Home / 2012 / Archives for April 2012

AdWords Expanding Phrase and Exact Matching Technology

4/18/2012 by 1 Comment

AdWords has started releasing a feature that will expand the matching technology for Phrase and Exact Match Types.  Essentially, Phrase and Exact Match Types will begin matching closely related search terms, when Google’s technology is able to establish that the search intent is the same.  The opt-out functionality is live today; All Advertisers are auto-opted into this feature as a new Advanced Campaign Setting.  The matching technology will not actually change until mid-May.  Advertisers can begin preparing now, and opt-out now if they choose to.

My Take: This is a classic win-win-win.  Users get better results, Advertisers expand coverage with good performance, and Google increases the relevance of their search results and ads.  Of course, they are most likely also increasing keyword competition, i.e.: profitability for the auctioneer.

AdWords Opt Out Settings

Eric Enge and I had the opportunity to catch up with Jen Huang, a Product Manager on the Google AdWords team responsible for the new matching technology.  Ms. Huang filled us in on some of the details of what exactly is changing.

"Jen Huang: The technology is attempting to expand Advertiser Keyword coverage for Phrase and Exact Match when we can match to the same user intent. Keywords that reflect similar user intent."

Background

Search Engines use matching technology to match a limited set of Advertiser Keywords to an infinite set of possible search queries.  Match Types help Search Engines match Keywords beyond their literal, character-for-character match, to all of those possible search queries.  While Phrase and Exact Match are as close as Advertisers can get to literal control, Search Engines also use normalization (or canonical form). That is a broader topic, which I covered previously on Search Engine Land with this article – Canonical Form: The Hidden Keywords in Paid Search.  Normalization has always normalized Phrase and Exact Match on capitalization, for example. [nasa] has always matched [NASA], before and after this release. This release takes that further, to closely related variants, based on the notion of user intent.

When Search Engines match various keywords, they have to play a balancing act.  If they can maintain relevance and also expand the coverage, then they increase competition and profit as a result.  However, if they expand without preserving relevance, then Advertisers would notice the decline in value, competition would decrease, and falling profitability would soon follow.

Interestingly, Broad Match Modifier already uses a similar version of this technology.  This AdWords Help Center Topic, states:

Each word preceded by a + has to appear in your potential customer’s search exactly or as a close variant. Close variants include misspellings, singular/plural forms, abbreviations and acronyms, and stemmings (like "floor" and "flooring"). However, synonyms (like "quick" and "fast") and related searches (like "flowers" and "tulips") are not considered close variants.

"Jen Huang: This new matching technology is similar to Normalization, and to the technology that finds close variants in Broad Match Modifier.  The whole point is to match to keywords with the same intent, helping users find what they are looking for, and Advertisers serve more ads."

I remember when broad-match was introduced with AdWords. It was not originally received well.  Advertisers found the technology was often way too far-reaching, and lacked relevance.  This was long before search query reports, when we actually had to do real work to parse server logs, extract the query string from the referrer_URL field… but I digress.  Most recently, AdWords added Broad Match Modifier, and adCenter is rumored to be following suit soon.

Long term, I am eagerly anticipating this shift towards intent-based advertising.  It will be an important shift that should enrich the user experience, provide some welcome simplification for Advertisers, and hopefully attract more direct-marketing advertising spend in an industry that has historically been dominated by branding and other non-ROI-focused advertising spend.

When Is This Happening?

The opt-out feature is live as of today, at the Campaign Level, giving Advertisers an opportunity to opt-out, or otherwise prepare.  The matching technology is planned to go live starting sometime in Mid-May, presumably in a rolling release.

Expected Impact

AdWords tested this feature in the marketplace with various Advertisers, across numerous verticals.  On average, with accounts already receiving greater than 33% of their traffic from Exact and Phrase Match Types, Advertisers generated 3% more clicks.  Performance metrics such as CTR, Conversion Rate, and Cost Per Acquisition were consistent.

In practice, we can expect there to be some variability between industries, and certainly amongst individual Advertisers.  With our clients, we will be advising that we take this one cautiously, and monitor performance closely.

Another impact for Direct Advertisers would be the ability to compress conversion data into one tracked keyword, while matching for more traffic on closely related searches.  This might be especially interesting for Advertisers with sparse conversion data.  On the other hand, Advertisers with a surplus of Conversion data, and the tools to support it, might choose to stay with the old technology, and the precision it allows.

What is Changing, Exactly?

Pardon the pun…

The technology uses the Google.com search engine to map similar keywords, and will generate a match along one of the following manipulations:

  • Spelling Correction
  • Word Stemming
  • Plurals
  • Acronyms
  • Abbreviations
  • Accents

It is worth noting that nothing else is changing as a direct result of this release.  In particular, Negatives remain unchanged.  They still get minimal normalization, and will continue to work with the same level of precision they do today.

