The broad modifier feature was rolled out in July 2010 in the U.S and was mostly aimed at scaling up those accounts not already using regular broad match type. Search marketers have to then build new broad keywords using “+” signs to effectively unlock the broad modifier feature.

Because it requires some time and effort, not all advertisers have actually implemented this feature – particularly those advertisers already using broad match type. Some of our clients reluctant to use standard broad match type did test broad modifier, and it turned out that it performed surprisingly well – with an incremental revenue volume up to 15% while maintaining efficiency on target.

The new matching behavior for exact and phrase match types announced on April 17th  (and just rolled out last week) is going one step further since it potentially impacts all advertisers by automatically updating the default matching behavior from standard exact and phrase matching to a more lenient matching behavior including plurals, misspellings, and other close variants.

In that sense, it can be seen as a logical sequel to the broad modifier feature. Advertisers have the option to opt out – however, most of them will allow the update to occur.

There has been a lot of speculation and skepticism about the latter one, as it seems it is just another way for Google to generate more ad revenue. Hence the question: what are the first takeaways a couple of days after the new matching behavior roll-out?

Where Can I See The Impact In AdWords?

While I couldn’t find any details in AdWords about those incremental queries corresponding to “plurals, misspellings and other close variants” in AdWords, you can now see that the “Other search terms” section shows impressions and clicks even for exact keywords:

In this particular case (a strong trademark keyword in exact match), we have measured an impressive 18% lift in impressions at a stable CTR. Conversions did follow with a slightly higher conversion rate and a slightly lower cost per order – not significantly though.

Overall Impact On Traffic & Conversion Volume

In this section, I will attempt to answer two questions: what is the average impact on traffic? And what is the average impact on conversion volume?

According to Google: “on average, the new matching behavior increased AdWords search clicks by 3%, with comparable CPCs”. Looking at 15 top brands managed through eSearchVision’s proprietary search query report, early findings show that clicks from queries not containing the actual keywords actually increased by 3.5/4.5% on average with comparable CPCs.

More specifically, the below graph shows that the percentage of clicks generated through exact and phrase without close variants has slightly decreased, while the percentage of “All Other Queries”, i.e. when the keyword is not included in the query, has increased from 23% on average the week before the roll-out up to roughly 26-29% the week after.

In the meantime, conversions have increased by 2.0/3.0% on average. However we can expect this number to go slightly up over time due to the post-click effect – since users who clicked over the last couple of days are likely to convert in the next couple of days or even weeks.

Main Takeaway

The new exact and phrase matching behavior seems relatively promising so far. As expected (and even a bit more than expected) we’re seeing more impressions and clicks at a stable CPC, as well as incremental conversions, even though conversions are not following as fast as the clicks for the time being.

As a result, search marketers no longer need to create additional keywords using the broad modifier feature. The new matching behavior seems to be doing pretty much the same job with no effort. Just keep an eye on search query performance and maybe add more negative exact and phrase keywords if you identify any poor performing or irrelevant queries since this roll-out.

However, a high click-through rate does not assure a high conversion rate. The two can even have an inverse relationship: an ad copy geared towards curiosity clicks will result in fewer conversions, percentage-wise, than an ad copy geared towards qualified clicks.

Keeping An Eye On CTR Helps Maintain Conversion Volume

Focusing on conversion rate optimization and neglecting CTR will keep you from scaling up your paid search program. You need to maintain a high traffic volume to maximize the number of conversion opportunities. If you don’t do so, you’ll end up with great conversion rates but low conversion volume overall.

A better way to go about it is to maximize the conversion rate within a certain CTR range, thereby keeping a good rank at a decent cost per click and maintaining traffic volume.

Even excluding the effect of CTR on Quality Score (and thus CPC), it is important to find the right balance between an appealing ad copy which generates lots of clicks, and a more conversion-oriented ad copy which generates fewer clicks at a higher conversion rate. The Quality Score factor makes it even more important to optimize the CTR as it helps mitigate the average cost per click and lowers the cost per acquisition (CPA) as a result.

