Archive for February, 2012

Auctions Beta Site – let the bidding begin

We’re launching the Members Auction site in Beta to get your feedback.  We want to know what you think, the good the bad and the bugs.

Take a look around the site and try bidding for the items, they’re real auctions after all. If you spot something that you think could be better or you have found a bug, drop us a note at mcfcgeeks@mcfc.o.uk or reply to this post. Include your browser so we can recreate what you see.

So what is Auctions all about??

The Auctions websites is where you can bid on amazing money can’t buy experiences, tickets or uniquely Blue items, but here’s the best bit… it’s free – there is no cash involved in bidding.

Who can take part in Auctions?

As a member, you will have acquired a certain number of auction credits, based on your spending with Manchester City since the start of the season.  These credits are what you bid with.

Open to all members– Whether you are Superbia, Platinum, Gold or a Blue member – you can take part now in the auctions beta web site.

If you’re not a member and you would like to take part – it’s just £10 to become a Blue member.

There are a few things worth noting about the City auctions web site.

Auction credits should not be confused with loyalty points – they are entirely separate.

Your credits can be spent on the auction site, whereas your loyalty points will continue to grow over the seasons as they always have done – untouched.

When you sign in, using your normal log in details, Supporter Number and password, the number of auction credits you have earned will be displayed. There’s a list of them in the My Account section too.

If you need to change your email or password you can do so via mcfc.co.uk or mcfc.co.uk/tickets

Have a look around the site – see which lots currently available you are interested in – or see what’s coming up soon if there’s something you want to save your auction credits for.

There are two types of auctions. The highest bid and the lowest unique bid.

For a highest bid auction – the supporter who bids the most auction credits by the time the auction closes, will win the prize. For the winner, the number of auction credits they have bid will then be deducted from your total. Similar to ebay.

We will then contact you with any further details necessary to collect your prize.

For the lowest unique bid – simply guess the number you think will the lowest unique number out of everyone who takes part.   It will cost you a set number of auction credits to make guess, regardless of whether you win or not.

When the auction closes, the supporter who guesses the lowest number, which no one else has also guessed, will win the prize!

The auctions site unfortunately is not open to our supporters under 18 – however, prizes are transferable so look out for prizes specifically for the juniors in your family.

Most of all – the auctions web site is just for fun.

If you would like to know more about the techy side of the development, we will follow up with a more detailed post once all the feedback is in.

We hope you like it!

The results are in. Or are they? Search.

As part of ongoing enhancements to the site over the last few months, we identified search as a key function with room for improvement.

Ever since the site launched in 2009, and until just before Christmas, the mechanic behind our site search is Lucene, an out of the box solution configured to display results by “Relevancy and Recency” across all of our content. On the face of it, a logical mechanic. But what are users searching for?

The lion’s share of top ten searches consists of players, and the “Relevancy and Recency” solution failed to deliver relevant results. Why?

A typical example.

A search term such as “Micah Richards”, would likely return results in the following hierarchy,

  1. Other players profiles
  2. Non related articles/videos (and plenty of them)
  3. Micah Richard’s player profile

Why would the Micah Richards player profile appear 3rd in the hierarchy of results? For a number of reasons.

Volume, we’ve posted thousands of news and video items on the site since 2009, and that’s on top of merging content from 2005 onwards from the previous site.

How is relevancy and recency defined?

Relevancy is the number of times a search item is found within the title, summary and body copy of piece of content posted on the site, all of which have equal value.

Relevancy also discounts the tags associated to a piece of content.

Recency, is simply the latest content item.

Based on this definition, the algorithm created the following chain of events…

Dear Recency,

I’m pleased to inform you that I have 16 mentions of “Micah Richards” across my article’s title, summary and body copy.

I hope that see you this fit and relevant, and prioritise my response above the lions share of other Micah Richards content items.

All the Best

Relevancy

reply – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – -

Dear Relevancy,

Hang on a minute, there you go again assuming you should come top of the tree…

What you fail to understand is that you were posted on the site in January 2011.

Since then, I’ve been inundated with articles that mention you once or twice, hundreds of them…

I don’t care if the article is a match report, or another players profile that refers to you in the body copy, or indeed a news item, that’s exclusively about something else that only mentions you once.

At the end of the day, they’re more recent, and must be prioritised above you.

No hard feelings.

Recency

Something needed to be done.

We want to return relevant and recent search results, particularly around the most popular searches. For that, we required a solution that would fix the short-term issues immediately, but with long term flexibility.

Our creative and development agency Aqueduct proposed that it would be possible to improve search without having to strip out and integrate a whole new service.

Searching for Balotelli

New search using "balotelli"

How could we build upon the current system?

1) Automated weight relevancy

We could apply weighted relevancy on article template fields (behind the scenes) so this would not be noticed by editors. The hierarchy of this relevancy would “score items” in this way.

  1. Title
  2. Summary
  3. Links in content (if relevant to the term searched for)
  4. Term in content

*note the enhancement here, beforehand, a,b,c,d carried equal weighting.

Should a user perform a search that has the search term in the Title, this will weigh higher than a content item with the term in the Summary.

Should a search term be contained within Title and Summary, then this will have a combined weighted ratio and thus put the article higher than if the term was just being used in the title.

On this basis, the user would be presented an article item at the top of their search results, with a date that is in fact older than the second placed article. The reason for this would be the search term was contained in both title and summary.

2)     Link terms to a taxonomy

We would also have a football taxonomy that we can link to and instruct the the algorithm through code and simple logic. This taxonomy would be based upon terms that users may logically search for, and where that item would be stored within the site category.

As an example, if a user searched for something containing the word “goal(s)” logically the content item would more than likely be within:

  1. Match Report
  2. Matchday Centre
  3. Player Profile
  4. Team News

This would instruct the algorithm to first look into the above 4 categories as a priority then, to scan the rest of the site.

3) Presenting Search Results

Search results have always been “jumbled”, resulting in a scenario whereby if a user runs a search, the date-ordered results are displaying as unfiltered content types.

We would introduce additional filters to allow users to re-order search results from ‘content types’ into ‘date order’

4)     City Recommends

Essentially a pre-defined selection of results based on the users search term, touching on the manual manipulation option. Think Google sponsored ads, this is an ideal piece of functionality to present popular content or partner offers to our users.  Try searching “Dzeko” or “Silva” on the site.

Why choose this option?

  1. No overhead to content editors
  2. Minor administrational needs (for the taxonomy)
  3. Ability to fully control the taxonomy and how it weights the category and terms
  4. Not re-inventing the entire search system
  5. Easy to enhance and build upon moving forward

What’s up next for search?

  1. We need to integrate recency into the search results by tweaking the automated weight relevancy
  2. We need to apply the taxonomy to key search terms in order to boost relevancy when viewing “all” returned results.
We’ll keep you posted.

John Kearney


About this blog

This is the official blog of the digital and media team at MCFC.

MCFC Geeks Official Twitter

  • You may have noticed some gremlins with our online & mobile match coverage. This was due to not receiving correct Opta data. Sorry. 1 week ago

Follow

Get every new post delivered to your Inbox.

Join 8,689 other followers