We took a bunch of customer questions, RFI's and RFQ's and combined them into a single Q&A page. 

Drill into the different themes through these links.

On Etuma and feedback analysis business

On analysis technology

On feedback channels

On Industry specific Codeframes

On insight distribution

On API's

On contact center ticket and feedback analysis

On multi-language analysis

On platform in general

On setup and maintenance

 


Want to learn about how to choose the right method and
tool for customer experience text analysis?

CX PROFESSIONALS GUIDE TO TEXT ANALYSIS

TERMINOLOGY

Feedback text analysis is a new industry and that’s why there isn’t a well established terminology (that everybody agrees upon). Understanding these four concepts makes the reading of this document easier.

SIGNAL: We call customer's and employee’s open-ended text comments Signals. Some people call them open-ended comments, text feedback, unstructured feedback or verbatims.

TOPIC: Topics are what customers talk about. They try to instill the meaning of a customer’s or employee’s intention. Topics are issues, concerns and ideas that customers express in their text comments. Etuma Topics are industry specific or tuned to customer’s specification. Some people call these categories or classes.

CODEFRAME: We call a set of industry or customer specific Topics a Codeframe or Topic Codeframe. Some people call this taxonomy, ontology, categorization system, or classification system.

SENTIMENT: Sentiment is how customers feel about different aspects of your company's operations. In Etuma’s solution Sentiment is measured at Topic level. Etuma gives a negative sentence in Signal a score of -1, neutral is 0 and positive +1. Customer’s and employee’s wishes are given a score of -0,5.

On Etuma and feedback analysis business 

[fa icon="plus-square"] What kind of company is Etuma?
Etuma is a computational linguistics application provider. More than half of Etuma's staff have a degree in computational linguistics.
[fa icon="plus-square"] What is Etuma's service called?
Etuma's text feedback analysis service is called Etuma Feedback Categorizer but most people just call it Etuma.
[fa icon="plus-square"] What is Etuma's business model?
Etuma's service is provided as a monthly subscription service. Minimum contract term is one year.
[fa icon="plus-square"] What kind of business domains does Etuma service?
Customer Experience Management, Contact centers, Competitor analysis (using social media), Market Intelligence, Sentiment Analysis, Social Media Monitoring, Voice Of The Customer, Employee Engagement, Chat analysis.
[fa icon="plus-square"] What kind of verticals does Etuma support?
Etuma has an existing multi-language codeframe for example to following industries: Consumer products, Consumer services, Education, Financial Services, Municipalities, Healthcare, Electronic devices, Industrial products, Insurance, Media, Hospitality, Hotels, Charter Travel, Airlines, Online gambling, Video games, Retail, eCommerce, Telecommunications, Logistics, Local transport, Car industry.
[fa icon="plus-square"] What are the main aspects/features/capabilities that separate you from your competition?
Multi-language high quality verbatim analysis. Etuma's NLP technology gives unique advantages: continuous articifial intelligence driven learning (=improves all the time), ready day one, tools for fast tuning and optimization, topic specific sentiment, topic specific sentence extraction. No vendor lock-in to a survey or visualization vendor. Can work with any 3rd party solution.
[fa icon="plus-square"] How do I know that your service really works?
First, we provide a free demo with your data. Second, Etuma has extremely low churn: Etuma has lost only few enterprise customer during the five years of service. 
[fa icon="plus-square"] Where does Etuma name come from?
It is a bit of play with words. I mean, we are a natural language processing company. It comes from a Finnish word Etumatka, which means being ahead of other people. We believe that we give our customers an advantage because they gain a deeper understanding of their customers. 

On analysis technology

This explains HOW Etuma analyzes Signals.

