Archive for the 'analysis and optimization' Category

Real-time Context Targeting

Targeting local advertising using the real-time context of local searchers is an effective alternative that overcomes the limitations inherent in simply advertising a local business as a potential alternative based on the category and location of the search.  In this post, I’ll describe what real-time context is, how it can be discovered and how models can be trained to target local advertising using this approach.

This post is part of the Improving Local Ad Performance series.  In The Challenge of Fine Segmentation in Local Advertising I discussed the challenges inherent in relying solely on a ‘like for like’ targeting approach in local search.  In a like for like targeting, local advertising is displayed by matching the category and location of the advertising business to the category and location of a search request.  In today’s post, we are going to discuss using Real-time Context Targeting as a complementary alternative to this approach.

What is Real-Time Context?

Let’s use the example of someone searching for a Taxi to call.  They might be looking up a number for a particular Taxi, looking for any old Taxi or even looking for the nearest available Taxi.  The following diagram illustrates various situations where a user might be looking for a Taxi:

The Real Time Context of Someone Looking for a Taxi

The Real Time Context of Someone Looking for a Taxi

The actual context of the user is hidden from us, but let’s assume for the moment we actually know it (or could at least take an educated guess at it).  In that case, we could advertise based on that context.  So, for example, if they are call a Taxi because they’ve lost their keys, we could advertise the Locksmith they will also need!

Why Not Just Advertise Taxis?

There are several reasons to consider advertising something other than just Taxi services:

  1. Other types of advertisers (a Locksmith for example) might be willing to pay you far more for a lead.  So, even if your response rate ended up lower than for a Taxi ad you might make more money because you’ve delivered more valuable leads.
  2. Some users may be loyal to a particular brand of Taxi or using a service that provides the nearest available Taxi.  As such, they aren’t really open to substitution – but may be more more open to something that meets another need in their context.
  3. You might not have Taxi ads in that locality.

If the advertising is relevant to the real-time context of the caller, they’ll appreciate and perhaps act upon them.

How Do You Determine Context?

Sounds great – but how do we discover the hidden context of our searcher?  We have to infer the hidden context from characteristics (or attributes) of  the search and current events.  Examples include:

  • time and date based attributes (time of day, day of week, weekday/weekend, holiday, seasons, etc.);
  • place attributes (requested location and current location of the searcher if known); and
  • real-time events such as weather, sporting events and cultural events.

In fact, by examining past search history (in a completely anonymous way that protects the privacy of individuals) and the historical event record we can apply machine learning algorithms to build models that recommend the best ads based on current events and the attributes of a local search.  An example of how we do this at Predictabuy is shown below:

Real-time context targeting of local advertising

Real-time context targeting of local advertising

What’s in the Event Stream?

The explosion of social sites and mobile usage exemplified by Twitter and Facebook provides a rich and evolving set of events that a context-driven local targeting engine can exploit.  As such, this approach will just get richer and more effective over time.  If you have a suggestion for events you think will be important in local advertising leave a comment and let me know.  I’d love to hear from you.

Learn More

For a free consultation on how you can use Real-time Context Targeting in your business contact me.

This post is part of an ongoing series on Improving Local Ad Performance.  Upcoming posts will cover the use of consumer behavior and preferences in the effective targeting of local advertisements.  To ensure you don’t miss any of the discussion:

The Challenge of Fine Segmentation in Local Advertising

This post is the first in a series on Improving Local Ad Performance.  My perspective is primarily that of local information publishers and application providers.  In this post I lay out one of the fundamental challenges of local advertising: the need to finely segment it by geography and business type.  I’m going to focus on local searches, but the principles apply to other types of local information and advertising as well.

Local Searches: Sliced Very Thinly

Local searches are highly targeted.  This makes  makes them both an advertiser’s biggest fantasy and their greatest nightmare.  It presents a number of unique challenges.

As a minimum, local searches are distributed across many different geographies and types of businesses — and this is only the tip of the iceberg since you can also take in to account the user’s context, behavior and preferences.

