Real Search Console data. Every number from our own GSC account. No projections, no benchmarks from someone else's study.
187 AI-assisted blog posts. 7 weekly Google Search Console snapshots. 1,522 total impressions. 13 total clicks. Only 17-37% of posts appeared in search results in any given week. The posts that ranked weren't the ones we expected. Volume didn't matter nearly as much as specificity. Here's every number.
Most "does AI content rank?" articles give you theory. We have data.
We published 187 blog posts on syxoai.com between February and June 2026. Every post was AI-assisted (drafted with Claude or ChatGPT, then edited for voice and accuracy). We pulled Google Search Console data every week via the API.
This is what 8 weeks of real ranking data looks like for a new site running AI content at scale. If you're building an AI marketing system, these numbers will save you months of wrong assumptions.
Source: Google Search Console API, pulled weekly via service account (automated script).
Period: 7 weekly snapshots from 13 April to 7 June 2026.
Domain: syxoai.com (registered late 2025, first content published February 2026).
Post count: 187 HTML blog posts as of 11 June 2026.
Content process: AI-drafted (Claude/ChatGPT), voice-calibrated against a brand voice profile, manually reviewed for accuracy. No auto-published raw AI output.
What we tracked: Weekly impressions, clicks, pages appearing in GSC, average position, position distribution by page, query-level data.
Across all pages. That's 217 impressions per week on average for 187 posts.
Overall CTR: 0.85%. Most clicks came from branded queries.
Best week (May 25-31). That's 37% of all published posts appearing in GSC.
Jun 1-7. A 70% drop from the previous week. Volatility is the reality.
This is the full weekly trend. Every number pulled directly from our GSC account.
| Week | Period | Impressions | Clicks | Pages visible | Avg position |
|---|---|---|---|---|---|
| 1 | Apr 13-19 | 276 | 1 | 31 | 38.7 |
| 2 | Apr 20-26 | 279 | 1 | 41 | 24.8 |
| 3 | May 2-8 | 117 | 3 | 33 | 19.1 |
| 4 | May 9-15 | 214 | 2 | 49 | 17.2 |
| 5 | May 16-22 | 283 | 4 | 47 | 43.3 |
| 6 | May 25-31 | 271 | 1 | 69 | 30.2 |
| 7 | Jun 1-7 | 82 | 1 | 52 | 49.2 |
The pattern isn't "publish more, get more impressions." Impressions spiked to 283, then crashed 70% the next week. For a new domain, this kind of volatility is normal. Google is testing your pages in results, pulling them back, and re-testing. If you're checking your stats daily, you'll panic for no reason.
We published 187 blog posts. In any given week, only 31 to 69 of them appeared in Google Search Console at all. That's a visibility rate of 17% to 37%.
| Week | Posts published | Pages in GSC | Visibility rate |
|---|---|---|---|
| 1 (Apr 13-19) | ~120 | 31 | 26% |
| 2 (Apr 20-26) | ~130 | 41 | 32% |
| 3 (May 2-8) | ~140 | 33 | 24% |
| 4 (May 9-15) | ~155 | 49 | 32% |
| 5 (May 16-22) | ~165 | 47 | 28% |
| 6 (May 25-31) | ~175 | 69 | 39% |
| 7 (Jun 1-7) | ~185 | 52 | 28% |
This doesn't mean 63-83% of our posts are de-indexed. It means they're indexed but not generating any impressions. Google knows they exist. Nobody is searching for what they cover, or they're ranked too deep to register an impression.
The practical takeaway: publishing 187 posts doesn't give you 187 ranking shots. It gives you 30-70 at best, and those rotate week to week.
These numbers come from running a real system. If you want to build one that doesn't waste 60% of your output, start with the workflow.
See the SystemWe categorised every page that appeared in our Week 7 (Jun 1-7) snapshot by position band.
| Position band | SERP location | Pages | % of visible pages | Clicks from this band |
|---|---|---|---|---|
| 1-10 | Page 1 | 27 | 52% | 1 |
| 11-20 | Page 2 | 4 | 8% | 0 |
| 21-50 | Pages 3-5 | 10 | 19% | 0 |
| 50+ | Deep | 11 | 21% | 0 |
27 pages sat at positions 1-10. They generated 1 click. Why? Because most of those "page 1" rankings are for ultra-low-volume queries with 1-6 weekly impressions. Ranking page 1 for "ai content service for coaches" sounds great in a report. It means almost nothing in traffic until search volume catches up.
