Keyword Referrals

Today Google announced that the default login behavior for users will be secure and encrypted. They also announced that query data for logged in users will no longer be passed to Google Analytics unless the user clicks a paid link. I think that default to secure is fine and good, but not passing the query data unless you are paying is shady. Knowing what people are searching for is very helpful for businesses in choosing what they change about their site. Not just for optimizing for traffic, but also optimizing for online experience and usability of the site in general. The people that are most likely to be hurt by this are the small and medium sized business that are dependent on Google products, and industries that are heavily skewed toward users of Google products. As an example:

• People who use Gchat
• People who keep Gmail open
• People that use Google Analytics
• People that us Adwords
• People that use Google+
• People that connect to Google Labs Products
• People that comment on Blogger
• People that comment on YouTube

Many of the changes that Google has made over the last few years have been working toward making ”Google” a more embedded experience on the web and more individualized. The trend is toward more people being logged in, which means we will see a continuing degradation of data. This is particularly worrisome for countries where Google has over a 90% share and for business that have fewer than 100 search visits in a day. For small businesses making smart data-driven decisions was already hard, now it will be harder.

Search Engine Rank Page (SERP) Results

Search is becoming an amorphous term. Some people argue that Facebook and Twitter are search portals. Even if we restrict the conversation to just Google there are multiple separate, but interconnected, search channels: Web, Mobile, Local, News, Blogs, Image, etc. Many people are very focused on tracking rank. The problem with tracking rank is that it is not entirely connected to traffic. There are vast difference in traffic between phrases in the same conceptual space and a very different position in the buying cycle between search channels. For some people traffic is the only thing being monetized, so all visits are equal, but most people have a conversion to consider. In many competitive cases you are also triggering universal search that pulls in multiple search channels, which means #1 Organic result can be visually the 11th result.

(Click to see full size)

No Organic results show up above the fold for that search: Locksmith. Who is getting those clicks? Rank is NOT a performance indicator. Traffic is a performance indicator, money is a performance indicator, phone calls are a performance indicator. Rank is an interesting bauble that has interesting information at the beginning of a campaign to see your competition and a very steep curve once you are in the top ten. There is often much more to gain from focusing on getting traffic from phrases that aren’t getting traffic yet, particularly because of partial match links. Applying strategies that value diversity and monetization of traffic over SERP rank will win in the long term.

We are having a book release party at Belltown Pub on 10/6 from 6-9 pm. If you can make it out we’d love to see you and will have new and old books on hand to autograph. We’ll also raffle of a book or two for free. There will be a free food spread and beverages available for purchase.  Please click one of the links below to RSVP:

If you have a Twitter account:

http://twtvite.com/2gy8te

If you have a Facebook account only:

https://www.facebook.com/event.php?eid=180906625317307

Or if you have neither, just contact us and we’ll send you all the details!

Thanks,

Jeff and Carlos

We decided to look at three key things to see if we could glean any info: mentions of @nfl or #nfl in tweets, actual number of tweets from @nfl, and the follower count for @nfl. Actual tweet counts from the main account allow us to understand how active @nfl was during this period. Follower counts show whether @nfl’s effort resulted in an engagement increase. Total mentions allows us to gain a robust understanding of the effects or “social prominence” of a term such as “NFL”.

Let’s review some key dates that will give context (all of this takes place in 2011):

• 2/6 – Superbowl
• 3/4 Proposed Lockout date
• 3/5 and 3/11 extended Lockout dealines (24-hour extension was announced on 3/3 and on 3/4 a 1-week extension was approved)
• 3/11 NFL Players Association Decertifies
• 3/12 “Lockout” begins
• 4/25 Judge rules in favor of Players and temporarily lifts lockout
• 4/28 Day 1 of the NFL Draft
• 4/29 Lockout is reinstated
• 7/21 NFL Owners vote to approve a new Collective Bargaining Agreement
• 7/25 NFL Players Association approves new Collective Bargaining Agreement
• 7/26 Free Agency Begins – the NFL is back

