Twitter for business Part 2: market research via keyword analysis

The most powerful concept in marketing is owning a word in the prospect’s mind.  – The 22 Immutable Laws of Marketing by Al Ries

According to wikipedia, a Unique selling proposition (USP), refers to “any aspect of an object that differentiates it from similar objects.” So go back to our previous coffee machine winter campaign example, what is the #1 word that can win a customer’s mind?

This is one of the main purposes conducting quantitative marketing research, which typically means to create a survey,  find a group of random people, invite them to answer a questionnaire and have some statisticians to come up with the conclusions.  Depending on the scope, market research costs anything from a few thousands up to millions of dollars, making it out of reach for many mid-sized to small businesses.  With the vast amount of personal opinions on twitter and a powerful analytics tool like Tribalytic, you can easily do research peoples real opinions in a few minutes. Here is how.

Given any search, Tribalytic will take a evenly distributed sample of 500 tweets and extract the top 52 keywords, providing a good summary of what people are talking about.  Here are the ones for “coffee machine”.  This can be a bit intimidating at first sight, but keep reading and we’ll group the words together in a way that soon you’ll be able to find the “hidden gems”.

Ignoring keywords with less than 2% (more on why this magic number later) the keywords can be put together into following groups:

  • Nugget: broken, working, fixed, buy (bought)
  • Where: office,
  • When: week, today, morning
  • Emotion: love, carefully,
  • Noise: china, #pawpawty, bone are from the same kind of tweets like this which can be safely tossed away.

By sifting through the natural noise coming with the tweets, now we can see that a common theme of people talking about the physical status of their coffee machine.

Nuggets in the tweets

If you drill down by selecting keyword “broken”, you may find a tweet like this sent at 4am on a freezing Australian winter morning:

Compare the above message with this one and it’s no surprise that people are expressing their frustration about “broken down” coffee machines.

Remembering the quote at the beginning of this post,  what’s the word you should own? I’d say it has to be something that will bring out the warm, cozy feeling from owning a trustworthy, hard-to-break coffee machine.

Before we finish this post, we’ll have to talk about the 2% magic number.  As you have seen, the valuable information in tweets can come with irrelevant stuff.  So how do you make sure the conclusion you derive from analysing tweets is statistically significant? To put it another way, how can you be sure that the number we get is unlikely to occur by chance? The solution is to make sure that a minimum percentage of tweets have the words.  Based on the current sample size: 500, 2% means 10 tweets, which is the minimum number we can trust.  While this is not strictly speaking statistically correct,  it works well enough here as a simple rule of thumb.

Tagged , , , , , , , | Leave a comment

Twitter for business Part 1: improve daypart targeting

There are many ways to catch trout.  One, which does not require either training or fitness, is to buy a hand grenade, remove the pin, throw the grenade into the pool, and, when it explodes, scoop out the bodies in a net. That is the way that many companies have traditionally advertised, but financial restrictions mean that they now have to find more skillful, intelligent ways of attracting and retaining customers.

Truth, Lies and Advertising by Jon Steel

This is the first part of a series posts on how businesses can use information mined from twitter to optimize their advertising campaign and increase their ROI. More specifically, how a market research tool like Tribalytic can help companies to improve their media planning, campaign effectiveness measurement and brand equity tracking.

This post will focus on how Twitter data can help with daypart targeting.  Radio introduced the idea of daypart targeting decades ago.  Typically radio separates a day into morning drive, midday, and evening drive with the “prime time” as morning and afternoon drive time.  TV picked up the torch and developed its own daypart scheduling.  Despite the well known success of Budweiser’s advertising on Friday afternoon and KFC Popcorn Chicken around lunch hours, daypart targeting has been underused online because it’s hard to discover when your target customers “want” your product during the day without significant and expensive market research data.

Continue reading

Tagged , , , , , , | Leave a comment

Top 10 Australian Twitter TV shows: Week ended 29 August 2010

Numbers were down across most shows this week. This may be due to the Ben Cousins “Such is Life” documentary potentially taking viewers from Gruen Transfer in particular. With the absence of the election, people seem to be less engaged, it will be interesting to see how this pans out in coming weeks.

Last night’s A Current Affair figures should make next weeks Top 10, with Bert and Patti Newton discussing Matthew Newtons various issues.

Top 10 Australian Twitter TV Shows

(lw) Rank Network Show #hashtag Unique Tweeters Total Tweets
(1) 1 ABC Question and Answer #qanda OR #quanda 2,464 8,100
(-) 2 Seven Ben Cousins – Such Is Life #bencousins OR #suchislife 761 1831
(2) 3 ABC Chaser – Yes We Canberra #chaser OR #ywc 467 790
(3) 4 ABC Lateline #lateline 414 883
(4) 5 ABC Greun Transfer #gruen OR #gruentransfer 185 260
(6) 6 ABC Media watch #mediawatch 180 263
(5) 7 ABC Four Corners #fourcorners OR #4corners 130 253
(10) 8 Seven Dancing with the stars #dwts OR #dancingwiththestars 127 304
(-) 9 Ten Neighbours #neighbours OR #neighbors 115 223
(7) 10 Ten Offspring #offspring 88 154

Please note – these are indicative figures only based on the Australian TV shows we are tracking. If you’d like us to include more shows or channels, please add them to the list. The methodology and approach for producing this listing is described in this post Why rate Australian Twitter TV usage.

