Text Analysis

The good, bad, and the f#ugly 

Let's talk

Sentiment Detection

Do you know what people are saying about your brand? These days, it’s no longer enough to rely on manual analysis to gauge customer sentiment. Leading organisations are using natural language processing (NLP) to detect and interpret customer sentiment from a range of data sources to improve products and services.  

Unlike other tools, Analytics8’s sentiment detection tools analyse multiple paragraphs with ease. They incorporate negation (i.e. “… don’t like…”) to accurately detect sentiment, and incorporate domain knowledge to detect words that have an unusual, industry-specific meaning. For example, the word “spoiler” typically has a negative meaning, but in an automotive context it refers to aerodynamic accessories (e.g. “I like the spoiler design”).  

Hashtags, emojis, & TLAs 

A hashtag, emoji or three-letter acronym (TLA) can significantly change the meaning of a social media update. So how can you be confident that you are interpreting customer sentiment accurately? Our advanced text analysis solutions automatically detect sentiment and infer meaning, even for emojis, hashtags, TLAs and profanity. 

Spelling mistakes and capitalisation

Most out-of-box text analysis solutions cannot analyse text that contains spelling mistakes or grammatical inconsistencies like persistent capitalisation (e.g. “…don’t EVER listen…”). Analytics8’s solutions, on the other hand, recognise spelling mistakes and other errors to ensure that all customer comments are considered.

Make data actionable

Can your organisation match customer feedback to the employee who delivered the service? Through common data manipulation analysis, we can matches descriptions about your organisation to existing organisational data. For example, if an employee is mentioned by name or department, customer feedback is linked to that employee identification.This is particularly useful when performing text analysis on sales processes or customer touch-points.  

Get smart about domain knowledge

Domain knowledge is the difference between an average text analysis solution and an expert solution that detects intricate details in the customer experiences. Many text analysis solution providers aren’t able to integrate industry metadata into their solutions.  

We build custom dictionaries to provide domain-specific knowledge. By analysing synonyms and association, we identify words or groups of words and their probability of use. Our tailored text analysis projects are enriched with thesaurus, custom dictionaries, and additional machine learning capability that improve the performance of out-of-box solutions supplied by major software and platform vendors. We also build text analysis solutions using R open source, either cloud hosted or on-premise.  

Quickly decipher text meaning

There is little business value in analysing text if you cannot decipher its meaning. This is why we go beyond data preparation and transformation to provide text analysis solutions that embed predictive and classification analysis. The result is an improved ability to correctly categorise and summarise surveys and customer feedback.  

Exceptions and outlier reports

Natural language processing is clever, but it sometimes makes mistakes. We catch these errors by allowing for manual supervision or review of text as required. This is particularly useful for occasions when sentiment detection, categorisation or text interpretation of text has a higher probability of error (i.e. incorrectly summarising the content and messaging of the text). 

Let’s talk.



Sydney: +61 2 9299 4430

Melbourne: +61 3 8605 4880

  • This field is for validation purposes and should be left unchanged.