Based on these dates, I would assume that the mention counts might look something like this:

Further down is a graph showing how accurate or (inaccurate this graph is). However to gain further understanding of the affects that the lockout might have had on the term “NFL”, we explored actual tweets from the @NFL account itself. The tweets were mostly news based tweets with links to stories about which teams might draft which player: (summing the Rowfeeder data by month)

• ~310 tweets in February (mostly during the Superbowl)
• 100 tweets in March (no football related events this month, but lots of news about the pending lockout)
• 500 tweets in April (NFL draft at the end of April, court rulings)
• Roughly 200 tweets per month for May
• Less than 100 in June and July
• 200 tweets a month seems to be the rough average throughout NFL season, not counting February (which is when the Superbowl is held) and April (which is the NFL draft) – thanks to Tweetstats for the monthly info prior to our data collection

Finally looking at total mentions, from mid February through mid April traffic for the term “NFL” averaged between 15-20K tweets per day. Peak days were as follows:

• 40K on 3/3 when it was announced that the deadline for the lockout would be extended 24 hours
• 34K on 3/4 when no deal is reached again and the deadline is pushed out one week
• 90K on 3/11 – the day that the NFL Players Union decertified
• 58K on 4/25 when a court ruling  temporarily lifted the lockout)
• 198K on 4/28 – Day 1 of the NFL Draft
• 77K on 4/29 – Day 2 of NFL Draft; Lockout is reinstated
• 52K on 4/30 – Day 3 of the NFL Draft
• 50K on 7/21 – NFL owners vote to approve the collective bargaining agreement
• 74K on 7/25 – Players Union votes to approves the collective bargaining agreement

Our prediction was a little off but was surprisingly accurate for most of the data (ok, maybe a cheated a little bit).

So what this seems to indicate is that the lockout news definitely generated social buzz and helped produce additional followers that might not have been gained had there not been a lockout. The NFL draft seems to be the biggest news interest for NFL fans, though we have no mention data for the time during the Superbowl or playoffs preceding, so we cannot say that it is the overall biggest story of the year. A follow up post might show that the days following the lifting of the lockout and corresponding signing of Free Agents/start of training camps and pre-season games will probably also show a larger volume of traffic. Also, it would be interesting to visit these numbers next year and see how the averages compare when there are no labor issues to discuss over the off season and also compare it to playoffs and Superbowl week.

Your web analytics data will also give you some clues as to where your current visitor base is arriving from, but be aware that there are opportunities here that you won’t see, because you aren’t represented or linked to from that platform yet. You might find that organically, your users visit Facebook prior to visiting your website. Or you might find that while Twitter draws a much lower amount of traffic, these visitors are viewing product and converting at a higher rate than some other type of traffic. These insights will be useful for finding new opportunities and identifying what success will look like on new platforms.

Data on web usage, upstream and downstream traffic flow, and demographics can reliably be found from organizations like Pew Research, Nielsen Group, Google Trends, ComScore, and Quantcast. If you are working with a consultant or agency this is where they will be finding their data. All of these providers regularly update their data on time scales ranging from monthly to yearly cycles. Each has an area where they are fairly reliable or offer a unique perspective; you should take advantage of all of their free content. Regardless of your findings it will likely be advantageous to engage multiple platforms.

Some demographics about the major platforms[1]:

• Twitter – 55% female; 45% 18-34; 69% Caucasian; 60% make greater than $60,000/yr
• Facebook – 55% female; 42% 18-34; 75% Caucasian; 62% make greater than $60,000/yr
• LinkedIN – 52% male; 38% 35-49; 83% Caucasian; 69% make greater than $60,000/yr

Bringing this type of data together with consideration of how people access these portals results in Targeted Social Engagement. Once you know who you are talking to you need to consider how they like to be approached. If you enter a new social media platform in a way that breaks their rules of etiquette you will be in for some rough times.

