I Tracked 50 Keywords for 90 Days: What the SERP Volatility Taught Me

It started as a bet with myself. I'd spent years doing SEO by feel — publishing content, building links, checking rankings every couple of weeks, and calling it "monitoring." I convinced myself that was enough. Then one of my sites dropped 40% in organic traffic over a weekend with no obvious cause, and I realized I had no idea what had actually happened or when.

So I started tracking properly. Fifty keywords, daily snapshots, three different sites across two niches. I set it up in mid-January and didn't stop until late April. Here's what I learned — and some of it genuinely surprised me.

The Setup (Because It Matters)

I used a combination of tools. SE Ranking for daily position tracking — it's cheaper than Semrush for high-frequency pulls and the interface doesn't make me want to close the tab. Google Search Console for actual impression and click data, which I pulled weekly and cross-referenced manually. And a basic spreadsheet where I logged any external events: algorithm updates, significant site changes I made, competitor posts going live.

The 50 keywords weren't random. I picked them deliberately across three tiers:

• Tier 1 (18 keywords): High-volume, competitive terms where I was ranking on pages 2–3 and trying to break into page 1.
• Tier 2 (22 keywords): Mid-volume terms where I was already ranking in positions 4–15 — the "almost there" keywords.
• Tier 3 (10 keywords): Long-tail, low-competition terms where I was already ranking 1–3 and wanted to see how stable those positions actually were.

That tiered breakdown ended up being one of the best decisions I made, because the behavior in each tier was completely different.

Week One: False Confidence

The first week was boring in the best possible way. Rankings barely moved. A couple of Tier 2 keywords ticked up by one or two positions, one Tier 1 keyword bounced from 24 to 19 for no reason I could identify. I thought: this is going to be easier than I expected. I might crack page one within a month.

I was wrong.

The First Real Volatility Hit

Around day 16, something happened. Seven keywords moved simultaneously — some up, some down — by five positions or more. A keyword I'd been sitting comfortably at position 6 for weeks dropped to 14 overnight. A Tier 1 keyword I'd given up on suddenly appeared at position 8.

I cross-referenced with MozCast and SERPmetrics. Both showed elevated temperatures for those days — signs of an unconfirmed update. Google didn't announce anything. But the pattern was unmistakable: something in the algorithm had shifted, and it affected different keywords differently based on — what exactly? That was the question I spent the next two months trying to answer.

What I Noticed About Volatility Patterns

After 90 days of daily data, a few things became genuinely clear:

1. Tier 3 Keywords Are Not Actually Safe

I assumed my long-tail rankings were bedrock. They weren't. Three of my ten Tier 3 keywords dropped out of the top 5 at some point during the 90 days — two of them temporarily, one permanently. The permanent drop happened after a larger site published a more comprehensive piece on the same topic. I'd been ranking at position 2 for a "how to" query with a 600-word post. They published 2,400 words with a video embed and schema markup, and within three weeks they owned that spot.

The lesson: low-competition doesn't mean low-maintenance. I was coasting on old content in a space that was slowly getting more competitive.

2. Positions 4–10 Are the Most Volatile Zone

This one genuinely surprised me. I expected the biggest swings in Tier 1, where I was ranking in the 20s and 30s. But the Tier 2 keywords — already on page one in positions 4 through 10 — showed the most week-to-week instability. Some of them moved four or five positions in both directions within the same week.

I think I understand why now. Positions 1–3 are usually locked down by high-authority pages that Google is confident about. Positions 11+ are far from page one and Google is still figuring out intent match. But that middle band — 4 through 10 — is where Google is actively experimenting, rotating pages in and out to test user behavior signals.

The implication for keyword research: when you see a "page 1 ranking" at position 7 in your rank tracker, don't celebrate too early. That position can evaporate in a week.

3. Freshness Signals Were More Powerful Than I Thought

I updated three existing posts around the six-week mark — not major rewrites, just adding a section, fixing outdated stats, changing the published date. Two of those posts jumped 3–5 positions within days. The third didn't move at all.

The one that didn't move was a post where I'd updated the metadata and one paragraph but left the core content unchanged. The two that moved had genuinely new information — a new subheading, new examples, 300+ new words. Google noticed the difference. A fake freshness update (just changing the date) does nothing. Actual new content in an existing URL, especially if it addresses search intent more completely, gets rewarded quickly.

4. Featured Snippet Wins (and Losses) Are Brutal

One of my Tier 2 keywords had a featured snippet attached to it. I had been holding position 1 with the snippet for weeks, getting roughly 60% of the clicks on that query. Around day 55, I lost the snippet to a competitor — they restructured their answer as a numbered list and Google preferred it. My clicks dropped by half even though my organic position stayed at #1.

That's the thing about tracking position alone: it doesn't tell you about snippet ownership, "People Also Ask" boxes, or how the SERP layout is changing around your result. I started using Semrush's SERP feature tracker midway through the experiment specifically because of this. The SERP is a page, not just a list of ten links, and you need to track what's happening to the whole page.

The Algorithm Update That Changed Everything

Around day 70, there was a confirmed core update — Google announced it officially about two weeks after it started rolling out. During those two weeks before the announcement, my data was already showing something significant: 11 of my 50 keywords moved by more than 8 positions in a seven-day window. Eight went up. Three went down hard.

The three that dropped all had something in common: the content was answering the keyword's question but wasn't demonstrating any real expertise or depth. They were "good enough" articles that had ranked because competition was thin. The update seemed to prefer pages that showed more topical authority — not just content that matched the query, but content that existed within a broader context of related content on the same site.

I'd been running these three pages on a fairly thin blog with no clear topical structure. The pages that held or gained were on a site where I'd built out interconnected content clusters. That context — the internal linking structure, the breadth of related content — seemed to matter more after the update than it had before.

What I'd Do Differently With Keyword Research Now

Ninety days of data changed how I approach keyword selection. A few specific shifts:

I now filter for SERP volatility before I commit to a keyword. Tools like Semrush and Ahrefs both have some measure of SERP volatility now. I look for keywords where the top 5 has been relatively stable over 90 days — not because I think I can predict the future, but because stable SERPs usually signal that Google has a confident answer and strong intent match. Breaking in is harder, but once you're in, you stay.

I stopped chasing keywords where a featured snippet dominates CTR. Unless I have a genuine shot at owning the snippet, I'd rather rank #1 on a query where 10 blue links get the clicks than rank #2 on a snippet-dominated query where the snippet steals 70% of them.

I pay much more attention to the full SERP composition. How many ads? Is there a local pack? Knowledge panel? Video carousel? A keyword that shows "position 5" might be the 12th element on the actual page. That changes the click math completely.

The Honest Conclusion

Tracking 50 keywords daily for 90 days is a lot of work, and I won't pretend I'd do it again at the same intensity indefinitely. But doing it once gave me a mental model for SERP behavior that I couldn't have built any other way. Reading about volatility is one thing. Watching a keyword you care about bounce from 6 to 18 and back to 9 over three weeks, with no explanation, is another.

The biggest takeaway: rankings are not a destination. They're a snapshot of where you are in an ongoing conversation with Google's algorithm. The sites that win long-term aren't the ones that find a trick to rank — they're the ones that build content and site structures that Google keeps coming back to as the right answer, update cycle after update cycle.

I'm still tracking. Just 20 keywords now, with more selective attention. That feels about right.