Local Searches are Thinly Sliced

Local Searches are Thinly Sliced

These thin slices have implications for both publishers and advertisers:

  1. Each category/locality combination receives a small percentage of all traffic.  So publishers need large volumes at a national level or they have to specialize in particular geographies, vertical categories or demographics.
  2. These highly targeted searches are often most meaningful to small and medium sized businesses serving the niche but acquiring the necessary mass of these smaller advertisers is extremely challenging.
  3. National or even regional advertisers have to find ways to make campaigns truly meaningful at a hyper-targeted level.

Like for Like Targeting Alone Won’t Get You There

Most people approach targeting of local advertising by having the advertiser define the category and location they want to target.  Then, the advertisement is presented when a user performs a search in that category and location.  This frequently takes the form of offering the user alternatives to their request.  I’m going to call this ‘like for like’ targeting.

While easy to understand, this approach has a number of lmitations.

In high value categories, demand exceeds supply.  Businesses in categories like Locksmiths or Attorneys are often willing to pay a large fee for a lead.  Unfortunately, searches in these categories are rare, so while the inventory is very valuable and sells quickly and at a premium price – you just don’t have that much of it!  In fact, as the diagram below illustrates – the value of a category (from an advertising perspective) has no relationship to the volume of searches it experiences!

Search Volumes and Value by Category

Search Volumes and Value by Category

Unless you have a huge number of advertisers, for the (vast?) majority of local searches you won’t have a like for like match on the basis of category and location.  At least not one that’s truly relevant.  Providing a user with alternatives that are too far away or always providing them with the same small number of national advertisers undermines the credibility of advertising suggestions.

And the flip-side of the above, is that for many truly local advertisers you won’t have enough traffic to give them a meaningful set of leads.

Finally, in many local search use cases, users aren’t open to substitution.  A true category search – where a user is  open to suggestion and recommendation (and thus relevant advertising) is a relatively small – albeit very valuable – part of a publisher’s search inventory.  Instead, the most frequent use cases result from a user trying to complete a transaction with a business they’ve already selected.  They are most often looking for a phone number or directions.  In these cases it can be better to provide them with an ad that complements their current choice and context rather than trying to get them to substitute their choice.

Tackling the Like for Like Challenge

There are several possible – largely complementary – ways to approach this problem.  I’ll be exploring these options in some detail in future blog posts as part of the Improving Local Ad Performance Series.  Follow me on Twitter, subscribe with an RSS reader or subscribe by email so you don’t miss any of the series.  A quick summary of some of the approaches:

  1. In addition to like for like targeting, target local advertising based on context, behavior and preferences.  With appropriate analysis and targeting models it is possible to deliver relevant and complementary advertising in a way that results in a better match between available inventory and available advertising.
  2. Focus your efforts on being the ‘go to’ destination for the higher value ‘category’ or ‘research’ type searches – either broadly or within verticals.  Yelp is an example of a company that has done this by focusing on creating a community of reviewers making it a destination for people seeking opinions.  The advertising Yelp provides is primarily of the ‘like for like’ type – which is appropriate given that most people viewing review pages are in fact open to suggestion.
  3. Participate in some sort of exchange or market where you can buy traffic (i.e. by using AdWords for example) or gain access to advertisers (i.e. by working with a Yellow Page publisher for example).
  4. Focus your resources from both a publication and advertising perspective on specific verticals.

The Challenge Becomes Even More Acute in Mobile

Increasingly, local searches are occurring on mobile devices.  On the one hand, mobile devices offer the promise of even richer context information (where you are right now).  On the other hand, the more limited screen real-estate means that providing the most meaningful suggestions (or advertisements) becomes even more critical.

This post is part of a series on Improving Local Ad Performance.  To receive future installments you can:

Google’s Place Pages are Designed for Optimization

Google’s new Place Pages are designed for optimization which potentially makes them great landing pages.  Is Google positioning itself to simplify advertising for local businesses?