There are 5 Categories this year:

Client/Practitioner of the Year (individual or group)

Most Influential Agency / Vendor (group)

Most Influential Industry Contributor (individual)

Web Analytics Rising Star (individual)

Innovator/Technology of the Year (individual/group)

Here is a full list of each category and each nominee: http://www.webanalyticsassociation.org/?page=awards2011_nominees

Carlos was nominated for the Web Analytics Rising Star award and based on his competition, I think he’s truly as deserving of the award as anyone on the list. I apparently am not the only one who thinks that. April Wilson (whom I have never met) feels like Carlos is one of the top three individuals on the list of Rising Stars. Thank you April, for your voice of support!

To reiterate some of the main points of why Carlos is nominated for this award: he attended three major conferences in 2010. He also hosted a slew of smaller gathering Web Analytics events where participation ranged from 5-30. He’s also very active in many of the Analytics forums including: #CROchat on Twitter, and the Web Analytics forum on LinkedIN. Carlos also finds a way to publish whitepapers and maintain this blog, in addition to writing his second book, which is all on top of maintaining his daily consulting job. Finally Carlos is always open to mentoring and advising younger analysts and giving them his two-cents worth.

So if you have not voted go vote and if you like Carlos and/or think he’s deserving of this award, vote for him!

The below graph shows average daily mentions for the term steelers throughout the season and highlight the high volume times throughout the season.

It appears a bit skewed thanks largely to the fact that there were 380K mentions on the day that they won the AFC championship. You’ll notice that this dwarfs the day of the Superbowl where the term had 103K mentions. In the case of the Steelers a major win translated into almost 4x the amount of mentions that a tough loss does even though the magnitude of the game was much larger in the loss. What is also interesting is that the previous game, on Jan 15, when the Steelers beat the Ravens, it too had many more mentions than did the Superbowl. Thus, it’s plausible that in this case positive sentiment overwhelmed and drove the mention count upward as fans celebrated their team’s victories.

Let’s look at some sentiment for mentions on Jan 23rd.

• 35% of all mentions contain the phrase “GO STEELERS” or contain a smiley.
• 15% of all mentions contain the words “win” or “love” and reference steelers (not jets, bears, or packers, or go, or smiley)
• Less than 10% of all mentions contain any sort of derogatory phrase toward the Steelers.
• 40% of all mentions are too ambigious to determine true sentiment.

So, roughly speaking there was 5 times more positive sentiment than there was negative sentiment, which is pretty good in the realm of sports where you have very opinionated fans, and then you have fickle fans. A team like the Steelers seems to have a very strong fan base who positively reinforce their team.

The question then might be who is tweeting?

Let’s take a look at some location data for January 23:

Mildly surprising that PA doesn’t hold a larger market share of tweets, however this state has a smaller population (and also splits the population with the Philadelphia Eagles). It’s not surprising to see NY as the top state, and CA as the third state. In fact, as the secondary graph shows, essentially the top 5 states by population are in the top 6 of percent of total mentions. So really this is unsurprising data in my opinion.

An interesting side note: of the mentions with the term steelers, here was the breakout of mentions that also contained a players last name for the 1/23 game:

Mendenhall (RB): 2,489 (121 rush yds, 1 TD)

Roethlisberger (QB): 2,296 (10/19 completions; 133 pass yds; 2 INT; 21 rush yds; 1 rush TD)

Tomlin (coach): 1,591

Ward (WR): 1,132 (2 catches for 14 yds)

Pouncy (C): 1,150 (broke a bone in the first quarter)

Polamalu (S): 832 (4 tackles)

Taylor (DB): 773 (3 tackles; game changing Forced fumble)

Harrison (LB): 266 (5 tackles)

Timmons (LB): 61 (10 tackles; leading tackler)

William Gay, a defensive back actually returned the fumble forced by Taylor for a 22 yard defensive touchdown that turned the game in the Steelers favor. However, filtering on the term Gay, did not render usable data, and at the very most it returned only a very small amount of data in total.

