For years I have steered clear of Google Business Reviews after I realised that Google Business has no systems to deal with these. If a client has a fake or harmful review posted on their Business Profile, I head straight to Google Legal and remove the review – thank God for legal sanity, sorry US businesses, this does not apply to you.
Last night in the Google Business Community, I came across a report of a business incentivising reviews. Diving into this showed me that the reality of Google Business Fake Review Detection does not exist.
These were customers (incentivised), so technically not FAKE any form of a basic algorithm should have flagged and filtered these reviews. Instead, they have been running unfiltered and undetected since 2017.
The Incentive for Reviews
This incentive is posted at the counter in the restaurant / take away.
Note:
Choose any keyword from these three:
- Best Kebab
- Best HSP
- Best Halal Burger
The Reviews
I have removed all other reviews.
Detecting Fake Reviews
Now I am not a programmer but in my mind an algorithm should look for a pattern and repeated use of the same words.
A little bit of digging and I found some published papers.
“A Survey of Fake Review Detection Techniques” by Jie Chen, Honglei Zhuang, and Xueqi Cheng. This paper surveys a variety of techniques for detecting fake reviews, including the use of repeated words.
“Fake Review Detection: A Survey” by Rui Wang, Zhiguo Gong, and Zhiqiang Sun. This paper also surveys a variety of techniques for detecting fake reviews, including the use of repeated words.
“Detecting Fake Reviews: A Natural Language Processing Approach” by Arvind Neelakantan, Mohit Bansal, and Andrew McCallum. This paper describes a system for detecting fake reviews that uses natural language processing techniques, including the use of repeated words.
Word n-grams: Word n-grams are sequences of n words. For example, the word n-gram “best halal burger” consists of three words. The frequency of word n-grams can also be a sign of a fake review. For example, if a particular word n-gram appears much more frequently in a review than it does in a large corpus of real reviews, this could be a sign that the review is fake.
Giving the Benefit of the Doubt
As I have already mentioned these were actual customers, they were in store and they used the QR code to leave the review.
So the algo was 100% certain they were there BUT any fake review detection should have flagged and filtered all of these reviews!
Non-Sequitur
Having worked in the industry for years, I can’t help stuff like this.
Current wholesale price of buffalo chicken wings is $12 per kilogram. Two wings, which would weigh approximately 100 grams, would cost approximately $1.20.
It would take approximately 10 minutes to cook and prepare two buffalo chicken wings. If you pay your staff $20 per hour, this would add an additional $1 to the cost of the wings.
Overhead costs would add another $0.50 to the cost of the wings.
So, the total cost of two cooked buffalo chicken wings in barbecue sauce would likely be between $2.70 and $3.70.
Taking the average at $3.20
The reviews above would have cost them around $705.60
Online Ownership does not take on review issues but we do offer Google Business Consulting, feel free to contact us to discuss the Legal avenue of removing reviews if you are in a country that hold publishers accountable for what they publish.