The downfall of most SEM offers to local merchants is that they deliver lots of clicks but few conversions.  That’s because too often nobody is optimizing the landing page (or has even defined what a conversion is).  Google has now put themselves in a position to address that by allowing the landing page to be optimized.  They could even have merchants use Google Voice if they want to optimize to receive calls.

What Does Designed for Optimization Mean

Recently I wrote in Picking Winners about the use of controlled experiments and A/B testing to optimize website performance.  Perhaps the most widely known application of this principle is the optimization of website landing pages using tools like Google Website Optimizer.

The basic idea in landing page optimization is to empirically test the performance of several different design options against some specified conversion goal.  For example, if your goal is to get people to ‘sign-up’ for something you’d test different page designs and see which one performed best.

If you want to do this easily – and especially if you want to do it using some automated process – you need to adopt a web page design that is amenable to such an approach.  Andrew Chen has written a great post on keeping the design consistent during A/B testing.  He says that the secret is to create an open design – and gives Amazon’s home page as a classic example.

Well, it turns out that Google’s Place Pages are another excellent example of open design that allows automated optimization.  Have a look at one of the example pages Google highlighted:

Google Place Page Showing Block Structure

Google Place Page Showing Block Structure

As shown above, the page is broken in to two columns and the content is organized in to various blocks.  This makes it easy for an automated process to vary both the placement and size of each of the blocks and the content shown within each block.  What’s more, you can select and optimize the look of the page based on where the traffic is coming from – varying the look and feel of the page based on how the user got there.  So, if you arrived at this page as a result of a search for ‘Tartine Bakery reviews’ the ‘review block’ might be much more prominently displayed.

The fact the pages are well suited for optimization doesn’t necessarily mean all that much.  Google is well known for being an A/B testing fanatic.  So, this may just reflect a desire to be able to more easily optimize how information is presented to users.

But it could also be a first step towards something more…

Could Google Try to Close The Optimization Gap?

Optimizing landing pages is a fairly well understood process.  Unfortunately, it’s a process that few smaller businesses have the time and expertise to perform.  So, it doesn’t get done.  And the end result is that small businesses don’t see the expected results from clicks and become discouraged.

But now Google has designed a landing page that it’s possible for a machine to optimize.

Imagine a tool that allows a local business to set up an Adwords campaign that automatically creates and tests landing pages.  The tool might suggest appropriate keyword alternatives along with appropriate landing pages and then start running the alternatives and select the combinations that deliver the best ROI.  All with minimal involvement from the business owner.  Google certainly has the scale and machine learning expertise to accomplish something like this.

What’s Missing

For one, Google would need local merchant’s to define some sort of ‘conversion event’.  This is conceptually as easy as defining a new ‘block type’ that will appear on the landing page and be optimized.  For example, a restaurant might view a phone call or an Open Table registration as a conversion event.  If it’s a phone call, I imagine the merchant could be encouraged to use Google Voice to provide a closed loop analysis of the conversion event.

Perhaps more likely than having individual merchants doing this (at least in all cases) would be a small army of SEO and SEM experts doing it on the businesses behalf – but within a closed looped system managed by Google.  Google could potentially create a whole new eco-system.

Updated (September 28, 2009): Lots of concern around a core issue of whether these pages are being indexed.  In fact, Google representatives have weighed in the comments on posts by Erin Schonfeld at Techcrunch, Greg Sterling and Mike Blumenthal.   Google is confirming that these new pages won’t be indexed directly, but they may be indexed if they are referred to by other sites.

They probably didn’t want to muddy the issue, but I couldn’t help but notice that they did NOT comment on my thesis about using these pages as landing pages!

Why I Like Adobe’s Purchase of Omniture

Ok, I know the folks at Adobe (yeah the Photoshop people) and Omniture (web analytics and optimization geeks) have been waiting for me to pronounce on their deal.  After mulling it over a bit, I’ve decided that it is good.

Judging from the twitter chatter, some found it perplexing.  And apparently the market didn’t much like it either.

