Traditional market research tries to “recreate” consumer truth by asking them questions. The number of questions, how they are formulated, the questionnaire's length, the quality of the recruitment – all these factors introduce bias and undermine the quality of the information gathered.
MASSIVE, on the other hand, directly gathers a complete census of spontaneous, influential consumer reviews based on first-hand, real-life experiences. Nothing is re-created in a market research "Petri dish"; the insights we provide are pure consumer truth.

Yes, we can. Fake reviews have a very specific characteristic - they are usually very generic and never point out specific aspects of the item described. Apart form the usual methods to spot fake reviews (similarity, length, etc.) our Artificial Intelligence can also identify generic, pointless review that doesn't say anything specific and pertinent about the reviewed subject. We have created a specific metric for this - the Review Quality Ratio - that clearly indicates what reviews are good fake candidates.  

Sampling live digital data (like consumer reviews, social media posts, etc.) is mathematically impossible. Yes: I-M-P-O-S-S-I-B-L-E. Statistically valid sampling is only possible when the size and the composition (a.k.a. "stratification") of the data population is known (like with the US population, or a production batch, for example). But digital data keep growing and changing by the second, thus any data extraction - however big it is! - will result in a so-called Accidental Sampling, which is statistically meaningless. And you do not want to base your business actions on statistically meaningless information, do you?
This is why the vast majority of Voice-Of-The-Customer (VoC) solutions available on the market (Like Social Media Listening or Brand Reputation tools) are flawed at the core. None of these tools is taking a continuous census of live digital data, they are just using Accidental Sampling for their convenience.
MASSIVE, on the contrary, collects and continuously updates a complete census of the consumer reviews (and other relevant info) of any specific category. This means that every data point in our analysis is a perfect, unbiased representation of reality, making our business insights much, much more reliable and statistically valid than traditional market research.

Sentiment Analyisis is but a small part of the CRA task. Extracting insightful opinions from reviews requires an accurate identification of what is the topic the consumer is describing in a specific sentence. And a review usually speaks about many topics contemporarily; one about a TV could be describing the remote control, the resolution, the speakers, etc.. That's why calculating the general sentiment of a review doesn't make sense at all - a consumer might have opposing views about different aspects oof the same product/service. And it would be utterly useless too - the review rating is accurately capturing the general sentiment of a review anyway.
So transforming millions of reviews into actionable business insights requires a much more complex task than just sentiment analysis, one that can only taken care of by a very solid Artificial Intelligence process.

All our data are arranged around the specific category market hierarchy, according to the following structure:
Holding Company > Brand > SubBrand > Line > Item (SKU, Venue, etc.)
Our data are aggregated from the bottom up, i.e. from the item up to the Holding Company. This means that the analysis carried at the Brand Level of Detail (LoD) will always include ALL the items sold under that specific Brand.
We call this data aggregation process "bucketing", and it requires extreme care. Different retailers give different names to the same product - sometimes the same product can be named in ten different ways in the same retailer. Our AI technology makes sure that all the reviews belonging to a specific item are grouped together, regardless of how it has been named, so that you can get an accurate competitive picture.

Yes, but very marginally. Our Artificial Intelligence is extremely sophisticated and powerful, and it's been designed to require minimal supervision, regardless the language. This is because at MASSIVE we are very conscious that human intervention introduces bias, so our human analysts only come into the picture for Quality Assurance tasks, checking if the AI is performing the job properly.

Our clients have direct online access to a variety of more than 10 dashboards and 40 data views, expressly created after their needs. We do not force our powerful data in a one-size-fits-all dashboard like the majority of VoC companies do. We proud ourselves to adapt our data delivery continuously and quick, as our clients keep coming with fresh new ideas about how to use  our powerful datasets.
Our clients can also request special analysis delivered in the for of Insights Reports, crafted by our Market Insight Analysts.
We also provide data by API connection, but as many of our proprietary measures (e.g. Reviews Quality Ratio, Net Opinion Score, etc.) are non-additive, API data connection cannot exploit our data to their full potential.

