Rank Tracker Accuracy: Is Your Tracker Lying to You Right Now?

An accurate rank tracker is hard to find this week: Semrush has confirmed that Google is returning inaccurate results to some automated ranking tools, and Google hasn't said a word. So that red line on your chart might not be a real drop.

Most people see a sudden fall in Semrush, Ahrefs or BrightLocal and assume they've lost rankings. Then they rewrite the page or send the client a worried email. This week, that fall could be bad data, a real Google update, or a mix of both.

Below: what's confirmed, how much data is affected, what to trust, what to check in Search Console, and a script for clients. Key claims are labeled Confirmed or Unconfirmed. It's written for business owners, freelancers and agencies.

Is your rank tracker accurate right now?

Not reliably, and not since mid-September. Semrush has confirmed that Google is sending some automated ranking tools "poisoned" results, pages that don't match what a real searcher sees, while Google itself has not confirmed anything.

Confirmed (by Semrush): Search Engine Land reported on October 27, 2026 that Semrush "confirmed that Google is returning inaccurate search results to some automated ranking tools." Ahrefs, AccuRanker, BrightLocal and other rank trackers have reported similar issues since September.

A poisoned SERP is a results page your tool collected that doesn't reflect what Google really ranks for that search. Semrush describes two types:

• Irrelevant results: results that don't match the search query, sometimes on page one.
• Partially inaccurate results: pages that look normal at first but contain loosely related results further down.

Semrush's example is almost funny, mind you. A search in Brazil for a specific baby stroller model returned a dental supply store's equipment page, because both products share the Portuguese word "carrinho." Store a page like that and the position for that day is fiction.

Unconfirmed: why it's happening. Semrush and other providers have suggested Google's efforts to restrict large-scale automated data collection, including collection for AI systems, may explain it. Google has not confirmed that explanation, and Google has not confirmed that it deliberately serves altered results to automated ranking tools. Semrush's own advice for unusual ranking changes: compare your reports with Google Search Console.

How much tracker data is affected?

Vendors report anywhere from under 1% to over 20% of results, depending on the tool, the test and the week. Each vendor measured something different, so read every figure as that tool's own sample, not one industry number.

In an October 22 update, Ahrefs said 5% to 20% of search results were affected, depending on the data provider. Here's what each vendor reported, as compiled by Search Engine Land.

Search Engine Land is clear on this: "The figures come from different samples and detection methods, so they are not directly comparable." So don't average them.

Same vendor, different test: Search Engine Journal reports that Nozzle's first-page count, which flags a first page when less than 50 percent of its normal first-page sites show up, came to 8.7 percent from Sept 13 to Oct 21, versus 0.17 percent in its baseline. That's a different measure from the table's 22.6% (Page 1 and 2 together).

If you do local SEO, there's a twist. BrightLocal said the problem "primarily affected rankings below position 10." But Semrush, AccuRanker and Nozzle also documented inaccurate results on Page 1, and Nozzle saw suspected poisoning on 17.9% of evaluated SERPs on Page 2 versus 12.2% on Page 1. So don't assume your page-one service + city keywords are safe. Where you really land still comes down to our local SEO guide.

Was Oct 27 a real Google update or just bad data?

Nobody knows yet. Volatility tools spiked around October 27, but Google hasn't confirmed an update, and those tools read the same scraped results pages that are being poisoned, so a spike could be a real update, bad data, or both.

Unconfirmed: on October 27, Barry Schwartz at Search Engine Roundtable reported "a larger spike in volatility" plus more SEO community chatter, and wrote that "Google hasn't officially confirmed anything yet." It followed other unconfirmed moves he reported on October 16, October 20 and 21, and October 24. His October 29 update: "things keep getting hotter," and he asked on X, "is it poisoned or is Google brewing something."

Confirmed: the Google September 2026 spam update started on September 24 and completed on October 8, 2026, per the same post. Barry also noted that "the Google tracking tools are struggling."

Even the tool makers say the signals overlap. DemandSphere acknowledged that algorithm updates, Google tests and normal ranking changes can produce similar patterns to poisoned results.

Now, the bit most people miss. Our read: volatility sensors are built from collected results pages too, so poisoned pages can skew them as well. Treat a sensor like a weather report for the whole industry, not a verdict on your site.

Is num=100 back, and why does it matter for your numbers?

Maybe, in a small way, but it's unconfirmed. Google switched off the 100-results-per-page parameter in September 2025, which changed both tracker data and Search Console impressions, so a return could shift your numbers again.

Confirmed history: in September 2025, Google stopped the num=100 URL parameter, which let anyone load 100 results on one page. A Google spokesperson told Search Engine Land: "The use of this URL parameter is not something that we formally support."

Search Console felt it too. In an analysis of 319 properties by SEO Tyler Gargula, 87.7% of sites lost impressions and 77.6% lost unique ranking terms. Search Engine Land noted some of that impression data may not have been real in the first place, because rank trackers and scrapers using the parameter were inflating impressions.

Unconfirmed now: on October 30, SEO Brodie Clark wrote that "Google seems to have brought back the ability to use num=100 in some capacity" and that "there were 99 organic results showing" after he clicked a knowledge panel result. Barry Schwartz tried it: "I tried replicating this across multiple browsers, and I cannot replicate it." Brodie also flagged /goto links bypassing previous workarounds, "likely an issue for some rank trackers too."

Our reasoning, not a reported fact: if num=100 comes back widely, deep-position tracking and possibly long-tail impression counts in Search Console could shift again. Write the date in your notes, and don't read a sudden impression jump as a ranking win until clicks agree.

