s result | Negligible search; judge on close rate, not traffic |
| Genuinely operational how-tos with screenshots and edge cases | Depth beyond what a summary can carry; earns bookmarks and links | Moderate; the long tail does the work |
| Original data you collected | The only reliable link magnet left | Judge on referring domains, not sessions |
And the list of what to stop writing is equally short: standalone definitions, “benefits of” posts, trend pieces with no proprietary data, anything under about 800 words that exists to hit a publishing quota, and any post whose entire content could be reconstructed from the first paragraph of three competitors.
Deciding which side of that line a topic falls on is a repeatable exercise rather than a judgement call, and it belongs before the calendar rather than after it. Our guide to content keyword research for Singapore businesses sets out the five criteria we score topics against — proof, commercial proximity, winnability, cluster fit and durability — and, deliberately, search volume is not one of them.
The Singapore adjustment
Three local factors change the calculation relative to the US-sourced advice most of this debate is conducted in.
Volumes are small, so quality of intent beats quantity of traffic. A Singapore commercial keyword doing 200 searches a month with genuine buying intent is worth more than a 5,000-search informational term, and it always was — the AI Overview simply removed the illusion that the 5,000-search term was ever worth chasing. Our keyword research guide covers how to read SG volumes without over-reacting to them.
Local intent is a defence. Queries with a Singapore modifier, an area name or a regulatory reference pull local results and localised answers, and the generated summary is markedly more hedged where the correct answer depends on jurisdiction. That is a structural advantage for a genuinely Singapore-based business writing about Singapore conditions, and almost nobody exploits it fully.
Competition is shallower than you think. Because volumes are small, a lot of the SERP is occupied by directory listicles and thin agency posts written to a template. A single properly researched page with real figures routinely outranks all of them. The bar is lower here than in a US market of the same difficulty score.
How to triage a blog that has stopped working
If you already have 40 or 200 posts and traffic has been sliding, do not start by writing more. Run a triage pass first. It takes a day and it is consistently the highest-return work available.
Step 1: pull sixteen months of Search Console data by page: clicks, impressions, average position. Sixteen months lets you compare like months and see the trend rather than the seasonality.
Step 2: sort every URL into four buckets.
| Bucket | Signal | Action |
|---|---|---|
| Working | Clicks stable or growing; commercial or local intent | Refresh annually. Add internal links to it from newer posts |
| Ranking but not clicked | Impressions flat or up, clicks falling hard | Check whether an AI Overview now sits on the query. If yes, rewrite for the decision the reader is making, not the definition |
| Overlapping | Several posts competing for near-identical queries | Merge into one strong page, redirect the rest. This is usually the single biggest win |
| Dead | Near-zero impressions for 12 months, no links, no commercial relevance | Delete or noindex. A thin archive dilutes the site and gives readers nothing |
Step 3: fix the overlaps before writing anything new. On a typical 60-post SME blog we expect to find eight to fifteen posts that should be three, and a long tail of ten to twenty that should not exist. Consolidating them concentrates links and relevance on fewer, better URLs — which is the pillar-and-cluster structure described in our content marketing guide for Singapore, applied retrospectively.
Step 4: only then commission new posts, from the top-right quadrant of the matrix above, with the on-page checklist applied before each goes live.
A worked triage
To make that concrete, here is the shape of the outcome on a typical Singapore B2B services blog of about 60 posts built up over four years — the pattern is consistent enough across accounts to be worth setting out as a template.
| Bucket | Share of posts | Share of clicks | What we do with them |
|---|---|---|---|
| Working | ~10% | ~70% | Refresh, extend, and link to from everything new. This is the actual asset |
| Ranking but not clicked | ~20% | ~15% | Rewrite around the decision, add the local specifics a summary cannot carry |
| Overlapping | ~30% | ~10% | Merge into a handful of stronger pages, 301 the rest |
| Dead | ~40% | ~5% | Delete or noindex; keep only those with earned links, redirected |
The uncomfortable implication of that distribution is that most of a four-year blog is overhead. The comfortable one is that the fix costs a fraction of what producing it did, and the merged pages usually recover more clicks within a quarter than the deleted ones were producing in total.
Why the surviving posts matter more than the traffic suggests
There is a second reason not to judge a blog purely on sessions. Gartner’s B2B buying research has for years put roughly 80% of the buying journey outside any contact with a vendor rep, with about three-quarters of buyers preferring a rep-free experience, and 6sense’s buyer research places first vendor contact at around two-thirds of the way through the evaluation. In a market as small as Singapore, where a shortlist may be three firms drawn from a LinkedIn thread and an afternoon of reading, the pages a buyer reads before they contact you are doing sales work that never appears in an organic traffic report. That is precisely why case studies and pricing pages belong in a content programme even when their search volume rounds to zero.
Measuring it now that clicks are not the whole story
The old dashboard — sessions from organic, month on month — will make good work look like failure, because impressions can grow while clicks fall. Four adjustments make it honest again.
Report impressions and clicks together. If impressions are rising and clicks are flat, you are winning visibility in a market where visibility converts less often. That is a different problem from losing rankings, and it has a different fix.
Segment brand from non-brand. Brand queries are demand you already created; leaving them in makes every content report look better than it is.
