80% Error Rate From Google Allintitle, Study Shows

The Google Allintitle search operator has fallen into serious disrepair. Regular users can see it clearly. But how broken is Allintitle? I did a research report for the Internet Marketing Gold community and the results were alarming.

This in-depth study shows that Google Allintitle has an 80% error rate. The search operator consistently returned flawed results, with both false positives (results that are not actual ‘all-in-title’ matches) and many missing true all-in-title (AIT) matches. 

Google Allintitle returned 8 results for all keywords…the true allintitle results counts (in the series) ranged from 0 up to 44.

Remember, the Google Allintitle search operator is meant to show users any pages in the Google index with all search term words (in any order) in the page title. Yet data inaccuracies are easy to observe in the wild.

First Signs of Trouble With Allintitle

The image above highlights one of the most commonly seen error types: results that are NOT a true allintitle match.

To regular users of the Google Allintitle search operator this is an all too common anomaly. What you can’t see is the true matches that should be there but are not. So, how common are these errors? 

The snapshot image below shows just a subset of the sample data keywords. For just this small series the Google ALLINTITLE search operator returned 8 results for all keywords. Yet, when SERP Sonar analyzed the same keywords the true allintitle results counts (“True PAT” below) ranged from 0 up to 44. 

Even greater discrepancies can be seen in the full study data set. NOTE: full data results visualizations are linked and viewable below.

Research Study Setup

This study started with a random selection of 150 keywords, with average search volume from 1-1000, which spanned many niches and various intent types. The final sample, after filtering out errors or missing data, was 138 keywords.

The tools used were a Chromium browser accessing US servers via VPN, and the SERP Sonar browser extension. The extension is what made it possible to easily assess page title matches, generate count totals, etc.

The Google Allintitle search operator consistently returns both false positives and fails to include results that should be there.

Allintitle Study Highlights

The Google Allintitle search operator consistently produces highly inaccurate data. The function returns both false positives and also fails to include results that should be there. 

Here is a breakdown of the results and takeaways:

• Google Allintitle usually includes false positives (results that are NOT true allintitle matches); 
• Google Allintitle often excludes true allintitle matches (results that SHOULD be included);
• On average the Google Allintitle search operator had only a 17.7% accuracy rate;
• More than 8 out of 10 results were NOT actual allintitle matches;
• Even when there are no true matches, Google Allintitle always returned results (false positives)*

*In the study sample set, the Google Allintitle results count (# of results listed at the top of the SERP) ranged from a low of 8 to a high of 113. In other words, none had 0 results, showing that Google will almost always return something (even if it’s inaccurate).

When using Google Allintitle, then, it would seem that every keyword had some allintitle matches. However, 45 (of 138) or 33% of keywords actually had zero (0) true allintitle match results, according to the SERP Sonar plugin.

…over 80% of sample set keywords included results that were not just mismatches (not allintitle matches) but were on completely unrelated topics.

Even if the the number of Google Allintitle results matched the true allintitle count provided by SERP Sonar (eg sample keyword “M” in the screenshot above) the results still included false positives (pages with titles that were not true allintitle). And these mismatches were not just relevant, on-topic pages with titles that were not allintitle compliant. In many cases the titles and pages are completely off topic and even spammy.

Hey, you got somethin’ to say? Leave a comment below! Is there a point in this post you disagree with, want to add to or totally love? Then let everyone know! Thanks!

Mismatched And Meaningless Results

Sometimes the results are NOT true allintitle but at least they broadly align with the topic and intent. But often the results returned are complete nonsense or spam pages (see image below). 

Based on visual inspection, over 80% of sample set keywords included results that were not just mismatches (not allintitle matches) but were on completely unrelated topics or spam.

This adds to the already frustrating data accuracy issues and makes the Google Allintitle search operator not just less useful but also a potential liability.

Google Allintitle vs True Allintitle

The results above highlight what is wrong with the search operator, but what does “right” look like?

The SERP Sonar extension finds allintitle matches among the top ~100 search results from a regular Google search. In the plugin this is referred to as Page All-in-Title (PAT). You will see below why I also refer to it as “true allintitle”.

…if Google Allintitle queries 200-300 pages and SERP Sonar scans just the top 100 pages then SERP Sonar should under-count results, right?

The general understanding is that a regular search on Google queries a subset of Google’s full search index or primary index. It has been suggested that this is an ever-changing short list of 200-300 pages, per keyword or topic. Of course, this is conjecture but it would be unfeasible to search the entire index of billions of pages every time any keyword is searched. So, it is likely a small subset of the whole at any given moment. And if that is the case for human searchers it’s doubly so for the search operators (which Google, clearly, care little about).

Still, if the Google Allintitle search operator queries only 200-300 records and SERP Sonar scans just the top 100 pages then we should expect the SERP Sonar results to under-count, on average. Right?

In fact, the number of true allintitle matches was consistently higher – 6X higher on average – than the Google Allintitle results count.

• True allintitle (PAT) count averaged over 6x higher than the Google Allintitle match count;
• In many cases the true allintitle was significantly higher (10x – 24x);
• Only 20 of the 138 cases (~14.5%) had PAT # equal to Google Allintitle match result #;
• But these keywords were all either 0 or 1 match result (ie extremely limited page supply);
• The more true allintitle (PAT) results, the less accurate the Google Allintitle results become.

full data set visualizations can be found here:

Google vs SERP Sonar Allintitle Result Count Comparison

Google Allintitle With Accuracy Rate vs True Allintitle Counts

Summary & Final Thoughts

Doing allintitle analysis can be extremely useful, especially in long-tail keyword research. The SERP allintitle results are a useful barometer of which and how many pages are directly targeting a search query. It can also help sanity check if Google ‘understands’ the keyword the way you do. 

SERP Sonar offers an accurate and free alternative to the Google Allintitle search operator.

Unfortunately, the current Google Allintitle search operator is broken and consistently returns highly inaccurate results. Google Allintitle both returns inaccurate false positives and also excludes true allintitle matches.

In time, Google might even sunset this search operator. Fortunately, SERP Sonar offers an accurate and free alternative. It even makes it easy to validate traditional Google Allintitle search operator results (do an ALLINTITLE: search and SERP Sonar will tell you how many results are true matches). 

Download SERP Sonar and give it a try, and check out all the other useful features too. Let us know what you discover in the depths of the SERP!