Why Your Rank Tracker Is Struggling: Google's Scraping Blocks

Your rank tracker is probably less reliable than it was in August, because Google has been getting better at blocking automated requests to its results pages. As of 14 October 2026, Search Engine Roundtable reports that several tools have posted notices, and Nozzle's research found that about 8% of the results pages it collected since 13 September looked normal but had their usual sites swapped out. Do not rewrite pages because a tracker shows a sudden drop: verify first, and report ranges, not single days.

Key Takeaways

• Google has tightened measures against scrapers through 2026: redirect-wrapped result links, more verification prompts, IP blocking and, per Nozzle, results pages that look complete but are wrong.
• Nozzle's sample (27.5 million results pages) found 8.0% were "poisoned" since 13 September 2026, versus 0.17% in its baseline. This is a vendor study, with its own definition of poisoning.
• Sistrix and Morningscore have each documented that some "drops" were data problems. Morningscore says false drops ran from 24 September to early October.
• Sudden drops that reverse within days across many keywords are a data-quality signal. Confirm with Search Console before changing anything.
• Switch reporting to trends, top-10 counts and share of visibility, and keep manual spot checks as a control.

What happened, in order (as of 14 October 2026)

The timeline pieces together from four sources that disagree on some details, so I list them separately.

• September 2025: Sistrix says Google capped each query at 10 results, so collecting the top 100 now takes ten queries instead of one. Many sites saw fewer Search Console impressions and better average positions from mid-September 2025, which Sistrix attributes to Google's change and not to lost rankings.
• June to July 2026: Search Engine Roundtable reported on 8 July 2026 that result links appeared to route through a google.com/goto address, first spotted on 23 June. The author could not replicate it at the time, so it looked like a test or gradual rollout. Sistrix says it now resolves these for organic results.
• 13 to 18 September 2026: A Roundtable post on 18 September said blocking may have begun around 13 September. Nozzle's Derek Perkins said most of his attempts to pull Google data were blocked, an approximately 80% drop in retrievable data, and DataForSEO saw the same.
• 24 September to 4 October 2026: Morningscore says Google began returning truncated results to many tracker queries at 22:10 UTC on 24 September, and that its tracker logged missing sites as real drops for almost six days.
• 7 October 2026: Sistrix published a statement. It says that in September 2026 it saw divergent results for some queries, now publishes only data that passes extra checks, and switched off on-demand updates.
• 10 October 2026: Nozzle published its "Poisoned Google SERPs" study.
• 12 October 2026: Roundtable's summary post collected the reactions.

Three kinds of bad data, and how each looks

Not all errors look the same. This is my own framing of what the sources describe.

1. Missing data (blocks and captchas)

The tool simply fails to fetch a page. Symptoms: stale "last updated" dates, keywords stuck, tools switching to reduced collection. This is the easiest to notice because the tool usually says so.

2. Truncated results

Morningscore describes results lists that ended early, with about 60% of searches in several markets returning fewer than 40 results instead of about 50 at the peak on 30 September and 1 October. A tracker that treats "not in the list" as "not ranking" records a false drop. Sistrix also says many results lists now end before position 100, so drops beyond the top 10 may reflect shorter lists.

3. Poisoned results

This is the one that worries me. Nozzle's definition: a page is poisoned when fewer than half of a keyword's usual sites (those in the top 10 on at least 60% of baseline-period pages) appear, and they return on a later page. On such pages, organic results kept 20.7% of the previous page's sites against 88.5% in the baseline, and a tracked site fell out of the top 10 on 58.6% of poisoned pages against 2.7% in the baseline. The pages carry no error or captcha. Nozzle attributes this to Google's response to AI scraping; Google has not confirmed that here, and Barry Schwartz suspects the measures target AI bots and SerpApi more than SEO tools.

Nozzle says poisoning peaked at about a quarter of pages on 27 to 29 September, then ran at roughly 5 to 11% a day through 9 October. Eighty percent of poisoning spells lasted a single page, and 77.9% of affected keywords had their usual sites back within three days.

How much weight to give these claims

Be proportionate. Nozzle sells rank-tracking data, so it has an interest in telling this story, but its method is stated openly: baseline period, definition, sample size and country coverage are all published. Morningscore is describing its own incident, including that its own bulk re-checks on 29 and 30 September recorded even more false drops, which is an admission against interest. Advanced Web Ranking is a vendor promoting its own methodology, so I use its article for the explanation of causes, not for product claims. Treat all of them as vendor-reported.

This is not the first crackdown: Search Engine Journal reported a similar scraping clampdown in January 2025, when Semrush and SE Ranking users reported refresh problems. Several practitioners quoted by Roundtable also say the tools were noisy before this. Lily Ray said trackers were too unreliable to share findings. Glenn Gabe compared tools on one site's drop and recovery and found Ahrefs tracked it best, Semrush lagged and Sistrix missed the recovery. That is one site, not a benchmark, and I would not buy or cancel a tool on it.

Why rank trackers differ from what you see

Advanced Web Ranking's post explains why tracked rankings no longer match what you see. Its three reasons: Google now decides how local to make a query, so a country-wide query produces a composite; Google appears to classify automated queries and may serve simplified pages to them; and clean tracker queries lack click-based personalisation. It also says Search Console is not a clean tiebreaker, because average position blends queries, devices and AI Overview citations. I agree with that last point from experience. Use Search Console for direction, not for judging one keyword on one day.