Watch the commercial pages, not the blog average. The blog exists to feed the pages that convert. If the cost and pricing pages are holding their clicks, the programme is working even if the explainer posts are not.
Accept that some of the credit is now invisible. GA4’s “AI Assistants” default channel captures referrals from chatbots such as ChatGPT and Gemini, but Google’s documentation is explicit that traffic arriving from AI Overviews and AI Mode is classified as Organic Search, not as an AI channel. A further slice arrives with no referrer at all and lands in Direct. There is no configuration that fixes this; the answer is to stop expecting single-channel attribution to be complete. Our attribution models guide and GA4 setup guide cover the practical configuration, and measuring content marketing ROI covers what to report to a board when perfect attribution is not available.
When the honest answer is “don’t blog”
We talk clients out of content programmes reasonably often. The cases where blogging is the wrong instrument:
Nobody is searching for your category. If keyword research turns up almost no demand — genuinely new product categories, highly specialised B2B niches — content can still build trust, but it will not capture demand that does not exist. Direct outreach and paid social usually beat it.
You need revenue this quarter. Content pays back over quarters. If the business needs pipeline in six weeks, run ads and revisit content when the cash position allows. We set this out in is SEO worth it in Singapore.
You cannot sustain it. A blog abandoned after four posts is worse than no blog: it signals a dormant business to every visitor who checks the date. If you cannot commit to twelve months, do something else.
Your site cannot convert the traffic. Sending readers to a site with no clear next step wastes the whole investment. Fix the destination first — see building a website for SEO.
The short version
Blogging did not stop working; a particular kind of blogging stopped working, and it happened to be the most common kind. The click data is real: Ahrefs measured a 58% shortfall at position one where an AI Overview appears, and Pew found organic click rate almost halving. Both losses land on generic informational queries. Commercial, local, comparison and evidence-backed content — the material that produces enquiries rather than sessions — is substantially insulated, and Google’s own guidance confirms there is no separate discipline to buy in order to appear in AI answers.
For most Singapore businesses the correct response is to publish less and mean it more: triage what exists, merge the overlaps, delete the dead weight, and put the freed-up budget into the handful of pages that answer a question a buyer would pay to have answered.
Publishing less is easier to say than to schedule, and the number you land on should come from arithmetic rather than a benchmark — our guide to how often you should publish works out the cadence your team can genuinely hold for twelve months, and why about 30% of that capacity belongs to improving pages you have already published.
Our Singapore case studies show what that looks like once it compounds. If you want the triage run properly — the Search Console pass, the merge plan and the rewrite priorities — our content marketing service starts there rather than with a publishing schedule, and our SEO service covers the technical side underneath it.
If the posts are good but nobody is acting on them, the missing piece is usually credibility rather than quality: E-E-A-T for Singapore businesses covers the authorship, proof and trust signals that decide whether a reader believes the page.
Frequently asked questions
Is blogging dead for SEO in 2026?
No, but generic informational blogging is close to it. Ahrefs found that where an AI Overview appears, position-one click-through rate fell 58% below what the general trend predicted, and Pew Research Center measured organic click rate dropping from 15% to 8% on searches with an AI Overview. Those losses concentrate on definition-style queries. Posts targeting commercial, local, comparison and decision queries — and anything carrying proprietary data or client results — continue to work.
How many blog posts do I need to see results in Singapore?
Think in clusters, not counts. Six to ten genuinely distinct posts around one topic, interlinked and pointing at a pillar page, is the point at which a cluster starts behaving like an asset. Twenty posts spread across ten unrelated topics will not, however long you wait. Because Singapore search volumes are small, depth on a narrow topic beats breadth almost every time.
Do I need to do something special to appear in AI Overviews?
No. Google’s documentation on AI features states there are no additional requirements to appear in AI Overviews or AI Mode and no special optimisations necessary, and that you do not need new machine-readable files, AI text files or special structured data. Eligibility is being indexed, being eligible for a snippet, meeting the technical requirements and following Search policies. Treat vendors selling a separate paid “AI optimisation” discipline with caution.
Should I delete old blog posts that get no traffic?
Often yes, but merge before you delete. Pull sixteen months of Search Console data and sort every URL into four buckets: working, ranking-but-not-clicked, overlapping and dead. Overlapping posts should be consolidated into one strong page with redirects, which is usually the biggest single win. Only genuinely dead pages — near-zero impressions for twelve months, no links, no commercial relevance — should be deleted or noindexed.
Why are my impressions going up while my clicks go down?
That is the signature of AI Overviews and other result-page features on your queries. You are being shown more and clicked less, which is a different problem from losing rankings and needs a different fix: rewrite for the decision the reader is making rather than the definition they can now get on the results page, and shift effort towards commercial and local queries where the answer cannot be summarised away.
Can I see how much traffic AI search is sending me?
Only partially. GA4 has an “AI Assistants” default channel for referrals from chatbots such as ChatGPT and Gemini, but Google’s documentation is explicit that visits originating in AI Overviews and AI Mode are classified as Organic Search rather than as an AI channel, and a further share arrives with no referrer and lands in Direct. No configuration resolves this fully, so report it as a known gap rather than pretending to a precision you do not have.