[fa icon="plus-square"] How accurate is the analysis?
Day one the analysis accuracy is above 80% and can get tot to close to 90% after six month of service. Sentiment analysis is typically around 80%.
[fa icon="plus-square"] What are your text analysis functions?
Syntactic parsing, word count, tokenization, end of sentence detection, part of speech (POS) identifications, roots, word stemming, sequence of stems (lemmas), roles, spelling correction, synonyms, semantic topic and sentiment detection.
[fa icon="plus-square"] What kind of comment level text analysis functionality do you have?
Concept extraction, coreference solution, entity extraction, lexical chains, word sequences, relationship mapping, topic level sentiment detection, word nets, standard codeframes (industry specific).
[fa icon="plus-square"] What kind of lexicons and dictionaries do you have?
Massive number of brand names (e.g. mobile phone brands and models). But at the beginning of each new customer relationship, Etuma recommends that the customer sends Etuma all their relevant brand names so that Etuma can ensure that they are captured.
[fa icon="plus-square"] How does your sentiment analysis work in reporting?
Etuma detects the sentiment taking into account the intricasies of feedback text. Etuma detects multiple topics per comment. Each topic is given a sentiment score. The whole comment sentiment is the average of all sentiment scores. Negative (-1) topic mention ("Cashier didn't smile" -> SERVICE ATTITUDE - Negative). -0.5 Sentiment for a wish or something missing. ( I wish you had XXX product in your SELECTION -> SELECTION -0.5). Neutral sentiment when there is no clear opinion. And positive (+1) when the comment is positive. 
[fa icon="plus-square"] How does your sentiment analysis work technically?
Etuma sentiment analysis tries to imitate the way human brain works. It has a large number of grammatical and language specific sentiment rules that calculate the sentiment using artificial intelligence. Etuma's technology has been developed during the past twenty years. It has close to 10 million lines of code. 
[fa icon="plus-square"] Can you analyze emoticons?
Yes, over 500.
[fa icon="plus-square"] Is the underlying technology to analyse the verbatim comments developed in house or is a 3rd party API/technology used?
Developed in-house. Etuma uses few 3rd party libraries but that company is owned by the same people as Etuma.
[fa icon="plus-square"] Can you analyze voice?
No, First the voice needs to be converted to text. Etuma recommends using voice to text solutions like Nuance. Once the voice is converted to text, Etuma can analyze it.
[fa icon="plus-square"] How many sub-level of granularity are available to define root-causes of detraction/ promotion?
Up to four layers: topic groups, topics, keywords, and the actual topic related sentence that the customer wrote.

On feedback channels

This explains what kind of text feedback you can bring into Etuma.

[fa icon="plus-square"] What kind of survey channels and formats do you support?
Any customer feedback channel as long as there is text in it. These channels include emails, webforms, contact center tickets, NPS/CES/C-Sat text comments, social media comments, forum comments, etc. Basically any discussion between a customer and a company or about a company between customers.
[fa icon="plus-square"] Does Etuma run surveys?
No. Etuma focuses on analyzing customer feedback. Etuma has connectors to many survey tools and solutions. If Etuma doesn't have an existing connector, a new one can be created.
[fa icon="plus-square"] Does your platform crawl social media?
Yes, Etuma can crawl public (company) Facebook sites and Twitter handles and keywords.

On industry-specific Codeframes

This explains the way Etuma categorizes Signals.

[fa icon="plus-square"] Can Etuma detect weak signals
Etuma's industry specific Codeframes have, besides industry specific topics. also the 'whole world'. What this means is that Etuma detects, besides all relevant topics in your industry, also unexpected new emerging topics (trends).
[fa icon="plus-square"] How do you detect 'trending' topics?
Etuma counts the Topic mentions. If, for example, people are talking more about WEBSITE then you will immediately notice that because the comment count is higher. There might be also a change in the sentiment.
[fa icon="plus-square"] What kind of lexicons and dictionaries do you have?
Massive number of brand names (e.g. mobile phone brands and models). But at the beginning of each new customer relationship, Etuma recommends that the customer sends Etuma all their relevant brand names so that Etuma can ensure that they are captured.
[fa icon="plus-square"] What is the method of categorization used by your platform? Are categories determined organically or can a predefined list of categories and levels be used?
Both ways are supported. Etuma has a ready code frame for over 20 industries. These are ready day one: just plug in the feedback channel and the analysis system starts working (in close to real-time). Etuma provides easy to use tools, with which the customer can tune the analysis for their needs.
[fa icon="plus-square"] Does your platform conduct sentiment analysis as well as categorization functionality?
Yes to both. Etuma categorizes feedback into industry specific topics and viewpoints and sets the sentiment for each categorization level. The whole comment sentiment is calculated as an average of all the mentioned categories in a comment. Etuma also detects whether there are just positive, negative or both positive and negative categories mentioned in a comment.
[fa icon="plus-square"] How can a codeframe be tuned?
Analysis Codeframe can be tuned to any industry, company and brand via a custom project. Usually this is not necessary: most Etuma customers use Etuma's 'standard' industry codeframes.
[fa icon="plus-square"] Can users create their own custom (keyword to topic mapping) rules?
Yes, Etuma has easy-to-use tools that a customer can use to tune the anlaysis. But often better Codeframe is created and maintained when Etuma's computational linguistics experts tune the system. 
[fa icon="plus-square"] What kind of codeframes do you have?
Etuma offers over 20 industry specific codeframes. It is important that the codeframes are industry specific because the topics are different, keyword mapping to topics is different and even sentiment analysis often requires industry specific tuning.
[fa icon="plus-square"] Can you import and tune codeframes?
Yes, but this requires a project. Etuma can provide tools for you to tune the codeframe.

