Crit Pick, three weeks of social: 453,000 views, 71 visits
Historical Crit Pick observations: platform views, attributed visits and search changes, with the limits of each measurement explained.
Across the period, the platform metrics totalled 453,370 views and Ghost attributed 71 visits to social sources. Those figures use different definitions and coverage. Search also changed, but this window cannot establish whether social contributed to that change.
That's the whole result in three sentences. The rest of this post is the detail, including the part I can't explain yet.
Seventeen weeks in, data through · Ghost web analytics (location-filtered visits), Google Search Console, and my scheduler's own post metrics · Crit Pick is the games site I run in the open as a live case study for SeeIndie.
- Social reach was real. Three weeks of daily posting on Instagram, TikTok, YouTube, X and Bluesky, 10 to 30 September, collected about 453,000 views.
- The view total and social-attributed visits are separate observations. They cannot be divided into a measured view-to-visit conversion rate because the included platforms, users and attribution do not match.
- Riding a trend worked for attention. 84% of all views came from posts about one game, Aniimo, which went live during the window. Its pages went from 2 visits in the two weeks before to 264.
- The available numbers do not establish social's effect on the other game sections or describe individual onward-browsing paths. Topic totals are not a measure of spillover from one post.
- Google changed the picture in week three. Search impressions jumped from about 50 a day to about 4,000, average position from around 18 to around 8, and clicks from almost none to 135 in seven days. It started the day after Google began rolling out its September 2026 spam update. I can't tell you whether that's cause or coincidence.
What I actually did
On 10 September Crit Pick started posting to five channels: Instagram, TikTok, YouTube Shorts, X and Bluesky. Threads ran alongside through Instagram. Three weeks, every day, scheduled in batches.
Most of it was about Aniimo. It was the game everyone in my corner of the internet was talking about, it went live in the middle of my window, and I had pages ready for it. Riding a popular topic is the oldest trick in social media, and I wanted to see what it does for a small site that mostly lives off search.
The rest was the site's usual mix: RimWorld guides, Guild Wars 2 builds, Palworld, WoW Forever, demo reviews and upcoming games.
Here's where the views came from. Most pieces went out on several channels at once, so each channel counts its own views.
| Channel | Views |
|---|---|
| 238,422 | |
| TikTok | 189,055 |
| YouTube | 25,122 |
| X | 771 |
| Bluesky | no view count |
| Total | 453,370 |
Bluesky doesn't report views to my scheduler, so it isn't in the total. It got nine likes in three weeks, if you're wondering.
Can this case show a social effect on SEO?
Separately, Ghost attributed 71 visits to social sources. The platform-view total and attributed visits have different coverage; this is not a matched conversion funnel.
| Source | 27 Aug – 9 Sep | 10 – 30 Sep |
|---|---|---|
| 0 | 38 | |
| Threads | 0 | 9 |
| Bluesky | 0 | 7 |
| 0 | 6 | |
| YouTube | 0 | 6 |
| X | 0 | 5 |
| Social total | 0 | 71 |
71 visits. For comparison, the site had 527 visits in the same three weeks once I'd taken the bot traffic out, so social was about one visit in seven.
And it faded fast. Social sent 28 visits in week one, 31 in week two and 12 in week three. Views fell the same way: about 247,000, then 158,000, then 48,000. The trend I was riding peaked early and my reach went with it.
One gap I want to flag rather than paper over: TikTok never shows up as a source in my analytics, even though TikTok was my second-biggest channel by views. Visits that come through in-app browsers often arrive with no referrer at all, so some social traffic is probably hiding in "direct". Direct visits did go up, from about 2.5 a day before the window to about 5.5 a day during it. I can't split that out, so I'm not counting it. Even if you gave all of it to social, the picture wouldn't change.
What riding Aniimo did
Aniimo attracted more recorded visits in the later window, but the comparison uses 14 days before and 21 days after, alongside release interest and search changes. It does not isolate the effect of the social posts.
The two most-viewed posts of the whole run were both Aniimo quick-reference posts: a list of the six free Prismana (57,000 views on TikTok) and three things to do first when you start (56,000 on Instagram). Short, useful, about a game people were opening that same day.
