AI vs trolls: neuro-audit of social networks, SERM systems and recognition of fakes

How systems find what they write about you

Negative reviews or comments are often left on the Internet:

There are always those who don't like something.It's important to identify unfavorable opinions.

Today there are several approaches to finding negativity on the web:

  • If a person doesn't like the service, then they will most likely want toAccording to a study by Sprout, Empower & Elevate, six out of ten consumers turn to brands on social media because they already have a good experienceInteraction.
  • Work with search results.With this tool, you need to be able to work correctly, carefully selecting keywords. For example, Google Ads can handle this - an AI system works with all queries in a search engine and, based on this data, gives statistics on the most popular words and phrases associated with a particular brand, person or organization. After that, you need to compile a list of keywords, enter them into different search engines, and from there select forums, marketplaces and other sites with reviews, including negative ones. The process can take a long time, and in order not to miss anything, you should use automated monitoring systems. One such service is Google Alerts. The system sends notifications on changes in search results. This approach is convenient for both users and companies, which, in turn, need to maintain a high rating. According to the American research firm FeedVisor, users are most likely to choose sellers on Amazon, with more than 90% of positive reviews. This is important because the user builds their brand identity based on the reviews and recommendations of other people.

How the Internet deals with negativity

People are 16% more likely to choose the communication channelwhich the company does not consider essential for customer support. This gap between what consumers expect and what the business has to offer needs to be bridged.

According to a ReviewTrackers survey, 53% of consumersexpect brands to respond to their reviews. At the same time, reactions to comments can be different: they can be neutral, positive, negative, or even slanderous. Any interaction is individual, but there are general recommendations. For example, if the review is bad or neutral, you need to first find out the reasons for the poor quality of service.

Sometimes it happens that to solve the problem peacefullythe accumulation of negative reviews does not work. This means that you need to use a Search Engine Reputation Management (SERM) strategy. This is a set of actions that move negative reviews or posts from the first SERP and replace them with positive information. For example, if Google selects negative review in response to a query, it needs to be optimized to offer the search engine a better article or review for the same keyword so that the information replaces the one that needs to be “hidden”.

Besides SERMs, there are other methods of working withnegative. Review management systems are another way to track, redirect, and respond to reviews when they are uploaded to multiple sites and forums. This type of negativity fighting gives you the opportunity to regularly post positive reviews on various online platforms by redirecting negative ones to a customer service representative.

Feedback management softwareCommunication is a system of processes that allows businesses and non-profit organizations to centrally manage and conduct surveys, ensuring the dissemination of information. Ideally, it is a web-based tool or portal designed to collect, disseminate and analyze feedback data so that it can be turned into strategic decisions for future developments. In addition to this, the software also allows you to grant roles and powers to users of various levels.

Another way to deal with negativity is to motivateusers to leave positive comments. For example, offer a discount on goods or conduct an SMS survey asking you to rate the service. Rewarding users can be a motivating factor as it often takes a little nudge even for the most satisfied people to leave a review. In such cases, any pressure to obtain an estimate should be strictly avoided, otherwise the confidence of potential buyers could be undermined.

How to predict what users won't like

To implement the idea, you need to establish feedback:it can be a separate site or, for example, a bot in Telegram. McDonald’s found an interesting solution: each restaurant visitor receives a unique code on the check, which can be entered on a special website to send a review. All comments are sent to the company's server, where they undergo primary processing and ultimately end up in the review storage system.

Another part of the work of anticipating negative reviews is a high speed of response.After posting an unfavorable comment, a person expects to receive feedback as quickly as possible, so it is important to respond promptly.

To do this, you can use, for example,Telegram bot. Basically, bots are special accounts that don't require an additional phone number. Messages, commands, and requests sent by users are passed to the software running on the side of the client application. Next, the intermediate Telegram server handles all encryption and communication with the API while the user communicates with the bot through a simple HTTPS interface that offers a simplified version of the Telegram API.

How to deal with pre-existing negativity

SERM is a reputation management system insearch results. Before you start implementing SERMs into your workflow, you need to do a little research. An SEO specialist needs to collect a semantic core (words and phrases describing a brand, product or service - "high-tech") for search queries. For example, the phrase “Web Design” will be high frequency, “Web Design Company” - medium frequency, and “Best Web Design Company in Moscow” - low frequency. The more popular a word or phrase is, the more competition the company faces. Links in search results are then analyzed for site quality metrics and grouped by content type and source to identify the most common negative links.

To work with the existing negativity you need on timeidentify it by monitoring search results. For example, Youscan.io or Brand Analytics can help track reviews on social media. Automated systems monitor every mention of a brand on the web, so it is important to respond to them: thank users for the feedback or do everything possible to change a negative review to a positive one. If a person left a slanderous comment, you can bring him to justice in court or try to figure it out on your own.

For example, Yandex supports brands and, when requested, can remove outdated or inaccurate information from search results on the "right to be forgotten".

A much faster and less costly method ischeating positive reviews, but this method also has its drawbacks. If this option is abused, the user will easily suspect that something is wrong and send a complaint. Identifying fake reviews is something self-learning AI is very good at doing. Such systems use language processing techniques to detect unusual text patterns, writing style, and formatting. For example, researchers at the University of Chicago in 2017 developed a machine learning system that was a deep neural network that relied on a dataset of 3 million real Yelp restaurant reviews.

How offline and online reputation are interconnected

Almost 91% of adults have a mobile phoneis on hand 24 hours a day, seven days a week, and 88% of consumers say they trust online reviews as much as they trust personal recommendations. Companies in such conditions already find it difficult to rely only on the interpersonal way.

Online reputation in the digital agetightly connected with offline. Because of this shift in consumer behavior, executives and business owners need to make sure their online presence is tangible. You can check this by keywords. For example, Wordstat allows you to view statistics in search engines. The system analyzes all user requests and collects information about them.

Omnichannel communication is the mostan effective strategy for hassle-free sales and service. Strategy is important because over 90% of consumers use multiple sources of information when making a purchase decision.

Many contact centers want to improveefficiency and improve the quality of service, therefore turn to the use of AI and bots. The use of AI technology, machine learning and natural language processing is helping companies reduce the time it takes to solve emerging problems.

In addition, the omnichannel approach includes collecting anddata exchange between various online platforms and sales channels. The impact of this practice is so significant that over 75% of consumers now expect constant engagement across all channels and platforms.

See also:

The first accurate map of the world was created. What's wrong with everyone else?

Ingenuity helicopter successfully takes off on Mars

NASA told how they will deliver samples of Mars to Earth