So again, it seems that positive actions combined with increased importance of a game in sports generate a lot of positive sentiment for a team, and thus drives mentions upward, while negative outcomes tend to lessen the amount of mentions. Also, higher profile games tend to generate more interest from a larger audience. I would wager that if you compared the location data for the AFC championship with that of a regular season game, a larger portion of the mentions would come from the local area. I shall write about that in a future post.

NY Daily News - Robbins/Getty Images

Over the next few posts I will be analyzing some social data for NFL teams and players pulled from Twitter using Rowfeeder and Scoutlabs interfaces.

Prior to the start of the NFL season, Carlos and I pondered whether a player could have more prominence than a team on a social network. Or, if not as prominent, then how does a significant player’s social prominence compare to a team’s social prominence. Further, would a significant injury, or amazing play show up in social data? Would it follow real time? Would there be continued talk about said injury or play hours or days or weeks afterward?

Several factors we considered: was the player active on twitter (and have a verified account); did the team have an official and active twitter account.

This post will focus on Troy Polamalu. We tracked mentions of “tpolamalu” – his official Twitter handle and “steelers” – the team’s official Twitter handle. Data was tracked from 9/8/2010 (the Wednesday before the season “kick off” game) to 2/6/2011 (the Superbowl). So it’s pretty amazing to have data for a player and a team from the start of the season all the way through the Superbowl (considering the chances of selecting the right teams to follow is 1 in 16).

TPolamalu received 106,985 mentions over the course of the season. Pretty impressive for the Defensive Player of the year – Avg of 718 mentions/day (but more on this in a couple paragraphs). Steelers received 2,518,000 mentions. 2.5 MILLION mentions on Twitter. So I’m going to devote a little more time to the Steelers data set before providing some analysis, but I promise to publish that soon. Again, thank you to Rowfeeder for tracking this data for us!

For TPolamalu there is a definite trend of game days being the most significant days, and moreso, big games like the Superbowl and AFC championship. More minutely, it does appear that the majority of the mentions happen during game time on game day. I’ll have to append a heat map for reference at a future date.

Here are TPolamalu’s top 6 days of Twitter mentions:

1. Defensive Player of The Year (1/31/2011)(No Game) – 4,166 mentions
2. Super Bowl (2/6/2011) (Loss)
3. AFC Champion (1/24/2011) (day after Win)
4. Game Day (12/5/2010) (Win)
5. Game Day (12/12/2010) (Win)
6. Bigest Online Shopping Day of the Year (11/29/2010) (No Game)

Late November and into December is where daily activity really started to pick up, probably due to increased awareness of the team’s Superbowl run. But it’s also holiday shopping season. To that end, surprisingly, Cyber Monday is also a good time to talk about TPolamalu because on the biggest day of the year for online sales, because it was the term’s 6th most active day with 1,815 mentions. Maybe that’s why his jersey was the #1 nfl jersey sold this season!

Troy Polamalu suffered an injury in the 12/12 game and did not play in the 12/19 game vs. the Jets (they lost), the 12/23 game vs. the Panthers (win) or the 1/02 game versus the Browns (win). These days all saw much less traffic than the two games mentioned above in December (when Troy played). So out of sight leads to out of mind? Possibly. It definitely seems to answer our question of whether an injury can affect social data: yes it can. He suffered an injury on 12/12 which was his 5th most active day and there were plenty of mentions of his injury in the data, and then the following 3 weeks when he does not play his social mention traffic dwindles down closer to the average seen earlier in the year. (It be interesting to compare this to a player who played all games in the month of December and saw a continuously higher mention count than in the previous months! So look for that soon.)

Further, during the month of December, TPolamalu mention activity averaged 200 per day vs. approximately 50 during the months of September, October, and November on non game days (and non award recognition days).  Daily average stayed approximately the same in January/February when taking out the two peak days in those two months. So, TPolamalu played 3 games each in September and October, and 4 games in November versus 2 games in December, 2 games in January, and 1 game in February, yet his mentions for the month are much more populous in the latter months. So it seems evident that as the team built towards a Superbowl appearance, and Troy was being recognized for his accomplishments, his social prominence grew significantly even when appearing in less games.