When two charts disagree, compare the pages and dates behind them before you compare the numbers.

A triage checklist for a "drop"

Use this before you open a ticket, brief a client or change a page.

1. Check the date range. Did the drop start between 13 September and 4 October 2026? Those are the windows the sources flag.
2. Check the shape. Many keywords dropping on the same day and returning within one to three days points to data. Real ranking changes persist.
3. Check Search Console. Are clicks and impressions for the affected pages down on the same dates? If not, the tracker is the suspect. Remember that Search Console changed too: see my note on multi-country filtering and AI performance reporting.
4. Check by hand. Search a sample of five keywords from a normal browser, with a location that matches your market. This is a spot check, not a measurement; results are personalised.
5. Check the tool's status page. Sistrix, DataForSEO, SE Ranking and Advanced Web Ranking were all named by Roundtable as having published notices.
6. Check whether it was a spam update. Google's September 2026 spam update activity overlapped these dates (Roundtable's 18 September post links the blocking to Google's September ranking changes). If a real change happened, you want to know before you blame the tool. I covered that in the September spam update guide.
7. Only then change the page.

What to change in your reporting

The best fix is not buying a better tool. It is reporting things that survive noise.

Annotate your charts. When I write a weekly review, I add a note line for any week where a tool posted an incident, and I keep the rule from my weekly review routine: never explain a one-day movement.

Handling the client conversation

Clients see a red line and ask what you broke. A script that has worked for me:

• "Our tracker shows a drop on these dates. Several tools reported data problems in the same window, and one vendor has published a study of it."
• "Search Console shows [clicks/impressions] did not move/did move, so we are treating it as [data noise/a real change]."
• "We will not change pages this week. We will recheck on [date] and report a range."

This is also a good moment to ask what the number is for. If the answer is "to prove we are growing", point to leads, signups and Search Console clicks. If the answer is "to see AI answer visibility", read my notes on whether AI visibility tools are worth it, which has the same caution about vendor data.

Should you switch tools?

Probably not because of one incident. My criteria, in order:

1. Does the vendor publish incident notices with dates? (Sistrix, Morningscore and others did.)
2. Does it say how it handles truncated or suspicious pages? Morningscore says it began discarding truncated results on 30 September. Sistrix says it publishes only data that passes extra checks.
3. Can you see the stored page (the SERP snapshot) behind a number?
4. Does it track location at a precise level?
5. Does it separate "not found in the list" from "list was short"?

If a vendor cannot answer the last three, treat its daily numbers as indicative only.

What this means for AEO and AI visibility tracking

The same pressure applies to tools that sample AI answers: they depend on automated access, and the data they return is a sample, not a census. Keep that in mind when you read a share-of-voice score; my share-of-voice definition and formula post covers sample sizes. Sistrix says it now also measures citations in AI answers through its AI Check tool, which is a vendor feature, not an independent standard.

Frequently Asked Questions

Why is my rank tracker showing sudden drops?

Possibly because of data problems, not real losses. Morningscore says truncated results made missing sites look like drops from 24 September to early October 2026, and Nozzle reports poisoned results pages from late September. Check Search Console before assuming a real change.

What is a poisoned SERP?

Nozzle's term for a results page that looks normal but has most of a keyword's usual sites swapped out, then returns to normal on a later page. Nozzle found 8.0% of its sampled pages fit its definition since 13 September 2026. It is one vendor's definition and has not been confirmed by Google.

Has Google confirmed it is blocking rank trackers?

In the sources I read, Google has not made a statement. Barry Schwartz suspects the changes target AI bots and SerpApi more than SEO tools, and Nozzle attributes them to Google's response to AI scraping. Both are interpretations.

Is Search Console accurate while trackers are noisy?

It reflects real searches, but it has its own quirks. Average position blends queries, devices and AI Overview citations, and Sistrix says the 2025 results-per-query change affected impressions and average position. Use it for trends.

Which rank tracker is the most reliable right now?

I cannot say. Glenn Gabe's comparison on one site found Ahrefs tracked a drop and recovery best, but it is a single example. Test two tools against the same keyword sample and your own Search Console data.

Should I pause SEO changes because of this?

Pause reactive changes triggered only by tracker drops. Continue planned work. Verify drops with Search Console and a manual spot check first.

What is the goto redirect in Google results?

Search Engine Roundtable reported in July 2026 that result links appeared to route through a google.com/goto address before reaching the destination. Sistrix says it started doing this in June 2026 and that it resolves them for organic results.

What should I report instead of daily rankings?

Weekly medians with ranges, counts of keywords in the top 10, Search Console clicks, and a share-of-visibility metric. Annotate incidents on the chart.

Do manual searches give me the real ranking?

They show one personalised view. Use them as a control on a small sample, ideally from a normal browser with a location matching your market, and not as a measurement.

How do I explain this to a client?

Show the dates, say which vendors reported problems, and show what Search Console says. Commit to rechecking after a set date and avoid changing pages in response to one-day movements.

CTA

I am Younus Fardeen, an SEO and AEO strategist with 4+ years of marketing experience working with edtech and startup teams. If you need help building a reporting setup that survives noisy data, or want a second pair of eyes on a suspected drop, see my work and reach out through the contact form at https://younusfardeen.in.

Want help with search and AI visibility for your own site? Get in touch through the contact form.