On insight distribution

This explains how you share the analysis results with the rest of your organization. 

[fa icon="plus-square"] Does your platform allow to add metadata to generate new filter parameters and give more contextual information to the existing information?
Yes, you can use any metadata in the survey or customer record to filter into the text analysis results.
[fa icon="plus-square"] How does Etuma support 3rd party analytics and visualization tools?
Etuma has a MySQL (AWS) database interface to e.g. SPSS, Excel, Tableau, Qlik. Customer can also use Etuma API to transfer the analysis results into their database.
[fa icon="plus-square"] Can Etuma automatically distribute verbatim to the corresponding department/ right ‘owners’ based on the root-causes of detraction/ promotion identified, further to another touchpoint?
Yes, Etuma customer can use EtumaReports to distribute customer comments in real-time to stakeholders. Customer can also do the comment routing in the platform of their selection based on the text analysis results. 
[fa icon="plus-square"] What are the standard "out of the box" reports that come with your platform?
EtumaReports is a browser-based tool in which a key user can create dynamic reports for customer experience stakeholders. Once a report is created then it is automatically updated when more feedback flows in.
[fa icon="plus-square"] Can you give examples of the reports?
E.g. If a person is reponsible for staff issues, they will get a report that lists only staff related categories. If somebody is responsible for a specific car part, then they will get all comments (or just sentences) related to that car part. Also process, invoicing, brand/marketing reports are available. Basically any report that has to do with text analysis results. There are too many to mention here.
[fa icon="plus-square"] What reporting or filtering data is available specifically for text comment analysis?
Any piece of information (meta data) and text analysis results (topic group, topic, keyword, sentiment) can be used to filter the analysis results and create a customized report out of it. Customer can bring into Etuma any kind of metadata on the customer comment via the API or as batch file upload CRM data, demographics, purchase behavior, location
[fa icon="plus-square"] Do you provide a reporting or dashboarding environment?
Etuma has its own reporting platform called EtumaReports. EtumaReports requires a browser. In it the key user defines the report by setting the filtering rules and then sharing the link to a report via email. For large scale insights distribution Etuma recommends using 3rd party visualization platforms (Qlik, Tableau, PowerBI etc.).
[fa icon="plus-square"] What parts of the reporting platform are flexible and which parts are standard and cannot change?
EtumaReports has a fixed easy to use structure that is built around the verbatim analysis dimensions (topic and sentiment). It has only few layouts which are specifically designed to visualize text analysis results and enable easy and fast drill-down to root-cause (topic specific sentences). Etuma has database access to other more advanced tools like Tableau etc, which are flexible and fully customizable.
[fa icon="plus-square"] Please explain the different types of KPIs/ metrics you are able to improve based on your own experience and what is the expected ROI for VSD for each amount of improvement.
Etuma is good at detecting the NPS/CES/C-sat drivers. Etuma helps you understand the correlation between e.g. NPS score and different service or products aspects. Etuma also helps you prioritize development efforts because you have an understanding of the order of importance of different product or service aspects.
[fa icon="plus-square"] Does your platform offer a place to conduct root-cause analysis?
Yes. In language analysis the meta data created from the open-text comment (viewpoints, categories, sentiment) are used to detect that something is happening either as a function of category volume or sentiment. The open-text root-cause analysis is done by drilling down from this pattern to the actual relevant text (not the whole comment but the actual sentence in which that category or viewpoint is mentioned). It is crucial that what ever text analysis vendor you choose, they can extract the category specific sentence from the text. This saves a lot of work and results in faster root cause detection and issue resolution.

API's

Introduction to Etuma's web interfaces.

[fa icon="plus-square"] What kind of text input types do you support?
Batch files, streaming (close to real time) Facebook, Twitter, Forums, Contact Centers, Feedback forms, Chat logs, Web forms, Email, Text messages, Transactional surveys (NPS and CES etc.).
[fa icon="plus-square"] What input text formats do you support?
CSV, Excel, XML, JSON, feedback forms, email (text, html, rich text).
[fa icon="plus-square"] What kind of connectors do you have to 3rd party systems?
Database (ODBC/JDBC), IMAP server crawler, RSS feed crawler, Social media listening (Facebook and Twitter), (S)FTP, existing connectors to dozens of CEM, CRM and BI platforms.
[fa icon="plus-square"] What % of the product functionality can be customized using APIs for OEMing, embedding, white labeling, etc.?
100%
[fa icon="plus-square"] What % of the product functionality can be invoked via external APIs?
100%
[fa icon="plus-square"] What are your output formats?
CSV, JSON, RDF, XML