Spelling Correction

This will work similarly to Google.com’s auto spell correction feature:

Spelling Corrections in Google

Essentially, Advertisers should not have to invest extra time in misspelling keywords.  Hopefully, it recognizes fat-thumb spelling errors from smart phones and helps get relevant searchers pointed in the right direction.

Word Stemming

This will match a root word with its various prefixes and
suffixes.  [snare] might match "snared", "ensare" etc.

Word Stemming Example

This one has more potential to conflict with the Advertiser’s intention when advertising on a word.

Plurals

For example, "car" will match "cars". Advertisers interested in distinguishing between singular and plural versions of their keywords might consider opting out.

Acronyms

For example, "NYC" and "New York City"  might be considered equivalent in this new technology.  Watch out for acronyms that might not work as intended, such as the state Abbreviations for Indiana (IN), Delaware (DE), Los Angeles (LA), etc.

Abbreviations

For example, "abbrev." might match with
"abbreviation", and vice versa.

Accents

This shouldn’t impact US-English traffic as much as it might Canadian traffic, or US-Hispanic traffic.  Of course, internationally this should provide some interesting results.

How Can We Measure the Impact?

As Advertisers, our ability to measure the impact will for the most part be limited to a "before" and "after" analysis.  We should be looking for an increase in traffic from Phrase and Exact Match
Keywords, with steady or tolerable performance metrics like CTR, CPC, ROI, etc.

My inner Excel Nerd is pondering if a more detailed analysis might work? Could we pull a Search Query Report? It might tell us the Keyword and the various expanded Search Queries? The current ones do not – they show us the normalized Search Query for Exact and Phrase. We may have to wait and see about that one…

Quality Score

Quality Score will continue to be calculated based on the performance of the original exact match keyword. Any close variants added by this new expanded matching technology will not impact Quality Score for keyword that triggered the ad.

Improving exact match and phrase match

Search Query Reports

Close variants matched by this technology will shows as a new derived Match Type &quote;Exact Match (close variants)" in the Search Query reports, once the technology is live.

Who Should Avoid It?

Advertisers who measure a decline in performance, for starters.  Additionally, Advertisers who use Brand terms with very different bids or ads, especially Brands that are intentional misspellings.  In this case, Advertisers may find the precision offered by the older technology allows them to maintain the control they need to treat Branded versus Non-Branded terms appropriately.  Likewise, any Advertiser who derives enough value from the precision of the old technology to offset the opportunity in the new version, should opt-out.

Summary

The opt-out feature is live as of today as an advanced Campaign Setting (you are auto-opted-in already), the matching technology changes in Mid-May.  You should expect good results, but I recommend keeping a close eye on things, as every Advertiser, and every Account, is different.

Filed Under: PPC, Search Engine Marketing, Uncategorized Tagged With:

Rel=Author Defined with Google’s Sagar Kamdar

4/11/2012 by 10 Comments

photo of Sagar Kamdar

Key Points from Interview with Sagar Kamdar

This interview transcript is from the March 13th, 2012 conversation I had with Sagar Kamdar, who heads up the authorship program at Google. This provided some simple, straightforward clarity on how it works, and some of the various scenarios for implementation. Here are some of the key points from the interview:

  1. Google sees the authorship program as a way to connect authors and readers who are a fan of their work.
  2. The original 3 link method is still supported. This is documented in detail in the main interview, along with a graphic image of how it works.
  3. They came up with the 2 newest methods because people were having trouble executing the 3 link method, in some cases because it was too confusing.
  4. The 2 link method only requires that the content on the site link to the author’s Google profile with a ?rel=author parameter, and the Google profile lists the site in the “Contributor To” section of the profile. This is documented in detail in the main interview, along with a graphic image of how it works.
  5. The email method only requires that your email address be verified and use the domain name of the site where the article is published, and that the email be included in the attribution for the article. This is documented in detail in the main interview, along with a graphic image of how it works.
  6. The email method was created because some authors do not have the ability to add a link as specified in the 2 link method.
  7. It is OK to put sites in the Contributor To section (provided you contribute to them) even if you don’t get rel=author links back.
  8. The Authorship Request functionality has been retired, though there is a simple form there now for verifying an email for the Email method.
  9. Anyone who properly implements rel=author is eligible for participation. It only requires that you have a high quality photo in your Google profile.
  10. When you first show up depends primarily on crawling and indexing time. This may take days or weeks depending on your site.

Full Interview Transcript

Eric Enge: Can you provide us an overview of your background and your role at Google?

Sagar Kamdar: Sure, I am a product manager on the search team. I have been at Google for about four years, spent most of my time on search, spent my time on a variety of projects like real-time search, and webmaster tools. Actually, my first project was external evangelism with search engine specialists and most recently I have been focused on authorship and some of our social search efforts that you have seen out in the wild.

