In March 2025, Asetena Pa drew 726,303 organic visits from Google. By April 2026 it drew 30,414. We publish that platform, the pages still work, and readers still need what is on them. Everything that changed sits in the gap between the page and the reader.
Publishing has never been easier and earning a visit has never been harder. Search results answer more questions on the page, AI tools summarise and cite, and a great deal of writing now gets consumed without ever being opened. A blog post can still bring search traffic, and it can also become source material, a proof point, a sales reference, a support page, a newsletter idea, a citation, or the clearest place to explain how a team thinks. The useful question is what work a blog is expected to do now.
The old traffic loop is weaker
For a long time blogging had a simple shape: find questions people search, answer them, rank, and let Google send the visits. That still holds where the reader needs depth, trust, or a current answer. For everyday informational articles it has become a shaky base, because search now handles the easy question itself.
Ahrefs found that AI Overviews were linked with a much lower click-through rate for top-ranking informational results in its December 2025 data. BrightEdge also found AI Overviews appearing across a large share of tracked queries, with many results pushed lower on the page. A 2026 study on Google AI Overviews found that question-style searches were especially likely to trigger AI summaries.
The details vary by study, query, country, and industry. The direction holds: much of the informational web is now read, compressed, and presented before the original page gets the visit.
Asetena Pa made it visible
Asetena Pa is a Ghanaian media and information platform we operate, covering mobile money, internet guides, education, and local services. Many pages answer practical questions people search when they need to do something quickly, which makes it a clean place to watch this shift happen.
The fall from 726,303 to 30,414 reads like a verdict on the writing. The pages tell a different story.
We wrote about the first shock in We stopped chasing traffic, which covers what we changed strategically once the visits stopped arriving. This piece sits with the numbers themselves and what they say about writing for a search that increasingly answers on its own.
The traffic graph shows the visible part of the shift. Search features changed, result pages changed, and some answers moved closer to the search box. The pages kept being surfaced while fewer people needed to open them.
Google Search Console shows the same thing from another angle. Impressions persisted while clicks and CTR fell away. By May 2026 monthly clicks had fallen to 857 from a peak of 23,265, and CTR moved from 2.32% to 0.35% across the period we measured.
Anyone still treating blogging as a straight traffic machine should sit with those two lines. The machine runs, with far more friction in it.
Some pages still earned the visit
The decline was uneven. Some Asetena Pa pages held their click-through rate because the reader’s need outlasted a short answer, where the information was current, specific, or tied to an action that mattered to get right.
| Page topic | Total clicks | Total impressions | CTR | Avg position |
|---|---|---|---|---|
| BECE grading system | 16,590 | 691,511 | 2.40% | 10.5 |
| MTN Up2U bundle codes | 11,949 | 170,112 | 7.02% | 6.6 |
| Telecel Ghana short codes | 7,012 | 792,767 | 0.88% | 8.3 |
| MTN Mobile Money charges | 6,971 | 254,300 | 2.74% | 9.0 |
| MTN fibre broadband packages | 5,696 | 255,716 | 2.23% | 8.8 |
| Telecel internet settings | 4,901 | 97,492 | 5.03% | 5.9 |
| MTN Ghana internet settings | 4,812 | 141,208 | 3.41% | 9.0 |
| Reverse MTN MoMo transaction | 4,514 | 291,166 | 1.55% | 7.9 |
| Check BECE school placement | 4,076 | 447,136 | 0.91% | 11.4 |
| MTN AFA bundle registration | 3,985 | 271,439 | 1.47% | 6.1 |
Writing quality explains part of that. The bigger factor was how replaceable the topic was by a summary.
The pages that held up better usually had a few things in common:
- It could go stale, so readers wanted the current version.
- It helped someone complete an action rather than learn a fact.
- It carried enough detail that a summary often fell short.
That is a better test than search volume. If the full value of an article fits comfortably inside a search result, the page is exposed before it is even published.
AI is now a reader too
One of the quieter changes is that a blog post now has two audiences. Alongside the person who lands on the page, it may be read by a search engine, an AI answer engine, a browser assistant, a research tool, or an internal system someone consults before they ever visit the site.