No. We have designed our proprietary AI technology with true Big Data in mind. As explained, any reliable analysis based on live digital data cannot take advantage of sampling. For trustworthy Consumer Reviews Analytics, we are obliged to take a continuous, growing census of all the review written in any category. This is why, from the very beginning, we designed our A.I. with humongous quantities of consumer reviews in mind, and we succeeded. The more data, the merrier.

There's no limit - it depends on how many are needed to properly analyze a category, and it can vary based on specific clients' needs. We develop our own specific scraping software for each source we monitor for quality purposes and to ensure that we respect its T&C and robot.txt requirements.

Brand Reputation tools are designed with just one task in mind - monitoring one Brand's reviews to manage potential crisis, and have an initial, superficial understanding about one Brand consumers' likes and dislikes.
They are only good for crisis management - i.e. managing angry customers or showing social interaction. But when it comes to doing competitive analysis, these tools can't perform it, because they only monitor a very small selection of items within a category. Market Performance is always a relative concept - it depends on the category standards and to what other competitors are doing. A 8.3 rating out of 10 might sound fantastic, but what if the market average is 8.9?
If you are simply looking for a Reviews Monitoring tool, and you are not interested in a competitive analysis that tells you about market trends, what really matters to consumers, which competitors to watch and what to do to improve your Brand performance, there are many tools in the market for this job.
MASSIVE, on the contrary, is built for Market Performance and Consumer Insights analytics. We are ALL about competitive analysis, so we analyze every single Brand or item sold in any given category, no one excluded. So when we evaluate the performance of an item we know exactly its true strengths and weaknesses, giving you the opportunity to really understand your market dynamics.

No we are not. As a matter of fact, we expressly avoid using Social Media data.
For many industries the impact of Social Media data on a Brand's reputation has been widely overestimated. When it comes to consumer insights, i.e. understanding the reasons behind people’s purchase behavior, Social Media posts (Twitter, Facebook, etc.) bear no information - apart from some specific industries like fashion.
The fact is that using Social Media data is actually harmful to any serious Brand performance analysis because they introduce a lot of noise and dramatically reduce the Signal/Noise Ratio (SNR).
In broad terms, Social Media posts provide a very generic indication of Brand appreciation. In the Consumer Journey (explained here), Social Media are relevant only in the Awareness phase. People do not go on Social Media to specifically discuss a service or product performance, unless they are complaining about something. All the Brand related posts are somewhat very generic and pointless in nature, expressing a  general sentiment that has very little business impact.
While Social Media Listening platforms can effectively cover the first 1/4 of the consumer journey, and they are irreplaceable for Crisis Management, Massive covers the remaining three fourths of it, and the much more important Conviction phase - where actually purchase decisions are made.
If anything, Social Media Listening and MASSIVE can be used as complementary tools to cover the whole consumer journey.

Social Media posts bear no or little consumer insights. Posts are usually very vague, providing an overall sentiment (like/dislike), and they often refer to the Brand as a whole, rather than to a specific product.
People don’t write reviews on Social Media – and they don’t make purchasing decisions there either. In information theory, this means that Social Media bears little signal and a lot of noise - and a low Signal-to-Noise ratio is just not good if one's searching for meaningful information.
(Signal-to-noise ratio (SNR) is defined as the ratio of the power of a signal (meaningful information) and the power of background noise (unwanted signal) where P is average power).

We prefer not. Yes, our A.I. technology is incredibly effective at analyzing huge quantities of text and understanding and classifying what is described there, so it could be successfully applied to other use cases (Customer Service Management, for example). But to be truly effective, an AI solution needs to be  “tailored” to a specific task - what is commonly described as "Narrow AI".
"Narrow" sounds unexciting and limited, doesn't it? Especially when the Media and some nutcases keep selling the idea of the imminent advent of a big, omniscient AI God*. But as Wikipedia puts it, “all currently existing systems considered artificial intelligence of any sort are narrow AI at most”.
With more than 100 scientific papers published on AI, our team knows a thing or two about Artificial Intelligence, and we are big fans of this so-called narrow approach. It might not get the headlines, but it works for real. We might not be invited as "futurists and visionaries" at the next "1000 weirdos watchlist" roundtable, but we know how to make A.I. actually work (now THAT's really weird). Our AI technology has been designed, developed and optimized specifically with ONE very specific task in mind – to identify consumer insights from online reviews. And we are the absolute best at performing this task.

