Twitalyzer and TweetReach — A Symbiotic Pairing for Twitter Analysis
In yesterday’s post, I laid out a framework of measurement tools for social media measurement and analysis. This post goes a little deeper into how I see the relationship between the two twitter tools I listed there: Twitalyzer and TweetReach.
Let’s say you’re Chipotle. You’re not nearly as awesome as Freebirds, but you’ve got a much larger market presence, and you’ve embraced social media. So, what do you do when it comes to measuring and monitoring Twitter?
Online Listening Tools, Perhaps?
Chipotle, I’m sure, is already using one of the major online listening platforms — Radian6, Nielsen Buzzmetrics (or whatever their current platform name is), Converseon, SM2, etc. Online listening platforms purport to index all social media so that queries can be built to assess the volume, sentiment, and influencers related to any topic in any and all social media channels. Here’s the challenge when it comes to Twitter:
The Physics of Twitter apply!
Only two entities have unfettered access to all tweets, and neither have made their full datasets available to anyone:
- The Library of Congress
Twitter has never allowed full access to all tweets through its API. I ran a test using TweetReach and one of the three online listening platform leaders according to Forrester, and the results bore this out. For the 3-day period where I compared two simple queries (it was a brand that ran a TV ad during the Oscars, and that date was the middle day in the 3-day period I evaluated), the listening platform reported 513 total tweets, while TweetReach reported 1,225. Comparing the tweets, there were 6 that the listening platform included that were missing from TweetReach, and a staggering 718 tweets that TweetReach included that the listening platform was missing.
I wasn’t surprised, but it illustrates the point. Online listening platforms are good for what they do, but Twitter detail does not count.
So, What Tool Should I Use Instead?
There really is no single tool that tells me everything I want to know about a brand’s Twitter presence. But, I do think two tools complement each other well: Twitalyzer and TweetReach. These are both paid tools, but the costs are nominal. There are scads of other paid tools, as well as free tools (for a pretty good run-down on Twitter tools, check out Michele Hinojosa’s Twitter Analytics presentation), but I keep finding myself coming back to these two.
The key distinction between the two tools is that Twitalyzer is user-centric, and TweetReach is tweet-centric tool. The platforms have some overlapping capabilities, but their underlying orientation is quite different. To illustrate, think about how Chipotle might view and manage Twitter. The Venn diagram below is a possible representation of the relevant Twitterverse from their perspective:
Twitalyzer provides a range of metrics related to @chipotletweets (and competitors, such as @tacobell) — it’s oriented towards the green circles above. These include both straight-up measures — follower count, number of lists that include the account, number of tweets, number of times the account is replied to and retweeted, etc. — and calculated metrics that take into account how often the account is retweeted/replied to, the number of followers of followers of the account (and the number of followers of users that reference the account), and so on to estimate things like the overall impact of the account, its influence, and it’s potential and effective reach (click on the image to see the full range of available metrics, as well as to see current data):
Using Twitalyzer, you can track an account’s performance over time — be that your own account or competitor accounts. Within the tool, you can identify members of an account’s overarching network who have the highest impact in the channel, and you can do the same for a particular keyword or topic. The key, though, is that the tool operates primarily as a user-oriented measurement tool. For deeper thoughts on Twitalyzer, check out Michele Hinojosa’s post on the tool, and, if you still want more, I wrote a post dedicated to the tool as well.
One key note about Twitalyzer is that it’s not able to violate the Physics of Twitter any more than any other platform. It relies on the Twitter API to pull user data, and an unlimited user history is not directly available. So, with the paid Twitalyzer accounts, the tool goes out and checks Twitter every day and updates its metrics — keeping a running history from that point forward. If someone else is already having those daily updates pulled for a user, and you go in and “twitalyze” the user, you’ll get a nice set of historical data. If someone twitalyzed the user 2 months ago and no one has twitalyzed it since, then, when you twitalyze (twitalyze, twitalyze, twitalyze!!!) that user, you’ll see choppy data as Twitalyzer tries to fill in the holes in its dataset.
In short, don’t think that Twitalyzer keeps some sort of running record of all Twitter users over all time. The volume of daily tracked users they do have is growing, but you really have to load in the accounts you care about to get consistent data from that point forward. It’s kinda’ like web analytics that way — if you don’t tag it, you can’t necessarily expect to measure it effectively!
TweetReach is, in some ways, more like a traditional online listening platforms (Hah! Traditional! A “tradition” going allllll the way back to 2006 or so), in that it uses keyword-driven queries to tap into the Twitterverse (via the Twitter search API) and then pull and analyze the subsets of tweets that fit those query criteria. Basically, TweetReach is geared towards the blue bubbles in the Venn diagram above. And, not just the blue bubbles, but particular areas of blue bubble intersections (“fast food” AND “burritos,” for instance). With TweetReach, you set up a “tracker” using Boolean logic to include and exclude tweets based on keywords and phrases. The tracker then taps into Twitter and starts retrieving and analyzing those tweets. Part of the output you get in near-real-time looks something like this:
For the date range you choose, the tool provides the total reach (the sum of unique people following people who generated one or more tweets that met the tracker criteria) of the topic, the total exposure (similar to reach, but not de-duped to unique people), total tweets, and total contributors. That’s useful in and of itself, because it quickly shows the impact of a campaign or event, as well as the volume of Twitter conversation occurring around a given topic over time.
In addition, the platform provides all of the tweets that drove those numbers and allows them to be sorted by “highest exposure” — the tweets that generated the most impressions, through a combination of followers of the person who tweeted and the volume of retweets (and the number of followers of the people who retweeted). The tool also provides a simple list of users who generated the highest number of impressions. Plug some of those people into Twitalyzer, and you can get a quick read as to whether they’re someone you may want to keep an eye on and even engage directly.
Since TweetReach is aimed squarely at Twitter, it does a better job of comprehensively capturing relevant tweets than other tools — paid listening platforms (as discussed early in this post) or free Twitter archiving services. By downloading the raw tweets for a relevant timeframe, you can do some basic text analysis — generate a word cloud, for instance — and get a quick view of the sentiment, tone, and word usage of these tweets. TweetReach posted a couple of great examples of word clouds from trackers after the Super Bowl. Neat stuff, huh?
You Said There Was Overlap?
The tools do overlap. Twitalyze can do some hashtag and topic analysis, but it doesn’t have the Boolean querying capabilities of TweetReach. TweetReach can do some user-based analysis by running reports and trackers using “from:<username>.” Both tools report “reach” numbers. But, with Twitalyzer, it’s the reach of the user, while, with TweetReach, it’s the reach of the topic. Both are valuable, but they’re different.
I’m sure there are other ways that they overlap, but I wouldn’t want to run solo with either tool! They complement each other fantastically, though!