In summary: for Troy Polamalu, news of his personal recognition as the best NFL defensive player of the year, and witnessing his injury caused recognizable spikes in his social prominence. More important games such as participating in the most watched television show in US history also led to a higher mention rate. His absence from the football field seemed to lessen his social prominence. Cyber Monday also seemed to cause a spike in TPolamalu’s mentions, probably due to a larger volume of internet traffic in general. But as prominent  a player as Troy Polamalu is, his social prominence is only 4% of the Steelers (this will be an interesting comparison to Peyton Manning and Tom Brady’s social prominence which will come in a later post).

My next post will compare some of the similarities and differences between TPolamalu’s mention data and the mention data of Steelers.  As well, in a future post I will break down some hourly and possibly even by the minute details where Troy made a significant contribution, or was injured and compare this to mention data for the Steelers as a whole.

Image hosted on the Denver Post

For the past two years, I’ve posted something about the Superbowl, however neither we’re very data driven. Two years ago, I wrote about branding during commercials. Last year I wrote about how brands were beginning to use social media as an alternative to the traditional paid media channels. This year’s Superbowl was truly all access, mostly due to the effect that social media has continued to have on our society.

To this end commercials were even geared toward a more social perspective. Groupon.com advertised several commercials explaining the benefits of group deal buying. Many brands again took to the social space ad launched campaigns there. Foursquare even had “the superbowl” as a top trending location to check into. However, after checking in I was disappointed to learn that my location was not close enough to the actual Superbowl to earn points or a badge.

Several brands attempted to create a tie in between traditional media and social media. Specifically, Audi and Mercedes-Benz. Each launched a twitter campaign with the hope of creating a trend. These campaigns were actually divulged well before the Superbowl, so I was able to collect some data on them. (Data for this article is from RowFeeder and from Lithium)

Audi

Audi claims to have been the first company to use a hashtag in a commercial. As such this was their campaign. I believe their intentions were good and I’m surprised that more brands have not used a hashtag in their commercials. They are quite abundant in online media, so it only seemed a matter of time before they become a regular occurance in traditional media. That said, Audi’s execution of the plan was poor and thus led to a poor showing in the results.

Audi’s recent ad campaign has been directed at Mercedes and it’s seemingly stuffy clientele. A play on the children’s story “Goodnight moon” was in a commercial that aired during the AFC and NFC championship games two weeks ago. It was well received. It’s commercial that aired on Sunday was also well received, as they made good use of a cameo by Kenny G. However the campaign itself and the use of the twitter hashtag, were convoluted and not quite connected to the campaign. They chose the term #progresseIs and it was show very briefly between the “end” of the commercial and before Kenny G appeared on screen.  To clarify, this is spelling “Progress Is”, not “Progress Els”. Audi’s mission for this current campaign is that their cars are for people who are living a more progressive lifestyle. So I can see the relevance, but I could write an entire blog on why this is a terrible hashtag. I can also see why they chose to capitalize the “I”, but again, the point here is pretty evident –poor execution on their campaign.

The data tells a very interesting story.

In the days leading up to the Superbowl, @Audi averaged roughly 50 mentions per day, until the story about their campaign came out on 2/3. On this day, Audi  is mentioned 446 times on Twitter. #progressIs has no mentions until 2/2 where it has 29 mentions (someone with some inside knowledge I assume!) Then on 2/3 when the story broke this hashtag had 722 mentions and became a trending topic (the hashtag ending “els” had an additional 229 mentions on 2/3). On Superbowl Sunday @Audi had 1,131 mentions compared to 113 mentions of “progress Is” and 29 mentions of “progress els”.

