This post has now been updated with an official response from Tweetmeme – the features have now been added to the development timeline.

The Tweetmeme site is a useful content source, but this article is mainly about the Tweetmeme buttons.

Adding something to your blog, especially something that can encourage viral behavior and add social proof is highly beneficial, and doing it correctly, without asking for passwords offers huge advantages.

Twitter is also a medium that is marketing content tollerant, as long as the people following you are marketers, unlike Digg, Reddit, and to a lesser extent Stumbleupon. 

Thus I have spent some time delving into the code and API for Tweetmeme and the existing WordPress plugin. It has just been updated, thus if you download the Tweetmeme plugin, it will be the latest version.

Important New Features

• Cleaned up cookie use a little, but it would be better if they set a cookie based on an action, rather than a display.
  The original RFC for cookies states that browser support for cookies should be a minimum of:-    
  • 300 cookies in total
  • 20 cookies per domain
  • 4095 bytes per cookie

  To be perfectly honest, I don’t want all those cookies, whether they are 1st party or 3rd party.

• Ability to brand the retweets with @andybeard etc, though you still get “via @tweetmeme” at the end
• The button is removed from excerpts using a filter on get_the_excerpt() – more on that to follow

However essential features of Tweetmeme and Tweetmeme Button are missing if it is going to be used by serious marketers.

Whilst I am not a huge fan of the @tweetmeme at the end of button tweets, it is something that users can just delete. However I have some additional problems / requirements / feature requests

1. The latest incarnation of Tweetmeme prevents the button, as standard, appearing on custom excerpts.

I can understand this from a user perspective:-

• Most bloggers are using automatic generated excerpts that might include the button javascript
• This appears ugly on the page, and is useless if they are also using that for their page descriptions

However some people use custom excerpts and custom meta descriptions, and like displaying tweet buttons on their front page without adding custom filters to functions.php or hacking theme files. Can the options be expanded with a separate set for use on the_excerpt? I can also see people wanting to use different buttons on full page, excerpt and rss, and only have one button on the_excerpt, but top/bottom on both feeds and posts.

2. The Headlines Used

Currently there is no way to define the text that is being used within a tweet – the text used is the Title tag from the header of a page. Many blogs use various titles for different purposes

• In the header for SEO
• For the title of a post
• Various navigation elements
• Custom titles for RSS feeds
• Calls to action for Social Media
• Calls to action for Twitter, or what Tim wrote about a few months ago, Social Bites

I really need to be able to define the text dynamically for each button display.

3. The Links Used

Tweetmeme like tracking things. Guess what? So do I…

I don’t so much care about branding of the final tweet links – it would be a bonus, but I need to be able to track the way I want to track.

• Within my short URL software
• Google Analytics for various goals
• Possibly for use with affiliates

I understand that Tweetmeme need the permalink to associate with tweets, but couldn’t I provide that, in addition to the link actually used to generate links? You might end up generating a lot more short links, but you would know in advance, without having to check headers, that all those links are to the same URL.

4. My Tweeps Like To Track Too

As a person who might occasionally click on Tweetmeme buttons, there is one fatal flaw.

If I use my own link shortener, I can track the response rate. I like knowing which topics and headlines generate a response from my audience. If something proves to be popular, maybe I will repeat it for people in a different timezone.

Tweetmeem Buttons remove my ability to test the popularity of the content I post.

The solution is actually fairly simple:-

The tweetmeme buttons are quite happy to write cookies all over a users system, why not read just one cookie, a unique user ID for Tweetmeme as well. The default user ID would identify which URL shortener they prefer, and their API ID. If you want to be extra secure, you could store both of these in a cookie on the user’s computer, rather than store them yourself. Links would ultimately go through the following redirects

• Tweetmeme Bit.ly
• Tweep (any)
• Publisher Short (any)
• Permalink + parameters
• Permalink

That is a lot of redirects, it might slow things down a little, and Googlebot might not like so many jumps, but ultimately this provides a better experience for users, and for once any SEO factors aren’t important – Google will have to cope. The 2 URLs to include with the javascript would be

• Permalink
• (optional) Publisher Short (any) / Permalink+Parameters

5. Split-testing

The observant among you would also notice that with all these parameters being set with each page load, it would be possible to split-test response to some degree. This level of functionality would add real value, and might even be looked on as a useful, modestly priced, premium feature – $10 per month/site  or $30 per month/site unlimited? For tweeps it would have to be free.

Thus I think Tweetmeme is a great idea, but to be turned into something that not only remains on my blog, but which I might even be willing to pay for, it needs significantly more flexibility.