On contact center ticket and feedback analysis

How Etuma can speed up problem cluster identification and reduce cost by automatically categorizing incoming emails and 

[fa icon="plus-square"] Can Etuma analyze and categorize contact center tickets?
Yes, Etuma has an automatic ticket categorization solution designed for contact centers. Service runs as a so called embedded analytics solution via an API (e.g. SAP cctr, Zendesk, Oracle, Salesforce).
[fa icon="plus-square"] Can Etuma analyze chats?
Yes, Etuma can analyze chat logs. For text analysis purpose it is crucial that the chat service tool can separate the agent's and customers comments from each other.

On multi-language analysis

Etuma is developed from day one to handle multiple languages.

[fa icon="plus-square"] What are the languages that are supported natively by your platform?
Our analysis languages are English, French, Italian, German, Spanish, Dutch, Danish, Norwegian, Swedish and Finnish. Analysis is done in source language (syntactic, morphological and semantic libraries for each language). Results can be presented in any of the ten languages (analysis results are "mapped" accross languages).
[fa icon="plus-square"] How would we be able to handle languages that are not natively supported by your platform?
Etuma uses Google Translate for other than natively supported languages. Google translate quality has increased substantically during the last year. If customer wants, Etuma can provide each customer comment and sentence in both the source and another language, so that managers in the central location can hear the end-customers unedited voice.
[fa icon="plus-square"] Can you automatically detect a language?
Yes.
[fa icon="plus-square"] Is your Platform able to consolidate the verbatim allocation results both in the local language and in English?
Yes, Etuma displays the analysis results in any of the ten languages. We recommend that one company uses one language in reporting.

On platform in general

Where Etuma servers are and how scalable is Etuma's service.

[fa icon="plus-square"] Can I run Etuma in my own servers?
No, Etuma runs only in Amazon Web Services in Ireland, EU
[fa icon="plus-square"] Can the product architecture scale up (add CPUs, RAM) and out (add servers)?
Yes. Etuma runs in Amazon web services and scales well.
[fa icon="plus-square"] Is the product architecture multithreaded?
Yes.

On setup and maintenance

What does it take to implement and maintain Etuma.

[fa icon="plus-square"] Does Etuma require a setup project?
Yes, but the set-up is very fast. The feedback channel(s) are connected via the API. Etuma codeframes are available off-the shelf. You can start enjoying the advanced feedback analysis results immediately. Etuma's computational linguistics staff tunes the system in the background.
[fa icon="plus-square"] What is the setup process for a new customer in your platform?
Service is live immediately after the web interface (API) is connected because Etuma has an existing industry specific Codeframe. Implementing the API takes usually less than a day. Codeframe is tuned in the background by Ph.d. Level computational linguistics experts. Codeframe is maintained for relevance and accuravy for the duration of the contract by tuning Topics, updating brands, mapping keywords to Topics and optimizing sentiment analysis.
[fa icon="plus-square"] Based on your experience, what is the average time to implementation of your platform?
It usually takes about a one working day for the customer to see the first analysis results. After this Etuma staff starts using customer's feedback, AI and machine learning to tune the analysis system to better meet the customers needs.
[fa icon="plus-square"] Based on your experience, what have been the biggest pitfalls and causes for delay during the implementation of your platform?
Getting an automatic downlink API running (if not standard REST possible) can take a few days. But before then, the analysis results can be manually uploaded in csv or excel formats.
[fa icon="plus-square"] Is the text analytics system adjustable (i.e. Platform continuously learns/ improves from glossary that will be adjusted and extended over time)? Does your Platform require a first sample of verbatims, or dictionary and word association, for its initial set-up?
Yes, system is built around artificial intelligence and machine learning technologies. No, Etuma doesn't need any learning material to do the first analysis but getting a corpus of historical feedback data will speed-up the analysis tuning. Analysis is improved in the background all the time by using machine learning, AI and Etuma linguistics experts to improve the accuracy and the relevancy of the analysis results.
[fa icon="plus-square"] Does the product use supervised (trained, human assisted), unsupervised (untrained, human free) or both techniques?
Both. The service that customers use is unsupervised but the analysis model is continuosly improved by machine learning and Etuma computational linguistics experts.
[fa icon="plus-square"] If categorization of verbatims evolve over time, can your Platform manage old categorization (= keeping the records of previous categorization)?
Yes, but customer needs to ask for us to store the old analysis codeframe.