So it wasn’t a complete loss for Audi. They did manage to create some buzz for their brand at the very least. And the author who broke the story of “progressIs” managed to create a ton of buzz for the hashtag.

Mercedes-Benz

Mercedes-Benz is arguably a more recognizable brand than Audi, thus it was not surprising to find that it averaged 1,800 mentions a day on Twitter for the week leading up to Superbowl. So the campaign by Audi, was seemingly just a small chink in the armor. However, while their execution of generating Twitter buzz may have had a slightly better result, the data seems to show that the execution may have been just as marginal. Mercedes-Benz attempted to host a ‘tweet-race’ in which four teams competed in a race based on how many tweets they received. Mentions of “mbtweetrace” were 2,995 on 2/3 when the story broke (the event had actually been going on for several days before the story however). 2/4 saw another 2,550 mentions, however the mentions fell off dramatically by Saturday with only 296 and on Superbowl Sunday there were only 91 mentions of #mbtweetrace.

That said, the individual teams which were “fueled” by Twitter mentions of their specific team names, or Twitter handles actually saw much more traffic than the race hashtag itself. There’s an interesting geo-location aspect here as well, since the teams traveled from four different locations across the country. The team from L.A. received nearly double the amount of mentions as did second place from NYC and almost triple the amount of mentions as the teams from Chicago and Tampa. None of these teams received significant mentions on Superbowl Sunday (150 mentions for the team from L.A., but every other team was well below 100 mentions).

So, in all I think there was some evidence of a bump in brand awareness on Twitter for these two brands, but moreso for Audi than Mercedes. However, as evidenced in both campaigns, a better execution will need to happen before another brand tries to influence a hashtag via traditional media. At the very minimum an easier hashtag should be used and it should be made more prominent in the traditional advertisement.

Coming later this week: tracking a team and player Twitter mentions for the entire season through the Superbowl.

credit: Nina Matthews Photography

After Barcamp Seattle and the Privacy Identity Innovation Conference (pii) I am struck by one painful truth about business and technology: All economic incentives favor inefficiency.

I gave a rant-style presentation at Barcamp titled “Your Product Sucks.” The central premise of the talk was that you have a single value to any given customer; either you fulfill it, or you suck. No matter how you strive you will not fulfill every customer need. The customer doesn’t care if you fail; they only care if you fail them.

pii focused on how we protect, legislate, and value personally identifiable information. There was an ongoing current of how people self-report and transport personal information through social media platforms like Facebook and Twitter. The introduction of Facebook Places has created an interesting problem: in order to differentiate from the existing self-report location services, Facebook has introduced other-report location.

During a screaming match passionate discussion I realized that Facebook Places is the continuation of a regular cycle. There are countless businesses that hemorrhage value. Every missing feature means money, every loose piece of data means money, and every support system means visibility.

“Broken” products and services create secondary and tertiary economies that become sustaining for the cycle and the starting product.  Think of tracking programs, cell-phone cases, and demographic data.

Two of the major secondary economies online are the protection of personal data and the reclamation of reputation space.  PayPal and Google Checkout both provide “security protection” by creating an intermediary between your bank instruments and online sellers. This comes at the expense of your credit data and personal data being centralized, the intermediary charges you for the transaction, and they keep a log of your transactions.

Think about that. Logging is a cost. The frightening truth is that there are very few laws about your information. Once people have the information there are few ways that they can’t use it. People can’t charge things to your credit card, but they can sell your e-mail address, home address, and what you like to purchase.

Every time you see information about who buys what, who uses a specific service, or any other identification of a demographic, is in an indication that information has been sold.

Facebook is a prime example of a business that takes advantage of the broken cycle they offer a clear function, connecting your online presence with your friends, but at the cost of sucking at privacy and security. They leave companies like Reputation Defender to clean up after the mess that Facebook “privacy settings” leave in their wake.

Read this infographic to see more data on how Facebook, Google and Apple use your data.

I strongly suggest that you visit the privacy settings at your free products and remember that your personal data is part of what these companies make money from.