13 clicks from 1,522 impressions across 7 weeks. Here's where they came from.
| Page | Click source | Clicks | Week |
|---|---|---|---|
| Homepage (/) | Branded ("syxo") | 9 | Weeks 1-7 |
| /guides/ai-marketing-systems-for-solopreneurs | Branded | 1 | Week 1 |
| /services/dfy-voice-system | Branded | 1 | Week 2 |
| /blog/content-strategy-one-person-business | Non-branded organic | 1 | Week 3 |
| /blog/build-ai-marketing-workflow | Non-branded organic | 1 | Week 5 |
11 of 13 clicks were branded. People searching "syxo" and clicking our site. Only 2 clicks came from non-branded organic search in 7 weeks.
That's the real number. Two organic clicks from 187 posts over nearly two months. Every blog post that tells you "publish 3x a week and the traffic will come" is leaving out this part.
Not all 187 posts performed equally. The pattern was clear by week 4.
| Post | Type | Best position | Peak impressions/week |
|---|---|---|---|
| Homepage | Brand | 2.8 | 27 |
| Voice prompts guide | Long-form reference | 5.1 | 18 |
| LinkedIn ghostwriter cost (UK) | Buyer-intent niche | 6.7 | 28 |
| What is a voice prompt | Definition/glossary | 5.7 | 6 |
| How to build a voice prompt | How-to tutorial | 5.2 | 5 |
| ChatGPT vs Claude for LinkedIn | Comparison | 9.4 | 7 |
| Build AI marketing workflow | How-to tutorial | 4.9 | 15 |
| AI prompts vs AI systems | Opinion/framework | 7.4 | 9 |
| Post | Type | Typical position | Likely reason |
|---|---|---|---|
| ChatGPT vs Jasper marketing | Comparison (competitive) | 67-82 | Dominated by high-DR sites |
| AI keyword research beginners | Broad how-to | 27-110 | Extreme volatility, high competition |
| AI landing page copy | Broad how-to | 54-92 | Saturated topic |
| LinkedIn ghostwriter alternatives | Listicle | 66 | Competing with aggregators |
| Best AI tools for LinkedIn content | Listicle (competitive) | 94 | Saturated, high DR competition |
The pattern is obvious in hindsight. "What is a voice prompt" (position 5.7) ranks. "Best AI tools for LinkedIn content" (position 94) doesn't. The first targets a term nobody else has written a definitive page for. The second competes with HubSpot, Hootsuite, and Buffer. New domains win on specificity, not on publishing speed.
One page's position across 5 weeks. This is /blog/ai-keyword-research-beginners, our highest-impression post.
| Week | Position | Impressions | What happened |
|---|---|---|---|
| Apr 13-19 | 48.2 | 120 | Initial indexing, broad terms |
| Apr 20-26 | 45.8 | 110 | Slight improvement |
| May 2-8 | - | 0 | Disappeared entirely from GSC |
| May 9-15 | 50.7 | 3 | Returned, much fewer impressions |
| May 16-22 | 28.5 | 14 | Position improved after title rewrite |
| Jun 1-7 | 27.8 | 4 | Holding but impressions dropped |
This post went from 120 impressions to zero to 14 in three consecutive weeks. It wasn't deindexed. Google was re-evaluating where it belonged. If you're making ranking decisions based on a single week's data, you're reacting to noise.
We grouped all 187 posts by content type and compared their average best position across the 7-week window.
| Content type | Posts published | Avg best position | % reaching page 1 |
|---|---|---|---|
| Definition/glossary pages | 8 | 5.4 | 75% |
| Niche comparison (2 tools) | 12 | 8.2 | 58% |
| How-to tutorials (specific) | 35 | 9.8 | 43% |
| Opinion/framework posts | 15 | 11.3 | 40% |
| Vertical ("AI marketing for [role]") | 22 | 14.7 | 32% |
| Broad listicles ("best X tools") | 18 | 42.1 | 11% |
| Broad how-to (competitive terms) | 25 | 55.8 | 8% |
Definition pages and niche comparisons crushed everything else. The worst performers were broad listicles and competitive how-to guides. Both content types that every AI writing tool defaults to.