Tweetmeme could be turned into a masterful, measurable viral marketing tool, with just a few tweaks.

Update: Tweetmeme Marketing Features Coming

Sarah from Tweetmeme has just left this message in the comments area

Hi Andy,

Your suggestions have been added to our todo list here in the office so do look out for the new features you’ve requested. They will take us a while to do but they are coming.

We’re using your post as our guidance, and if you are up for it when we have something to show, would you be up for looking over it and seeing if there’s anything we’re missing? We would like to get it right and have a really strong offering for marketing experts so feedback and working closely with experts like yourself is essential really.

Lets keep the conversations going between us and if you drop me an e-mail then I’ll make sure you are on our early adopter program :)

Many thanks
Sarah

I think this is a superb response and I want to thank Sarah, Nick and the rest of the Fav.or.it team for accepting my challenging post “head-on”.

I spent the better part of a week and a few hundred dollars ($300-$400) on various WordPress sales letter and affiliate themes – I came to the conclusion that they weren’t something I would use.

• Some I would class as pretty but lacking substance (polite version of crap)
• Every single theme was designed around the concept that you run a single theme on your blog at a time
• They all suffered from what I regard as a horrible plague – theme option pages – if you need to set up lots of options after you have uploaded a theme, you have lost a sale, unless I can also import and export settings.
• The worst aspect of theme options pages is using them for the content that might appear on a page, because it is a nightmare to set up split testing, and when editing, you really want all the words on the page in front of you.

So I set about creating a solution for myself…

I started of with Thematic which I had already been playing around with for a good 8 months. Themeatic is an offshoot or branch of Sandbox of which I was also a fan, but built upon a grid system that hopefully I won’t totally destroy when I finally get around to theming on this site.

The first challenge was custom CSS for every page, and navigating my way through the maze of which functions initialized at which time within the WordPress core, such that the style sheet used is determined by the page selected in the default custom layout selector.

The end result is only 160 lines of code in functions.php, plus some in individual custom theme pages, and that may well get shorter as I optimize things, but the end result is something unique and useful.

I am a strong believer in doing things once

Once I have created a custom sales page theme for this system, though it is not limited to sales pages, it can be used without any setup overhead other than selecting it within a dropdown list and hitting save.

And of course, everything is created based upon a child theme of Thematic, thus when Thematic has one of its frequent updates, there will be very little if any pain in upgrading.

WordPress Split Testing With Google Website Optimizer

There are a number  of plugins that claim to provide support for Google Website Optimizer. Most only work with A/B split testing as the authors couldn’t figure out the best way to add tags to post content.

In addition often the GWO code was placed in the wrong place in the header, it needs to be after Doctype, but before the CSS to effectively fully replace the CSS – I realise CSS is meant to be cascading, and you can force your way through when it doesn’t quite work out, but sales pages are meant to load fast.

So our single CSS file by default no matter what style the page is gets wrapped with GWO selectors by default. The most important element on the page, the post title which most of the time is used as a headline also gets wrapped in selectors by default.

You don’t have to test them all the time, but they are there when you want them. 

I eventually selected one plugin solution, from an Italian company who do landing pages. I may end up tweaking things more, moving various thing from functions.php into the plugin, though many of the hooks I have used are specific to thematic, even the one I ended up using within the plugin to get the correct placement of the GWO code in the header.

Whilst I claim that this video would only be 5 minutes at the start, it is actually 8 minutes, but demonstrates the full process of setting up split testing with my current solution.

I am sure you are also curious about how it works, so here is the demo sales page from the video, with split testing currently enabled.

The reason I haven’t yet implemented shortcodes are 2-fold.

• I don’t need them
• Using PHP, it is quite easy to use most of the work I have done with any other split testing software, either javascript or PHP based, for those with tin-foil hats (most affiliates using Adwords extensively?)

The PHP plugin I discovered rocks – I tweeted about it a couple of weeks ago.

Implementation for other theme frameworks isn’t immediately planned, and code after some more tweaking will be made available in various ways (for the geeks whose shoulders I had to climb on to get this far), and I will probably package things in some way for those less technically inclined looking  for a “solution, along with modified versions of various themes I have rights for.

He had a great article, he submitted it himself, and it flopped on Digg.

Doesn’t that happen to us all? Multiple Times?

But John didn’t give up, and applied some things he learned on other sites.

He then submitted the exact same article, but with different headline, and a new URL (changed page slug), to all the same sites.

The new version of the article has brought him almost 5000 visitors.