Eric Enge: There has been this longstanding problem of people who are the original author of content not necessarily being the first to show up for their content, and I think that it is really great that we are starting to have a method by which authors are identified. Can you provide some thoughts on what the authorship program is all about?

The main thing that we are trying to address is the faceless nature of the web.

Sagar Kamdar: The main thing that we are trying to address is the faceless nature of the web. For many years people have been clicking on content not knowing who created it, and not knowing who commented on it. What we are seeing is that users really want to know who created that bit of content. Users know who their favorite authors are, and we’re trying to make it easy for them to communicate with those author(s).

We want to make it easy for authors to get setup and then for searchers to be able to find the content they create. So if they search for “iPad” and they like David Pogue, they are more likely to see what David Pogue has written about the iPad. Authors benefit as well because they get attributed to their content and also they can engage with the users in a way they haven’t before.

Eric Enge: I think it is particularly interesting when you look at it because many well-known authors are published on many different sites.

Now we have this feature where users can say show me all content by Ben Parr, for example.

Sagar Kamdar: It is actually one of the coolest features we have for authors. We have all these people that are publishing on five or six different publications and their own blog and users can’t keep track of all that content. Now we have this feature where users can say show me all content by Ben Parr, for example.

Editor: This is how it looks in Google’s search results:

More Articles by Ben Parr

Editor: Here is the Author Page for Ben Parr on Google:

Ben Parr Google Author Page

Continue Reading…

Filed Under: Google, Search Engine Algorithms, SEO, Social Web

The Future of Search and Social Integration, Interview With Andrew Shuman

4/5/2012 by 2 Comments

photo of Andrew Shuman

Key Points from Interview with Andrew Shuman

I had a fascinating chat with Andrew Shuman about the adCenter UI and the integration of search and social media. He provided a lot of great insights on what some of the opportunities are, and some of the challenges as well. Here are some of the other key points from the interview:

  1. re: search and social integration: “There is also a very interesting scale problem there. As with traditional search, we have to query a massive numbers of servers, but then we also have to take my social graph and overlay that on the search results, modify them, and return this all in milliseconds.”
  2. “You also have the case where all the people in Seattle seem to be Liking this document. That’s less about me (as a searcher) specifically. Some of those micro populations become very interesting.”
  3. re: the deeper meaning behind a particular search query: “I am not just searching for a restaurant, but I am planning a whole night, and what those things mean for a user, how we can bring together the data in a related fashion.”
  4. “A restaurant is related to movie theatres that are nearby, right. This creates a new kind of linkage between objects that goes beyond the link graph or the social graph.”
  5. “It would be really nice that as you just get to restaurant intent, the related intents are there in part of the experience so that as I mouse over or hover over or whatever we come up with it’s probably just another tab then.”
  6. “Related searches is something where we think we can really experiment a lot more with the placement. For some searches it may make sense to be more aggressive in the UI with that, as it may be more common for the user to perform follow on searches.”
  7. “You also have to differentiate a person who does the same search frequently, such as they like to search on Bellevue restaurants a lot. Perhaps the third result is always the restaurant they go to.”
  8. re: the strength of social as a signal: “It’s an interesting challenge though, because the more generic signals across the whole web are a much stronger signal. You have billions of clicks versus a hundred friends on Facebook – there is a different science involved in that.”
  9. re: how varied people’s intents are: “… not everyone comes to Bing with a specific task in mind, sometimes they are in an exploration mode …”

Full Interview Transcript

Eric Enge: Can you provide a little background on yourself?

Andrew Shuman: I’ve been in Microsoft about nineteen years. I started in Outlook as a developer, and also spent some time with the MSN team. But, now I am at Bing, and it’s been by far my favorite job in Microsoft.

It’s a pretty unique place where you get to work with so many people of varied background in terms of linguistics, and statistics, and math, as well as computer science and that’s fairly fascinating. But then, the problems faced are so rich when you think about the fact that you can type in a three word search query, and we will show you several results.

But, from a user interface point of view and an application point of view, we are so far away from the joy of a traditional library where you get a sense of the volume of data around you and a sense of where you are within the system.

To help with that, we are constantly looking at different UI models; things like speech and touch that help create the feeling.

Eric Enge: Can you talk a little bit about the challenges of integrating social data and search?

… just because a friend of mine liked a restaurant doesn’t necessarily make it more relevant to me.

Andrew Shuman: One of the interesting areas I was thinking about a little bit this morning is the challenges of tying together the social signal with the other signals, and that it is a very interesting scale problem for us. It obviously is a very different signal to the relevance engine and to the user just because a friend of mine liked a restaurant doesn’t necessarily make it more relevant to me.

Facebook Likes in Bing SERPs

Continue Reading…

Filed Under: Microsoft, PPC