Online Marketer BootCamp presentation on the less obvious landing page optimisation tips and tricks

Are you an online marketer responsible for media budgets? Then you should at least have a quick look through the below slides (if you didn't attend our session at the Online Marketer BootCamp last week) to make sure all that media spend doesn't just evaporate but actually converts.

The slides mention the usual landing page optimisation best practice (i.e. headline, calls to action, social proof, etc) that everyone is writing about but then go into some of the things we don't see as much content on such as the importance of auto address completion, social subscription mechanisms, heat maps, statistical significance and unique phone numbers to optimise the offline user experience.

Click here to download:
201108 Datalicious Landing Page Optimisation V1.pdf (5.67 MB)
(download)

Check out the Datalicious Supertag: Container tag for smarter tag management

How to successfully implement Facebook's Open Graph to generate insights and start contacting fans

What is Facebook's Open Graph

The Open Graph is an open protocol for semantically labelling web content, but more importantly it provides the underlying logic for seamless integration into Facebook's social graph. Each configured URL becomes an object in the graph with various properties, like a title, author, image, URL, etc. People can be linked to objects through things like the Facebook like buttons (i.e. Johnny likes Datalicious). These relationships are then formatted nicely in Facebook news feeds and profiles to provide a means of virally sharing web site content. They also serve as semantic signatures, which can help search engines deliver more personable results.

Open-graph
Why would you bother implementing the tags?

There are many advantages to adding and configuring these tags properly. The key reasons are as follows:
  • Social relevance and influence is now officially part of Google/Bing search engine algorithms, so if you're into SEO and you're not looking at the Open Graph, then you're destined for rough times ahead. The "like" button is acting as a form of popularity score, much like Google's innovative PageRank, but each "like" is linked to a persons profile, allowing it to be authenticated and weighted accordingly.
  • Objects in the Open Graph drive traffic as they appear in news feeds and profile streams. For many sites this traffic is now greater than traffic from search engines. Without Open Graph tags and social widgets, you ARE losing potential traffic.
  • Properly formatted objects allow you to determine how the object will look in peoples profiles and searches. You can set the picture, title, description and many other important tags. This allows you to optimise your image.
  • Probably the most underutilised capability is the ability to contact users who have clicked like on your object. Effectively each like is functionally equivalent to subscribing to an email list, except it doesn't cost to send messages into users news feeds. Every "like" has a $ value, tapping into this new communication tool can be an extremely valuable exercise.
  • Objects in the graph are searchable in many other applications and likely will become a greater part of Facebook's built in search engine.
How to begin Implementing

There are several key pages on Facebook that explain the installation (see below), but don't believe everything you read, some of it is wrong (ironically eventually this post itself will likely be wrong). And to make matters more difficult, Open Graph tags do not create an object in the graph immediately (it takes time, sometimes a little, sometimes a lot!). The other thing that seems to influence the creation of the object is whether people have "liked" it, this appears to be important (although how many likes are required is also not clear).

Once you have worked through the creation of the meta tags on the site and they're testing ok (see the Linter below), you should be on track to start accumulating likes and generating traffic. To make use of the traffic you need to associate yourself as an administrator of the objects, to do this skip to the next section.

Resources
Facebook's Open Graph Developer Page
The Open Graph Protocol Page
The Facebook Linter - use this to check if things are working ok
Good blog article on how to get your web sites into your Facebook Insights

Sending messages into Users feeds

To do this you firstly need to make sure the URL has an object ID in the graph and secondly you must be an administrator of the object. To check the URL is indeed an object, go to the following address in your browser, but replace http://www.datalicious.com with the URL you want to check:

You want it to return something like the below (note the highlighted part means the object has an ID). Note: When you do this for thousands of URL's you won't do this manually, but initially it's a quick way to troubleshoot before attempting to post messages!