That raises the value of being clear, current, and specific. A thin post still gets indexed while giving those systems almost nothing to work with. A useful post becomes part of how a topic, product, service, or company is understood.
Muck Rack’s AI citation research found that non-paid media, including corporate blogs and journalism, made up a large share of links cited by AI tools. Few individual posts will earn a citation. The point is that owned writing remains part of the information layer these systems draw from.
This is where blogging has become more interesting. The post is often the source behind the answer, the link someone checks later, or the piece that makes a claim easier to trust.
The writing has to carry more
A blog post that repeats the first page of Google has little room left. The web has enough of that, and AI produces more of it every day.
The stronger work has a reason to exist beyond coverage. It has lived detail in it: what happened, what changed, what was harder than expected, what the numbers actually showed, what trade-off was made, what someone should know before making the same decision.
It can still be practical and still answer a search query. The value should sit in the judgement around the answer, which is the part a summary struggles to replace, because judgement depends on context, taste, constraints, and evidence.
What we would write now
If we were building a blog today we would still care about search, and we would stop letting search volume decide everything.
We would write more of the pieces that make a business easier to understand:
- Case studies with the uncomfortable parts left in.
- Build notes that explain decisions as well as outcomes.
- Guides tied to real actions people need to complete.
- Comparisons that help someone choose rather than rank keywords.
- Research from products, clients, markets, or systems we can actually observe.
- Process notes that show how the work is done when nobody is performing.
That kind of blogging is slower and much harder to fake. It gives human readers something to judge and AI systems something better than another reshuffled answer.
What to measure instead
Traffic still matters as one measure among several. A page with low traffic can answer an important objection. A page with modest reach can support a proposal, help a returning reader, or make a company easier to cite. A page with high traffic can sit there earning nothing.
| What the content is for | Better signs to watch |
|---|---|
| Proof | Referenced in sales conversations, linked from credible sites, reused in proposals |
| Signal | Cited by AI tools, linked by other writers, showing up around narrow topics |
| Conversion | Qualified inquiries, faster decisions, better-fit conversations |
| Owned distribution | Email signups, replies, return visits, subscriber-to-customer movement |
| Search presence | Impressions on useful queries, branded search, clicks on pages that require depth |
The metric has to match the job. Search traffic is one job. Trust, memory, citation, education, and conversion are others.
Small things still help
None of this needs a new doctrine. Much of the practical work stays small and unglamorous.
Make sure crawlers can reach the pages that matter. Keep the sitemap clean. Maintain an llms.txt file if the site has a set of pages that should be easier for AI systems to understand. Submit the site where search engines still expect to receive sitemaps. Show the author, date, and source of the work clearly.
Inside the writing, make sections strong enough to stand on their own. Use specific claims. Put the useful answer near the useful question. Cut the long passages that sound thoughtful while saying nothing clear enough to carry forward.
And give good content somewhere to send people. Not every article needs to end in a sales pitch. But if a piece earns trust, there should be a sensible next step: another useful post, a newsletter, a product page, a contact path, or a case study with more depth.
Where the data came from
The primary data in this article comes from Ahrefs and Google Search Console data for Asetena Pa, covering December 2024 through May 2026. The traffic figures, position metrics, CTR percentages, and SERP feature counts come from that data.
We also reviewed outside research from Ahrefs, Authoritas, BrightEdge, Chartbeat, WordStream, Muck Rack, Search Engine Journal, the Princeton/Georgia Tech GEO-bench study, and a 2026 study on Google AI Overviews to understand how our own numbers fit into the broader shift.
The version that still works
The easy version of blogging was publishing around keywords and waiting for traffic. That version has run out of room.
The version that still works asks more of you: write what you know, show the evidence, keep it current, make the thinking visible, and give people a reason to open the page when a summary already exists.
That is a higher bar than the old one, and it is the bar we are holding ourselves to here. The Asetena Pa numbers are why. We would rather publish fewer pages that survive being summarised than a pile that disappears inside someone else’s answer.