Tagged , , | Leave a comment

Measure your ROI on Twitter – New Tribalytic Release

If you need to accurately measure brand effectiveness on Twitter and research how to target your marketing campaigns, then you’ll love our new features launched over the weekend.

Our new Share of Voice measure lets you stop just counting tweets and instead focus on understanding exactly what percentage of the Australian Twitter Audience were actually engaged, all delivered within seconds by our new engine, meaning you can find the answers faster than ever before.

The Hourly Engagement feature shows when people are talking about your brands, letting you understand how and when to target them both in Twitter and through other mediums.

To experience these new features now, all Trial Accounts have been reopened for another seven days.  Go to Tribalytic.com and log on with the email address you registered with.  If you’d like to know more, please read on!

Continue reading

Tagged , , , , , | 1 Comment

Coverage in The Next Web

If you’re interested in a little bit more background on Tribalytic and the benefit we think it brings for Australian Business, you might like to read this article over at The Next Web.

I was interviewed by TechFluff.TV (published in The Next Web) about Tribalytic and where we are going with it – it was a lot of fun and I think the whole series will be interesting to follow (of course I am biased).  I’m looking forward to hearing from some of the other Melbourne startups and about their stories.

Tagged , , , | Leave a comment

Sneak peek of the next release – hourly graphing

Normally we keep all the goodies until AFTER we’ve released the product, but after a hard slog with the new engine over the last 6 – 8 weeks, it is such a relief to be moving forward so quickly now that I really wanted to share something coming up very soon that I’m really excited about.

We’re including two (maybe three) significant features in the next release which we think really add a lot more depth to our analytics.

The feature I want to share is the graphing of activity by hour, inspired by our previous analysis of Australian Tweets and Users by Hour.

The above graph shows the trend for #qanda over the last 2 weeks, then the related keywords and now a graph of Tweets and Users by hour (FYI these times are AEST and QandA screens from 9.30 – 10.30).

Continue reading

Tagged , , | 2 Comments

Top 10 Australian Twitter TV Shows: Week Ended 22 August 2010

The Top Four remained the same with QandA still holding number one spot, although it’s overall numbers were down slightly. Chaser’s Yes We Canberra jumped back up to second and slightly improved, while LateLine continues in number 3 for the third week in a row. Gruen Nation dropped from second back to fourth.

60 minutes returned to it’s more usual volumes (13th this week) after the Mark Latham affair the week before had pushed it up to 5th spot.

Continue reading

Tagged , , | Leave a comment

Get more insight into Australian Twitter users faster with new Tribalytic release

Tribalytic has always been about Social market research made simple, providing you instant results at your fingertips.  With our latest release incorporating a brand new analysis engine, we are delivering our in-depth research in a simple to understand way faster than ever before.

Now even the largest queries return within a few seconds, allowing you to explore even more issues “at your fingertips” with your clients and stakeholders.

This significant (100x) speed improvement also allows us to start adding more users into our research panel, providing an even greater pool of Australian twitter users to analyse for your business queries.

Another major improvement is in our related keywords which provide insight into the results of your search.  We’ve significantly expanded the number of related keywords from 15 to 52 (that’s right Fifty Two *) and included the occurrence percentage so you know how frequently the related keyword was used.

Continue reading

Tagged , , , , , | 1 Comment

What do Australian Twitter users first names reveal?

A genuine business question about Twitter is about the make up of the Australian audience. How is it made up, what’s the average age and where are people coming from?

We are taking our first steps down the path to answering some of these questions with Tribalytic. I was doing some research today into popular names and I thought it would be interesting to compare these to the popular names from each decade and to try and “guesstimate” the age of people using Twitter.

It turned out to be really interesting. I haven’t done any hard analysis on this, but it does give some high level general conclusions about the likely Australian audience makeup and confirms other statistics I’ve seen that suggests that the average Australian Twitter user is late-20′s and up.

Continue reading

Tagged , , , , | Leave a comment

What are Australian Twitter users lovers of?

OK, I admit, this one comes from the department of trivial information again, but I couldn’t help myself.

Extracting the twitter profiles tracked by Tribalytic a couple of months back I produce this post about how Australian Twitter users describe themselves on Twitter.

Gavin Heaton (@servantofchaos) picked up on this today, noting that we are “lovers”.

But lovers of what?  I extracted all the profiles that included the phrase “lover of” (n=932) then normalised to terminate at the next punctuation mark (, ! . ; etc).  I then took this portion of the profile and produced the following tag cloud.

Continue reading

Tagged , , | 4 Comments