Why do trackers disagree on how often AI Overviews show (62% to 82.5%)?

Because each figure comes from a different tool, sample and query mix. It's the same lesson as rank positions: every number is one tool's sample, not a census of Google.

Search Engine Roundtable reported AI Overviews now show for "over 80% of Google Searches," based on data from Semrush, SE Ranking and seoClarity, yet SE Ranking's 82.5% covers "US brand searches" while seoClarity sees them "for over 62% of queries but it can be 55% to 84% depending on industry." Rather than trust anyone's average, measure your own AI Overview exposure on your own keywords.

What should you trust, and what should you ignore this week?

Trust Search Console and your leads first, a manual spot check second, and your daily rank tracker last for now. Ignore one-day swings on page two until they hold for a few days.

Search Console is real Google data from real searches. Per Google's help page, average position is "the average position of the topmost result from your site," and results "are specific to the time, place, device, and recent history of the person searching." So it will never match one tracker check exactly. Here's the trust ladder I'd use this week.

And here's how to read the two together. This matrix is our working rule of thumb, not a Google or vendor rule.

Bottom line: this is a shitty week to make big decisions off one chart. Two sources that agree is a signal.

If you can't trust the rank trackers right now, the real question becomes what you can track. When the charts go weird, I stop staring at position swings and climb this ladder instead. It's how I get a data-driven read of what's actually working:

1. Search appearances. Are you being shown at all, in classic results or in AI answers?
2. Citations. Is search or an AI answer citing your page or your brand?
3. Mentions. Are people naming you in a good way, even when nobody clicks through?
4. Clicks. Did someone actually visit?
5. Conversions. Did it turn into a call, a booking or a walk-in?

Mind you, conversions can show up with no click at all. Someone gets a recommendation in ChatGPT or a Google AI Overview, then phones you or walks in without ever hitting your site. So positive mentions and clicks are the useful mid-funnel signals. Conversions close it. Rank positions are still worth watching over weeks, but this week the ladder above is the shining star for deciding whether anything real moved.

What should you check in Search Console and DataWise before you act?

Before you rewrite a page or warn anyone, run this check on the flagged keyword in the standard Search Console Performance report.

1. Open the Performance report (Search results) in Search Console.
2. Turn on all four metrics: clicks, impressions, CTR and average position. Position down with flat clicks is a very different story from both falling.
3. Filter to the query or page your tracker flagged. A site-wide chart can hide a one-keyword problem.
4. Match the country and device your tracker checks. A US desktop tracker and an all-countries view won't line up.
5. Switch to daily view and compare the flagged day with the days around it. Google notes the newest data can be preliminary and might still change.
6. Compare the last 28 days with the previous 28 days.
7. Wait about three days before you act. Nozzle found 67.9% of affected SERPs returned to their usual results within three days.
8. Check that the ranking URL didn't switch. In the Pages tab for that query, see if a different page of yours took over. A swap looks like a drop in a tracker.

If clicks and impressions really did fall for several days, it's no longer a data question. Work through our SEO audit guide on that page before you change anything else.

If you're an AI Ranking member, DataWise connects your Search Console and tracks keyword positions over time, so you can look at both numbers for the same query in one place. Its tracker reads Google results too, so check it against Search Console like any other.

A real example from our own site, pulled on November 2, 2026:

That's the branded query "ai ranking skool." The tracker says 4 this week. Search Console says 7.3. Neither is lying. A tracker records one check from one place at one moment; Search Console averages your topmost position across real searches over a date window. A tracker and Search Console measure different things, so this gap isn't proof of poisoning or of a Google update.

Site-wide, our Search Console daily data shows no drop around October 27: 12 clicks and 2,189 impressions on October 27, and 5 clicks and 2,832 impressions on October 30.

How do you explain ranking swings to a client (or your boss)?

Lead with outcomes, then positions, and always show the date the positions were collected.

Old reports are a trap right now. Search Engine Journal suggests that if you resend September or October data, ask your vendor whether data stored for those dates was updated, because older exports may not match what the tool shows today. Some tools also hold numbers back: SEOmonitor "may keep showing the previously recorded rank for up to two daily updates" after a big drop while it checks. And BrightLocal says there's "no single fix that solves it completely."

• Report clicks, leads and Search Console trends first, tracker positions second, with the collection date next to them.
• Mark the known dates on your charts: the confirmed September 2026 spam update (Sept 24 to Oct 8) and the unconfirmed October 27 volatility.

Here's a plain-English script you can adapt. It's a template, not a real client message, so swap in your own numbers:

"Hi [name], quick note on the ranking chart. Since September, several rank tracking tools, including Semrush, have reported that Google is returning inaccurate results to automated tools, and Google hasn't commented. So some daily positions in our tracker may be wrong. Here's what Google's own data shows: in Search Console, clicks to [page] were [number] in the last 28 days versus [number] the 28 days before, and we had [number] leads. We'll recheck [keyword] on [date] before changing anything. If the drop shows up in Search Console too, our next step is [plan]."

I reckon that one email saves heaps of panicked calls.

Don't panic, verify. Search Console and your leads tell you what happened; your tracker tells you where to look.

Not sure if a drop is real? Bring your tracker chart and your Search Console to a Premium live Q&A or a website review inside AI Ranking Premium, and we'll look at it together. Not ready for that? Start with the free AI Search Kickstarter.

Cheers,
Nico