"http://www.datalicious.com/": { "id": "116002505130390", "name": "Datalicious | Data > Insights > Action", "picture": "http://profile.ak.fbcdn.net/hprofile-ak-snc4/hs463.snc4/50260_116002505130390_7413361_s.jpg", "link": "http://www.datalicious.com/", "category": "Website", "website": "http://www.datalicious.com/", "description": "Smart data driven marketing. Actively helping companies to optimise their marketing programs by providing accessible reports and actionable insights generated from solid data platforms.", "likes": 4 } 

Once you've confirmed your URL has an ID, you need to make sure you're an administrator of the object. This is usually easy to see, as next to a Facebook "like" or "recommend" widget you will see an "admin" button like below:

Screen_shot_2011-01-19_at_5
Note: The admin button does not appear next to like buttons without the "show faces" set to on, so if you have a button where you don't see peoples photos when they click like, then you won't see an admin button. If you're using a comments box, sometimes you need to click on "like", then "unlike" to make the admin button appear. Failing all of this, you need to check the open graph meta tags to ensure you're either giving admin access to either 1. An application, or 2. To specific user ID's, you should see something like one of the following on every page with OG tags:
<meta property="fb:admins" content="USER_ID1,USER_ID2"/>
<meta property="fb:app_id" content="1234567"/>
To send a message to the "likers" of your Open Graph Object, you can click on the "Admin Page" link and you will be taken to a page that looks similar to a fan page, from there you can post to the wall, which effectively pushes your post to all the people who have liked the object. Note: The post can also include URL's, etc.

If you want to send messages programatically or to thousands of objects at once, then you need to look at the programmable solution, the one thing to note here is that the "id" field must be numeric, it cannot be the URL as is incorrectly documented on the Facebook page and shown below:

Open_graph_issue
Case Study: GoPetition.com

Below shows the statistics from a site where the Open Graph tags were implemented properly. The site has recently peaked at over 3500 new likes in a single day, which is approximately an order of magnitude above new daily email subscribers. The staggering statistic is the ability to generate viral traffic and accumulate new subscribers to a key communication channel in a single step, with efficiency far beyond most other mechanisms.  

Fb_growth_dec_10

Contact us now if you need help implementing Facebook Open Graph on your website!

About the author

Check out the Datalicious Supertag: Container tag for smarter tag management

YouTube channel with easy to understand videos on statistical concepts such as regression analysis

For those of you who would like to get into the heavy stats side of data mining (or have accepted the fat that they probably have to), I just came across a great YouTube video channel thanks to Shailendra Kumar's blog, the current data mining lead at Woolworths (his blog is a great reas as well but a little more involved).

The StatisticsFun YouTube channel contains some really great videos with step by step explanations of statistical concepts including the below on regression analysis which is basically the mechanism behind the prediction of customer behaviour based on past transactions or the holy grail of data mining. How about watching a few stats videos of the Christmas break?

Check out the Datalicious Supertag: Container tag for smarter tag management

Stop misallocating media budgets using multi-channel media attribution with Google Analytics

Update: Video on new Google Analytics multi-channel funnel reports

Thanks to Google we're all used to measuring campaigns on a 'last click gets all the credit' basis (or occasionally first click). That is, all conversions (i.e. sales, leads, form completions, etc) are tied back to the last (or first) media channel that a person responded to respectively clicked on before converting.

Unfortunately, this method ignores all other touch points that a consumer might have responded to leading up to a conversion which also contributed to some extend, resulting in the undervaluation of certain media channels and ultimately the misallocation of marketing budgets (i.e. some channels are more likely to introduce a product rather than closing the deal and these miss out using last click attribution, see graph at the bottom comparing first and last click attribution).

Google is trying to address this through their new AdWords Search Funnel feature, however the Google reports do not include any organic channels or direct to site visits which do play a significant role in a consumer's path to purchase. In fact, organic search terms that include brand keywords and direct to site visits stimulated by some other form of media (i.e. TV, radio, print, etc) account for the majority of conversions on most websites (so we think they should be included).