But look at what happened to everything else in the first two weeks, while social was at its peak:
| Cluster | 27 Aug – 9 Sep | 10 – 23 Sep |
|---|---|---|
| Aniimo | 2 | 169 |
| RimWorld | 36 | 23 |
| Guild Wars 2 | 24 | 26 |
| Palworld | 30 | 33 |
Page visits, location-filtered visits. Page visits count per page, so they don't add up to the visit totals above.
The figures shown here are section totals, not tracked journeys between pages. They cannot establish how many Aniimo visitors continued into another game section.
Week three: Google moved, and I don't know why yet
Then the last week of the window happened.
| Search Console | Clicks | Impressions | Avg. position |
|---|---|---|---|
| 27 Aug – 9 Sep (14 days) | 3 | 1,061 | 31.0 |
| 10 – 23 Sep (14 days) | 6 | 781 | 19.1 |
| 24 – 30 Sep (7 days) | 135 | 19,018 | 8.2 |
On 24 September the site had 48 impressions. On 25 September it had 523. On 26 September it had 4,280, and it has stayed around 3,000 to 4,000 a day since. Pages with at least one impression in search results went from 143 in the first two weeks to 537 in the third.
I don't think this is social. Social was already fading by then, and in the first two weeks of posting, while it was at its peak, search impressions were lower than in the two weeks before, not higher. Nor does it look like a wave of newly indexed pages: the index count rose in early September and was flat in the latest index report I have, from 21 September. That report hasn't caught up with the jump yet, so I can't fully rule it out.
What the page-level data does show: about half the pages that appeared in search that week were newly recording impressions in this export, including the creature database on its new subdomain. They collected impressions and almost no clicks. Nine in ten clicks came from pages Google already knew and suddenly ranked higher.
One thing in the table could look like a social effect, so I'll deal with it: average position improved from 31 to 19 in the first two weeks. But it did that on fewer impressions, not more. My best guess is that some low-ranking pages simply stopped being shown, which lifts the average without anyone seeing the site more often. It's a guess, and I'll check it against the page-level data.
What I can tell you is the timing. Google started rolling out its September 2026 spam update on 24 September, and as I write this it's still rolling out. My jump starts the next day. That's timing, not proof. I've been here before: in the 263-articles post I had a decline sitting on top of the August spam update and couldn't separate the two. This time it's a rise, and I'm treating it exactly the same way. What Google is doing with its updates this year is covered in Google core update 2026.
What I can also tell you is where it landed:
| Cluster (24–30 Sep) | Pages with impressions | Clicks | Impressions |
|---|---|---|---|
| RimWorld | 21 | 52 | 4,423 |
| Aniimo | 10 | 22 | 4,117 |
| Palworld | 413 | 5 | 3,586 |
| Guild Wars 2 | 34 | 27 | 3,294 |
The RimWorld animal guide had 203 impressions and zero clicks in the 28 days to 6 September, the window from my last post. In this one week it had 1,999 impressions and 36 clicks. The 413 Palworld database pages got shown a lot more too, and turned that into five clicks. The written guides took the clicks. The generated database got the impressions.
What I'd tell anyone posting on their own
The platform metrics recorded 453,370 views and Ghost attributed 71 visits to social sources. Those observations cannot establish a conversion rate or the value of each view.
The available attribution does not show how attention elsewhere on the site related to the social posts. A missing attributed visit is not proof that no indirect effect occurred.
Someone may encounter a game more than once before taking an action. These data do not establish a required number of exposures or show how many people later searched for it.
Search and social do different jobs. In three weeks, social gave me attention and Google gave me visitors. More of my thinking on what each channel is actually for is in social media for creatives.
What's next
Social posting continues, with less volume per channel and more attention on what actually gets clicked. I'll also start watching searches for Crit Pick by name in Search Console, which is the closest I can get to seeing those repeat contacts. The Search Console re-measurement continues on the dates I already named, and the next one will say whether the week-three level held once the September spam update has finished rolling out.
Common questions
Does social media help SEO?
This case cannot answer that causal question. Search and social changed over different windows, with other events and site changes in play. A sequence in time does not establish an effect or its absence.
Does traffic from social media help rankings?
I cannot establish a ranking effect from these data. Social can bring interested players to a site, but the relationship between those visits and search rankings needs a separate test.
Is SEO still worth it in 2026?
Can I check any of this?
Follow the experiment
The next set of Crit Pick numbers, and whatever I am currently testing. No fluff, no pitch.