Tweettronics

General Information

What is Tweettronics?

Tweettroncs is a search and analytics platform built to help you measure and monitor campaigns on Twitter, by determining the reach, volume, influence and impact of your message.

Brands, People, and Twitter

Tweettronics both monitors and measures what people are saying. We listen to conversations on Twitter to discover general sentiment, who is participating and why, and different ways of characterizing a speaker's public persona and influence.

Active listening helps you to understand your customers and engage in relevant discussions. Identifying influencers helps you to understand the impact of your message and build relationships with potential brand evangelists.

Sentiment Analysis

What is Sentiment Analysis?

Sentiment analysis involves classifying text into categories that are "positive", "negative", or "neutral" to determine the attitude, or sentiment, of the speaker.

Tweettronics automatically estimates the sentiment of each tweet and categorizes it accordingly. Within the search dashboard results, you can modify sentiment and reclassify text as positive/negative, if you feel it was automatically categorized in the incorrect category.

Please note that cultural factors, linguistics nuances, and differing contexts make categorizing sentiment difficult for humans, and even more so for machines. Sentiment analysis on machines relies on humans to classify/edit whether words are positive or negative. If you'd like to contribute to making the sentiment more accurate, please let us know.

Sentiment analysis is limited to English at this time, but we encourage you to contact us regarding requests for additional languages.

The Search Dashboard

Creating New Searches

New searches are created on the dashboard. Select "New" to begin a search. On the search page, there are advanced options and you can choose to enter in multiple terms or exclude Twitter names. The "hints" tab provides search hints and examples. Search tips and techniques are also listed below:

Unless otherwise specified, completed searches store all tweets

Editing a Search

If you change the search terms of a search that has already been running for a while, the new results are added to those already accumulated. So if you do not want to add to the previous results you need to start afresh.

By default, listed tweets are sorted by date/time. Tweet options, date ranges, and sentiment classifications can be changed.

Removing a Search

When you remove a search, all stored tweets are removed and data cannot be recovered.

Comparing Searches

Head to head comparisons are useful for competitive reporting. In the search dashboard, you can select two or more saved searches to compare. The results are graphed by volume trends, sentiment trends, and user trends.

Paid accounts allow for any number of searches to be compared with each other.

Exporting Data

Paid accounts can export data from the search dashboard in CSV format.

We have the ability to provide more ways of processing the data for results. If you prefer different data fields that we are currently supplying or have specific requests for the data sets you would like exported, please let us know.

Search Tips

Try Out the Search

Try a quick search on the dashboard by entering in a search term and selecting “Try Out the Search”. Results will appear on the same page.

These searches run faster than saved data searches, and can be useful if you want to check that your search terms are returning relevant tweets. Once you've run a search you want to keep, you can choose to save it.

Excluding Users

You have the option to exclude the tweets of specific users in your search. Excluded tweets are not removed but are filtered from the report.

Please note that it may take a while for data to load when changes are being processed, especially for high volume searches.

Automatic Updates

Searches saved in the dashboard are automatically updated several times a day.

If a search has a high volume of daily activity it will be effectively sampled throughout the day, as every possible tweet cannot be obtained due to current data restraints. However, Tweettronics may provide improvements for the handling of high volume searches/trending terms. If these types of searches are important to you, please let us know your expectations for managing high volume queries.

Alerts

If a search has unusual changes in its volume, sentiment, or audience size an alert will be generated and sent via email. You will be alerted for a given search no more than once per day.

Finding Mentions and Hashtags

You can find mentions of users by searching for
@TwitterName

You can find hashtags by searching for
#HashtagName

You can find all instances of a name by combining these together:
"person name" "TheirTwitterName" "#TheirHashtag"

Search Dashboard Results

Trends

Trending graphs depict the volume, reach and sentiment over a given period of time. Lines indicate values and moving averages. Tweettronics automatically calculates if data movement is statistically significant using the average and standard deviation to determine low and high boundaries. It also computes when near term trend is crossing the moving averages. When averages cross these lines, it suggests a significant change in audience attitude.

Words

The word cloud represents the most commonly found words within the searched tweets. The number next to each word represents frequency (how many times each word was found).

Phrases

Key phrases are categorized as positive or negative and grouped accordingly. This is an experimental feature Tweettronics is developing to discover how word grouping in sentiment analysis improves Search Engine Optimization.

The number next to each phrase represents the frequency of the word grouping (for example, "Twitter conference metrics" 5, indicates that these three words were found grouped together 5 times).

This feature provides insight into how key phrases are used in connection with a brand/name/topic search. We welcome your feedback and encourage you to contact us with any comments or suggestions on the use of key phrases.

Tweets

The tweets listed can be sorted according to your preferences.

The first column, “+/-”, indicates if tweets contain positive or negative sentiment. The “Text” column includes all tweets with the specified search terms. The “Author” column shows the user's name (Twitter handle).

Tweets are sorted by date/time but date ranges, tweet options, and sentiment classifications can be changed.