If you're using AI to write "10 Best AI Tools for Marketing" on a new domain, you're producing content that will rank at position 42 on average. That's page 5. Nobody scrolls to page 5.
Our top non-branded queries over 7 weeks, showing what people actually searched to find our content.
| Query | Total impressions | Best position | Clicks |
|---|---|---|---|
| keyword research in 90 minutes | 52 | 5.3 | 0 |
| seo workflow | 97 | 77.8 | 0 |
| chatgpt jasper (all variants) | 48 | 71.0 | 0 |
| ai keyword research (all variants) | 127 | 55.5 | 0 |
| how much does linkedin ghostwriting cost | 10 | 6.7 | 0 |
| voice prompts (all variants) | 16 | 5.1 | 0 |
"AI keyword research" generated 127 impressions at an average position of 55.5. We have the content. We have the impressions. But we're on page 6. That's a signal to consolidate and strengthen the page, not to publish more keyword research posts. "Voice prompts" generated 16 impressions at position 5.1. Lower volume, but we own the space. That's where the next click comes from.
1. Publishing speed doesn't correlate with ranking speed. We shipped ~25 posts per week in our biggest sprint. It didn't accelerate rankings. Google evaluates pages individually. A batch of 50 posts in a week ranks no faster than 3 good ones.
2. 60-80% of your posts won't generate impressions. Of 187 posts, a maximum of 69 appeared in GSC in our best week. Most weeks it was 31-52. This isn't failure. This is how new domains work. The 20-40% that do appear are your growth surface. Focus there.
3. Definition pages rank fastest. "What is a voice prompt" hit position 5.7 within weeks. Definitions are underserved, they match exact search intent, and they're structured in a way that Google likes to surface. If you're planning content for a new site, start with definitions your competitors haven't written.
4. Position volatility is brutal for new domains. Pages moved 20+ positions between weeks routinely. Our keyword research post went from 120 impressions to zero to 14 in three weeks. Don't optimise based on a single snapshot. Wait for 4+ weeks of data before making title/content changes.
5. Clicks come last. We expected our first consistent organic clicks around month 3-4. The data confirms this. 13 clicks in 7 weeks, 11 branded. Non-branded organic clicks are the trailing indicator, not the leading one. Impressions and position improvements come first, often by weeks.
If you're building a content strategy as a one-person business, these timelines matter more than any best-practice article. The real answer to "how long until SEO works?" is longer than you expect and more volatile than anyone shows you.
We've documented the full AI content workflow: voice capture, batch drafting, quality control, and SEO tracking. It's the same system we used to publish 187 posts.
See the Full WorkflowYes, but slowly and inconsistently. Of our 187 AI-assisted posts, 17-37% appeared in GSC in any given week. The posts that ranked best targeted niche, low-competition topics. Broad competitive terms rarely broke page 2 regardless of content quality.
New posts typically appeared in Search Console within 1-2 weeks. But appearing in the index isn't the same as ranking. Most indexed posts sat at positions 30-80 for weeks. Some never improved. Median time to page 1 for any query was 4+ weeks, and most posts never reached page 1.
Very low. We recorded 13 clicks from 1,522 impressions over 7 weeks, a CTR of 0.85%. Most clicks were branded queries. Non-branded organic CTR was near zero for the first 8 weeks. This is normal for a new domain.
Publishing 187 posts gave us 31-69 visible pages per week. Volume alone didn't drive rankings. Tightly focused, low-competition content outperformed everything else. A site with 20 specific, well-targeted posts could outperform 187 broad ones.
We saw no evidence of an AI penalty. Posts were indexed normally. The issue wasn't detection. It was quality and competition. Generic AI content ranked poorly because it competed with established sites. Niche, voice-calibrated content performed far better on the same domain.