Taking Advantage of Twitter Search
This post was made Jan 21, 2010 by Carlos del Rio
This is an excerpt taken from Making the Most of Twitter: Search, Tools and Analytics. Visit our download page to buy the whole 25-page article.
This post covers in-depth how to use the search function to monitor Twitter. This will help you to find the people that you want to talk to and to understand the fast moving torrent of information and thought that Twitter provides.
One of your primary opportunities for leveraging Twitter is their own site search (search.twitter.com). Once you are familiar with this feature you will have better understanding of what third party tools are working with, why there are limitations and a clearer sense of what Twitter is building toward.
Twitter Search works very much like the search engines you are used to (Google, Bing, etc.); it employs 16 operators, or additives that affect the search’s meaning.
| Operator | Meaning |
| No Operator –> user driven | Containing both “user” AND “driven” |
| Quotation Marks –> “user driven“ | Containing the exact phrase “user driven”. |
| OR –> user OR driven | Containing either “user” or “driven” (or both). |
| Minus –> user -driven | Containing “user” but not “driven”. |
| Hash –> #measure | Containing the hashtag “meaure”. |
| From: –> from:UserDriven | Sent from username “UserDriven”. |
| To: –> to:UserDriven | Sent to username “UserDriven”. |
| @ –> @UserDriven | Containing the username “UserDriven” |
| Near: –> Sale near:seattle | Containing “sale” and was sent near “Seattle”. |
| Within: –> near:98122 within:2.5km | Sent with in “2.5 km” of “98122”. |
| Since: –> sale since:2009-10-04 | Containing “sale” and sent since the date “2009-10-04”. |
| Until: –> sale until:2009-10-04 | Containing “sale” and sent up till the date “2009-10-04”. |
| :) –> delivery :) | Containing “delivery” and positive sentiment. |
| :( –> delivery :( | Containing “delivery” and negative sentiment. |
| ? –> sale ? | Containing “sale” and a question. |
| Filter:links –> funny filter:links | Containing “funny” and includes a link. |
| Source: –> funny source:tweetdeck | Containing “funny” and entered via “TweetDeck”. |
Several of the operators are old favorites Quotation Marks, OR, and Minus are available in most popular search engines. The new tools are the user, location, temporality (time) and sentiment (feeling).
User and Hashtags
To: and From: are most powerful once you have identified key influencers for your customer group. From: will allow you to section out all of the tweets sent by the specific user and can be paired with search terms to identify the specific issues that you need to address. To: is used to identify who is speaking to, or dialoguing with, a specific account. The to: operator returns all tweets that start with the selected username.
Example:
@userdriven where is the meet up?
This will show up for to:userdriven.
I am going to the meetup @userdriven is sponsoring.
This will not show up for to:userdriven
If you want to return all mentions of a username then you use “@username” so @userdriven will return all tweets to the account and all uses of the username in tweets.
The hashtag # is used to tie together tweets on the same subject. One of the best examples of hashtag use is #journchat. Journchat has a large group of participants and a regular time each week when they hold a discussion, and everyone who wishes to participate can employ the hashtag to take part and follow the action. The hash is what allows Twitter to accomplish limited multi-person dialogue.
User and hashtags are the only operators that function as stand alone queries.
Location
Near: and within: will allow you to find out what people are saying in the vicinity of a specific location. For near: you can use a large area like City (Seattle), a smaller area like a Zip Code (98122), or extremely specific Longitudinal/Latitudinal values (47.625193,-122.3574).
Within: requires the near: operator and works best when paired with kilometers (km). Currently the dropdown that is found under the map on Twitter Search is in kilometers so you will do better to maintain consistency. Also, open Google Maps to verify that Twitter is correctly assessing your near: operator in the way you expect.
If you have both online and brick and mortar locations leveraging near: and within: will be especially valuable to you. Consider running vicinity searches around real world locations in tandem with online campaigns to drive awareness of your physical location and to expedite customer service resolution.
Temporality
Knowing when people react to your account, brand or product can be very useful. Unfortunately, Twitter has very finite server capacity. The result is that it does not hold data for very long. For the operators since: and until: Twitter only returns results going back 72-hours. This means that if it takes you more than three days to recognize that Twitter has spiked you won’t be able find the issue directly through Twitter Search. This means that even though you can use these operators to segment a single day (sale since:2009-10-04 until:2009-10-04) you have to do it very close to that date. The technical limit for the operator is seven days, but most phrases are purged more frequently than that.
For example:
If today is October 6, 2009 then you can use since: and until: for October 3rd, 4th, 5th and 6th.
This means that Monday is extremely important for your Twitter campaigns. This is the only day that you can assess the run into a weekend spike.
Sentiment
One of the major concerns of social media marketing and brand PR is sentiment–how people feel about your product, service or company. Twitter addresses this with :) and :(. These operators, signified by a smile and a frown emoticon, pull back tweets with positive or negative buzzwords (or the actual emoticon).
While great in concept the execution of these searches can be spotty. Both operators will return a set of emoticons that are essentially neutral, causing overlap in the returned search. Also, you will see a variety of results where you are not the subject of the sentiment. The sentiment operators are a good way to take a snapshot of the Twitter’s feelings.
The last operator that relates to sentiment is ? . It doesn’t actually affect sentiment, but it does allow you to find actionable issues. Appending ? to your queries allows you to see tweets that ask a question. So if you pair it with :) or :( you are more likely to find a tweet that has a clear solution you can bring to the author.
Special note: sentiment modifiers can disrupt location modifiers.
Source and Filter:links
Source: returns all tweets that were delivered through a specific channel–like Tweetie, UberTwitter or the main site. This is useful for charting the value of advertising in third-party tools or web portals.
Filter:links return all tweets with links in them. At this time I am not aware of other valid filters, but Twitter does plan to add retweets and home timeline segmentation to the search function, which will likely be other filters.
The real power of these operators is in combining them together to find opportunities for visibility or service of your customers. Combining Location and Sentiment can help you overcome a localized issue. Combining Location ? can help you find a business opportunity (e.g. lunch near:98122 within:1km ?) like finding people looking for lunch in your neighborhood.
Now that you know how to construct your complicated advanced search let’s cover some methods for saving you some time retrieving the results.
If you use the search box in the right navigation of your profile page you will find a “Save this search” button under the “update” button of your profile.
This is a good way to respond to searches quickly. You have your update field available, but the interface is a little confusing. Be sure you aren’t typing your new searches into your updated field when you use this method.
The second tool Twitter offers for leveraging search in a persistent way is a search widget. You can find this widget in the Goodies section of Twitter. It will allow you to set up a widget with any search that you want and define how you want it to display. We have set up a simple reputation portal that lets you enter a query to see positive, negative, tweets with questions, tweets employing the #fail hashtag, tweets with links and retweets using via. You can download the page here: http://www.userdrivenchange.com/twitrep.php
To get the full 25-page PDF visit our downloads page
Great post!