Reach

Reach is represented through impact, name, and number of followers. Impact is the number of messages the user has sent on a topic multiplied by the number of followers they have.

The size of the overall conversation is represented by tallying the number of Speakers directly engaged, and, separately by the sum of their follower counts.

The speaker list can be sorted and date ranges can be modified.

URLs

Here are the URLs mentioned in the tweets. Tweettronics autmatically extracts these and provides the most commonly mentioned domains, and the most commonly mentioned URLs.

The “Count” is the number of times each URL appears in tweets.

The “Actual Page” URL is the page where the referenced URL points. The number next to referenced URLs shows the number of actual URLs referred to.

The “URLs To Be Processed” is a list of unknown URLs, meaning that Tweettronics has not yet been traversed to find where URLs are pointing. Refreshing the page will cause Tweettronics to continually traverse.

The URL lists can be sorted and dates can be modified.

Influence Analytics

What is Influence?

Influence is the reach and impact of a message, and the ability of a user to affect others through their character and behaviors.

What is Reach & Impact?

Reach is the impact, name, and number of followers.

Impact is the number of messages the user has sent on a topic multiplied by the number of followers they have.

You can search by name to find the influence of a particular Twitter user or you can search by topic.

What is Characterization?

The system makes automatic assessments as to the sociability of a user. We do this by characterizing what sort of networking role they present along two dimensions 1. being a consumer or generator of information, and 2. as a participant or organizer of social groups.

These assessments are intended to help in the determinination of the role a person would serve in communicating the messages about brand you are interested in promoting.

The system estimate how likely it is a person will read a tweet that is sent them. This affects how likely they are to be affected by the information and how likely they are to react and respond to it (via retweet or otherwise). So for instance, celebrities are difficult to market to because they are unlikely to see your communication.

Searching Influence by Name

Friend and Follower Relations?

The system also estimates how close they may be to their followers, which affects their influence on them. People who are close to each other will discuss different matters and have a deeper influence than those that we are not close too.

Some people try to follow everyone who follows them. But people who follow a high number of friends on Twitter have a lower probability of reading all of their friends' tweets (unless they are bots). The “effective friendliness” of these people is downgraded accordingly, and they may be classified as “strangers” rather than “friends”.

Where do the percentages come from?

We mine millions of Twitter accounts, and the number is still growing as more people join Twitter. We sample a large random set to establish a statistical baseline. With this baseline, we can make normative statements about Twitter accounts and behavior. The baseline is updated daily.

As a result the system knows and presents the relative rank of a user's friend and follower counts interms of the entire Twitterverse.

Effective Transmission Influence

These are two numbers that estimte the likely maximum reach a message to this person will have at most if, 1. the user sends it to his followers, and 2. if they sent it on to theirs. The number includes only unique users.

High Friend or Follower Counts

We do not provide assessments for Twitter users who have very large numbers of friends or followers (5,000 or more). However, please let us know if you want to customize results to include these users in your influence information.

Searching Influence by Topic

Why search by Topic?

Searching by topic allows you to quickly pull a list of people who are engaged on any given topic.

You can quickly see who is speaking and their impact. From their you can learn more each individual by clicking on their name to see their influence and characterization.

The results list impact, Twitter name, number of followers, and number of tweets on the topic subject.

Finding those engaged with a Brand

By searching for a brand name, or a competitors name, you can see who is speaking about them.

You can also choose to exclude a Twitter name in your search. This can be useful if you want to search for mentions of your or competitor's brand (ex. Tweettronics), but do not want to include tweet mentions that you've posted from your Twitter handle (ex. @Tweettronics).

Finding those mentioning a Twitter Name.

By searching for a Twitter name, or a competitors name, you can see who is speaking with and about them.

The Twitterverse

What is listed in the Twitterverse

These charts show how Twitter users rank against each other in the Twitterverse (based on the number of Friends and Followers).

The current top 10 users are also listed, and depicted as having the highest Friend or Follower count in the Twitterverse

These charts are updated daily.

Commonly Asked Q&A

How is data collected?

Data is obtained from Twitter 24 hours a day and is stored locally for no more than 1 week. We compute results using immediately available data from Twitter, in addition to data that's locally stored.

What if Twitter is down?

We rely on Twitter for our data. Sometimes Twitter may be over capacity, cannot fulfill our request, or we've exceeded our allocated Twitter resources. When this happens, we advise you to wait and try your search/request at a later time. Usually the problem is resolved quickly, and the good news is that these types of problems are happening less.

Why is my search taking longer than expected?

When you enter a new search on the Dashboard, Tweettronics pulls all of the available data from Twitter for the specified search. Initially, this includes approximately 1,500 tweets. Then a variety of analytics are performed on the tweets. For this reason, loading data can take a while.

Why do I see search results for just one day?

If you are searching for a popular trending term (like "Michael Jackson"), Tweettronics will only obtain the most recent tweets. Due to the high volume of these tweets, it's likely that results will only contain data from the current date (one-day range). Please let us know if this type of data is important to you, as Tweettronics may improve on the handling of high volume trends.