* According to some A.I. church pundits, the A.I. God will come into this world 20 years from now, give or take. It was 20 years also 20 years ago, so there's nothing to worry about, really.

We are a DaaS (Data-as-a-Service) company, but we see ourselves first and foremost as Insight providers. We think our clients want powerful insights, not mere data points. And we know most people don’t have the time to dive deep into a huge data pool like ours. This is why, aside from our data dashboards, we do provide Market Insights services in the form of special reports. Custom analysis about specific products or segments can be carried out upon request.

When people write meaningful reviews they usually express one or more opinions about specific aspects of their experience. This is where the true insights are. A review about a Face Cleanser might describe its fragrance, its cleansing power, how gentle it is on the skin, the texture, and so on. All these topics describe the consumer's real-life experience, and influence the judgment of everyone reading the review. The general sentiment of a reviewer is already expressed in the Rating - there's no need to calculate that again. But someone could have written a 5 star review about a product/service she's enthusiastic about, and yet she could have described negatively  a few aspects of the overall experience. And these negative aspects might be key to other consumers, convincing them not to buy.
At MASSIVE we run our analysis at an Opinion level. Our A.I. scans the reviews in search for sentences about key category performance topics, and then calculates the sentiment around each specific sentence. As a result we can profile the performance of products and services with extreme precision - one review might express many Opinions, and the quantity of Opinions is also a measure of the quality of a review (see our Reviews Quality Ratio).
Summing up: an Opinion is a Sentiment expressed about a Key Performance Topic in reference to a specific Brand, SKU or Item. And as we collect ALL the Opinions about ALL the products /services sold in a category, we provide the most comprehensive performance benchmarking analysis ever.

No, they are two  different metrics, although they are both aimed at measuring consumer experience. The problem of the NPS score is that it can be misleading and simplistic; it lacks the necessary depth to describe a consumer experience in all its nuances, and therefore it doesn't provide insights. about what to do to improve customers satisfaction.
The Net Opinion Score (NOS), on the contrary, is a proprietary MASSIVE's composite metric that can be easily split into all its performance components, thus revealing what are the specific areas of improvement that will have a positive impact on consumers' satisfaction. The Net Opinion Score is a weighted metric that calculates  the sentiment of the Opinions about a specific market feature - be it a product, a Brand or a Topic; this is then weighted against the normal sentiment distribution of the category.
The Net Opinion Score stems out of MASSIVE's unique data structure, as it relies on every market item being profiled on consumers' opinions about every key category performance topic.

The vast majority of text analytics solutions are based on 2 key concepts:
a) keywords search
b) frequency
These two characteristics combined are a sure recipe for disaster, and provide a very frustrating user experience too. The burden of the insight discovery is 100% on the client, who has to identify a specific topic and select a set of related keywords to search. This introduces an incredible amount of selection bias into the analysis; further bias is then introduced by skewing the dataset towards high-frequency items - a typical feat of "word clouds". Only the most frequent terms get visibility, while the long tail is concealed from view, causing a distorted perception. Here's an explanation:     a soap's fragrance can be described in a thousand different ways, but many of these terms are hidden in the long tail (low frequency - e.g.: "patchouli"). If one's searching for "fragrance", "scent", "smell" or other similar general terms, only a small fraction of the opinions related to the topic will be identified, and its real importance will be underestimated. In summary, traditional text analytics tools are not designed to manage Big Data. Quality and quantity of the data are arbitrarily selected to reduce the amount of managed information. It is a typical example of  "Iceberg paradox"; the area below the water is invisible, and yet its magnitude is significantly bigger of the visible part.
To avoid these issues,  MASSIVE uses AI technology that doesn't rely on frequency at all, and that automatically identifies the key topics, making  keywords searches obsolete. The magnitude of a topic is clearly seen, regardless of keywords frequency; what really count is keywords' meaning in the specific context.This technical choice ensures 100% statistical validity, a feature of paramount importance for consumer insights.