To solve this issue for one of our clients, NDS (Carecareers), we used some custom JavaScript to record a stack (or path to purchase) of all campaign touch points across paid and organic channels in one of the Google Analytics custom segmentation variables (see chart below for a sample of the raw data). For simplicity sake we only recorded top level channel (i.e. SEO, SEM, direct, etc) in this case but this could be as granular as you want (i.e down to ad groups or even search terms).

Screen_shot_2010-09-27_at_6
The data on its own however is not very useful, you just end up with a long list of unique channel combinations (the above is only showing the tip of the iceberg) so we had to export the raw data from Google Analytics and analyise it using the Tableau business intelligence software. To make sense of the data and accommodate the various different purchase path combinations we decided to follow the ClearSaleing model and classify all touch points as either introducer, influencer or closer (see graph below and at the very bottom).

Picture_9

Looking at the simplified example above we realized that paid search responses play an important role both as introducers and as closers, but not so much as influencers (we're thinking that unbranded terms make up most of the introducers and branded terms most of the closers but we don't have the data yet). Conversely, the importance of organic search, direct to site visits and emails (activity just started) might have been understated in the standard last click based reports up until now as they are more likely to act as influencing channels (SEO might pay off after all).

Given the above results and relatively simple data collection and analysis method we think there's really no excuse for marketers anymore to keep relying on last click media attribution so please drop us a line at insights@datalicious.com if you would like to find out more.

(download)

Check out the Datalicious Supertag: Container tag for smarter tag management

How to add an 'email this post' link including Google Analytics tracking to your Posterous blog

Social media buzz or not, email is still the most frequently used tool to share interesting things with your network so why not make it as easy as possible for them?

Would you like to add an 'email this post' link with Google Analytics tracking to your Posterous blog similar to the below (Thanks @ianlyons, for prompting us to write again)?

Emailthis

The email would like similar to the below once people click on the above 'email this post' link, as you can see the email is ready to be send including perma link to the post and Google Analytics tracking parameters.

Email1

1. Ok, first you need to generate a URL with Google Analytics tracking parameters so any clicks are picked-up and recorded. Skip this step if you don't want to track responses from forwarded emails. 

To generate your URL go to the below page and fill in the form as shown as shown, generate the URL and copy it.

Campaign Source: email-post
Campaign Medium: email
Campaign Name: email-post

Urlbuilder1

The tracking URL should look similar to the below depending on how much you changed the above form text.

http://www.domain.com/?utm_source=email-post&utm_medium=email&utm_campaign=email-post

2. Now you need to prepare the custom mailto link code that you can insert into your custom Posterous theme including the appended Google Analytics tracking parameters. To do that go to the below site, fill in the form as shown, generate the mailto URL and copy it.

To: AAA
Subject: BBBTitleCCC
Body: I thought you might find this post interesting.
DDDPermalinkEEE?utm_source=email-post&utm_medium=email&utm_campaign=email-post

Make sure you add the Google Analytics tracking parameters highlighted above to the body text behind the 'DDDPermalinkEEE' part.

Encoder2

The mailto URL should look similar to the below depending on how much you changed the above form text.

Codea

3. Also, before we can insert the code into the Posterous theme we need to first clean it up a little and turn it into an actual link.

A) Completely remove the 'AAA' in the above URL.
B) Change the 'BBB' to '{' and the 'CCC' to '}'.
C) Change the 'DDD' to '{' and the 'EEE' to '}'.

Please make sure you only replace and remove exactly what I highlighted above otherwise the code may break. Once done the code should look similar to the below.

Codeb

D) Finally wrap the encoded URL into an actual HTML link tag. Once done the final code should look similar to the below.

Codec

4. Ok, you're ready to insert the code into your custom Posterous theme. For more information on how to install a custom theme visit the below page.

To have the link appear in the same position as on our blog copy and paste the HTML link code into the following position.

Code1
Check out the Datalicious Supertag: Container tag for smarter tag management