The Consumer Intelligence Blog - Infegy

Building a powerful social intelligence web app

When building a customized web application based on social listening intelligence, you have to start with a solid data framework and clear goals in the story that you want to communicate to your prospective audience.

This year, we had the goal of completely rebuilding our Most Trusted Brands Dashboard to provide even more customization and drill-down options to brand strategists and analysts. Our end results were fantastic. In the end, we offer both 27x more data per brand than last year’s iteration and interactive ways that you can drill down into the brands that interest you most.

We accomplished this massive increase of displayed data and interactivity with API calls to our new Infegy Starscape API, which enabled dynamic and customized aggregations.

Using Infegy IQ, our in-house AI which powers sentiment analysis and emotional identification, we initially query 2000 brands sourced from the Forbes 2000 to get a ranked list of the ones with the most trust. Then, for each of those brands, we query trends, linguistics, and geographic data. Each of those subcategories have both three different types of data and three different time frames. We precache our results to help with scaling. Because all the data pulls are automated, this data is live and updated weekly for your trust-tracking pleasure.

Let’s jump into the specifics of how we built our new dashboard. We’ll walk you through how we initially sift through brand data and display it in our interactive dashboard.

Infegys Most Trusted Brands of 2024
Figure 1: Most Trusted Brands 2024 relaunch.

How we built 2024’s Most Trusted Brands

Infegy's 2024 Most Trusted Brands dashboard has been completely rebuilt from the ground up with the new Infegy Starscape API to bring you even more insights about the brands you know and trust. Instead of 2023's static document, 2024's Most Trusted Brands data is updated weekly.

The initial collection

Unlike previous years, we expanded our brand list by 20x. We used the Forbes 2000, or the world's top two thousand largest companies by revenue, profits, assets, and market value. 

We send 2000 queries to Infegy's servers each week using the brand's Entity and name. Infegy uses Entities, or proper nouns representing big concepts like brands, to detect posts that might not mention the specific company name but still relate to it. For example, think of brands like Coach or Apple. You'll have many irrelevant results if you search just by those words, but Entity detection gets you content beyond the initial search query. Infegy IQ, Infegy's in-house proprietary AI, enables our Entity searches.

Related: Infegy’s Starscape API Harnessing Flexibility and Speed in Social Listening - A Webinar

We look at a rolling window of the past 365 days. Finally, we add filters for the Trust emotion and positive Sentiment. We specifically added the sentiment filter to control for the negativity that consumers could feel towards certain brands ("I'll never trust this company after they did this.”)

Once we run those 2000 requests to Infegy's servers, we sort the results by the number of posts containing Trust. These aggregated results represent a monumental amount of data (at the time of publication, Amazon, our leading brand, had 2.6 million documents containing Trust for the last 365 days). After sorting, we return a list with the top twenty brands. 

Now that we've compiled a list of the past year's top twenty brands ranked by Positive Trust, we'll explore each brand in more detail. We'll examine Trends, Linguistics, and Geography. For each of these examples, we'll use Nintendo, our third most trusted brand at the time of this publication.

Figure 2 - Initial query structure
Figure 2: Initial query structure for each of the two thousand largest brands.

Digging deeper into each brand

For our app, now that we've compiled a list of the past year's top twenty brands ranked by Positive Trust, we'll explore each brand in more detail. We'll want our users to have the ability to examine Trends, Linguistics, and Geography. For each of these examples, we'll use Nintendo, our third most trusted brand at the time of authorship.

We divided these three details into three different sections in our app for further analysis.

3. Nintendo-1
Figure 3: Branded page for Nintendo, one of our initial most trusted brands.


Trends shows how the conversation around Nintendo, or any brand within our dashboard, has changed over time. We show normalized post volume, sentiment, and emotions for each brand. Additionally, each view has three different time views (one year ago, one month ago, and one week ago). This sorting allows you to zoom into the most relevant period to see how the conversation has changed.