I like it because it recognizes that the creation of web content should be done hand in hand with activities like Search Engine Optimization (SEO) and Search Engine Marketing (SEM) and the tools used to the analyze and measure the effectiveness of a web-site.  In today’s world, there are those who create websites and those who do SEO and SEM – and they are often blissfully unaware of each other.  That’s unfortunate given that the whole point of a website is to engage people and accomplish some commercial goal.

For a long time, the technical hurdles associated with the mechanics of creating a website have dominated the equation.  But that continues to get easier – as it should.  So rather than focus on the mechanics, you can naturally expect people to start thinking more about how to use a website as a truly effective tool.  Which should lead you to think about how you are going to structure and evaluate the content.

So, I can see Adobe creating entire new classes of tools where the very way you think about and create web pages becomes much more oriented towards optimizing and measuring the content.  For example, you might have a tool where the first thing you do for a new page is define the ‘objective’ for it (i.e. this page is intended to get people to register for the site).   From this objective you would then have tools that would advise you on ‘best practices’ for achieving it.  You would design the page from the get go to evaluate several alternatives.  The code needed to manage this would just disappear in to the woodwork.  The tools required to manage the revision of various content elements are part of the tool-set.

And there is potentially a very nice network effect.  The more people use your tools to create pages and analyze them the more data you (can potentially) collect on what works and what doesn’t.  This means you can do a better job of pro-actively advising people on what they should and shouldn’t do.  This allows the creative people to spend more time exploring new things that might work rather than wasting their time on things that are pretty unlikely to work.

Think of it as ‘objective driven design’.  I’m gonna let Adobe use that phrase if they like.

Of course, all that’s easy when you say it fast.  And difficult to execute in practice.  And I’ve glossed over at least one really important point.  What one means by an ‘effective’ web-site is a moving target.  Changes like social media and real-time media – not to mention changes in what people expect or want – mean that the very definition of ‘well designed’ is always shifting.

But that just makes it an interesting problem worth tackling.  Time will tell.

Picking Winners

Web applications allow us to quickly try out new features, presentations and approaches.  But people are terrible at predicting which changes are beneficial and which ones are neutral or even harmful.  That’s one reason why a systematic approach to the analysis and optimization of changes through controlled experimentation is important.

At it’s simplest, controlled experimentation is just trying different approaches to a problem (which can be as simple as the color used on a web page) and measuring how user’s respond to these changes.

The paper “Online Experimentation at Microsoft” (PDF) was presented at KDD 2009 and provides a great overview with many concrete examples of actual experiments run at Microsoft.  Here’s one example:

The MSN Real Estate site (http://realestate.msn.com) wanted to test different designs for their “Find a home” widget. Visitors to this widget were sent to Microsoft partner sites from which MSN Real estate earns a referral fee. Six different designs, including the incumbent, were tested.
treatmentsA “contest” was run by Zaaz, the company that built the creative  designs, prior to running an experiment with each person guessing  which variant will win.  Only three out of 21 people guessed the winner…

The winner, Treatment 5, increased revenues from referrals by almost 10% (due to increased clickthrough).

In general, the paper documents that even experienced experts can only pick the winners less than 1/3 of the time.  Meaning, the other 2/3 of the time they are recommending changes that are at best neutral or at worst actually harmful.

This is a non-technical paper that provides motivation for taking an experimental approach.  They also describe the many cultural barriers they encountered at Microsoft.  Overall, a very good read.  Highly recommended.

Of course, they recommend a very sophisticated approach.  But the same principles apply in a broad range of situations.  One common activity that falls in to this category is the optimization of landing pages for SEM and SEO efforts.  In these situations you are usually assisted by tools that make it easy to get the statistical analysis right.

The take home message is that successful companies are learning how to fail fast forward rather than getting stuck in endless rounds of paralysis and internal arguments.  Real-world experimentation can be the final arbiter.

via Greg Linden


Twitter Updates

Archives

June 2017
M T W T F S S
« Jul    
 1234
567891011
12131415161718
19202122232425
2627282930