You're greeted with a universe graph when you load a brand page. This graph shows your brand's normalized post volume over the last year, month, and week. We normalize the volume by dividing by the total number of posts we've collected during that time period. We normalize to account for the year's data collection fluctuations, so you can see relative branded spikes in post volume.

3- Nintendo-1-1
Figure 4: Normalized post volume for Nintendo


Next, we grab branded Sentiment. We show the overall positive and negative share of conversation for each particular time frame. Usually, because we've filtered to positive Trust, the brands reflected on this dashboard should have very high average positivity. Despite the average positivity, you can see large harmful spikes in Sentiment when specific negative stories break over social media.

3- Nintendo_sentiment
Figure 5: Normalized Sentiment for Nintendo


Finally, we show Emotions. Infegy analyzes ten emotions (Trust, Joy, Love, Anticipation, Surprise, Anger, Fear, Hate, Disgust, and Sadness). Like with Sentiment, because we filtered specifically for Trust, Trust should be the leading emotion for each of these companies. Like Sentiment and Universe, we also normalize here so you can see relative fluctuations in emotional conversation.

3- Nintendo_time
Figure 6: Normalized Emotions for Nintendo.


After the Trend data, we grab Linguistics data. Linguistics data represents the underlying aggregated words that make up each particular post. Most Trusted Brands of 2024 shows you three types of Linguistic data: Topics, Hashtags, and Mentions.


Topics represent the underlying nouns, verbs, and adjectives frequently used within the underlying posts about a brand. These can have Sentiment attached to them. In this word cloud, we color by sentiment, with red terms appearing more in negative documents and green terms appearing in more positive documents.

Now is also a great time to talk about language detection. Entity searches are language-agnostic. Because Nintendo is a Japanese company, we're getting a lot of Japanese topics (we pull both Kanji and traditional Japanese character conversation).

Word cloud_sentiment
Figure 7: Top Topics for Nintendo.


After Topics, we also show hashtags. These terms, marked with "#," often can allude to more branded content. These hashtag attributes are especially evident in Nintendo's case, with hashtags like #NintendoSwitch and #TearsOfTheKingdom rising to the top.

Figure 8: Top Hashtags for Nintendo.


Finally, within Topics, we show mentions. These are accounts that are tagged most frequently within the Nintendo conversation. Mentions can be helpful in showing which accounts are driving the underlying conversation. We also color them by sentiment.

@hikakin @NintendoAmerica
Figure 9: Top Mentions for Nintendo.


Finally, we'll dive into our last tab on Most Trusted Brands: the Geography Tab. We aggregate data collected by country and show you which countries have the highest levels of conversation, Sentiment, and Trust.


First, we'll show you the Universe tab, which demonstrates the highest level of conversation. Nintendo has a giant market in the United States, so we see a high concentration of Nintendo-related conversation there. We also see pronounced discussions in Brazil, France, and Japan.

Figure 10: Post Universe by Country for Nintendo.


Next, we show Sentiment by country. This sentiment "s "is organized on a scale with countries shaded in green as more positive and Nintendo's shaded in yellow as less positive.

Sentiment by Country
Figure 11: Net Sentiment by Country for Nintendo.

Trust score

Finally, we show which countries have the highest concentration of branded trust conversations. This type of query wasn't possible in Infegy Atlas but is possible with Infegy Starscape's new API. We aggregate the country field with positive Sentiment and Trust to accomplish this with one query. Our API returns a JSON object, which is super convenient for parsing and graphing.

Map 2
Figure 12: Normalized Trust by Country for Nintendo

Sample query structure for Trust
Figure 13: Sample query structure for Trust by country aggregation.

Takeaways for your brand

In order to build an insightful social listening web app, you need the right type of API as your data foundation. Using just a few automated queries, we’re able to automate and scale insights for thousands of brands. By drilling down into the specific endpoints and time periods, our users can zero into the data points which are driving branded conversations into the largest companies in the world. We’re most excited about the recurring updates, which mean that our dashboard will now be able to live on its own in the background, continuing to bring new knowledge to our users in 2024.

The API behind this app is from Infegy, if you want to know more about how your application or project can be built using this data, contact us.