The best performing ad copies are rarely those with the highest CTR or the highest conversion rates, but typically somewhere in the middle for both metrics. They are appealing without being deceptive or too generic. Best-performing ad copies usually have a strong CTR and an ok conversion rate – as opposed to an ok CTR and a strong conversion rate – because of the Quality Score factor and its consequences on CPC.

Beyond CTR & Conversion Rate: Maximizing Profit Margin

In the following chart (inspired from actual data) ad #1 has the lowest CTR and the highest conversion rate, while ad #10 has the highest CTR and the lowest conversion rate. Again, these metrics do not always have an inverse relationship, but it is the case very often.

If you are still unconvinced, you can give it a try for yourself with two ads in rotation – the first one clearly geared towards curiosity clicks, the second one clearly geared towards qualified clicks – and this logic should be verified.To get back to our sample data below, it is worth noting that neither ads #1 nor #10 are the best-performing ones overall.

If you look at the conversion volume, ad #6 is doing the best without having the strongest CTR, nor the strongest conversion rate.

If you look at the CPA, ad #8 has the lowest CPA. Ultimately, ad #6 performs best with respect to the effective profit margin.

In a nutshell, one should keep an eye on the CTR when testing ad copies in order to maintain decent traffic volume and maximize overall profit margin. The right balance might take time to find, but the important thing to keep in mind is that the best ads in terms of CTR or conversion rate are most likely not the best-performing ads out there. There is a middle ground where profit margin can be maximized.

Also, while Quality Score is only available at the keyword level (and unfortunately not at the ad level) it does make sense to monitor this metric when testing multiple ad copies geared toward qualified clicks rather than curiosity clicks.

Google engineers designed their paid search algorithm in such a way that the inverse relationship between CTR and CPC forces search marketers into coming up with ads which don’t necessarily generate the most revenue per visit, but the most clicks. It is mostly about relevance to the end users, Google says – where relevance is defined based on clicks, whatever their engagement level is.

• How to organize a paid search program to reflect your site map and product offerings
• How to maximize quality score
• How to categorize queries by user intent (informational, navigational, transactional)

Yet, I have not found any literature addressing how to put those three concepts together.

In a world where more and more campaign management solutions allow search marketers to allocate revenue across impressions and clicks involved in a conversion, it makes sense to organize a paid search program to fully leverage attribution modeling.

With the right structure, you can fully measure the interaction between various sections of a website, as well as the interaction between informational, navigational, and transactional queries; as well as branded and non branded keywords to get a good understanding of consumer behavior and brand awareness.

Taking a step back, user intent can be broken into three categories:

• Informational queries look for a specific fact or topic, and account for roughly 80% of all search queries
• Navigational queries involve locating a specific website, and account for roughly 10% of all search queries
• Transactional queries look for information related to buying a specific product or service and also account for roughly 10% of all search queries

More details about user intent can be found in “Determining the informational, navigational, and transactional intent of Web queries” by Bernard J. Jansen, Danielle L. Booth, and Amanda Spink.

For the purposes of this article, let’s break down a site map into three levels (there could fewer or more levels, but the logic remains the same):

• Category: for instance “office supplies” and “breakroom supplies” are category-level keywords.
• Sub-Category: for instance “binders” and “clipboards” are sub-category keywords within the “office supplies” category.
• Product: for instance “ideastream storage clipboard” is a product-level keyword within the “clipboards” sub-category

It makes sense to differentiate Brand vs. Non-Brand for several reasons: to maximize your impression share, monitor specific KPIs, implement a specific bidding strategy, use sitelink ad extensions, etc…

As a result, with the opportunity to build a paid search program from scratch based on current knowledge and tracking tools it would make sense to organize your paid search campaigns into all possible combinations of site map levels and brand types.

There are 18 campaign types:

1. Non-Brand | Category | Informational
2. Non-Brand | Category | Navigational
3. Non-Brand | Category | Transactional
4. Non-Brand | Sub-Category | Informational
5. Non-Brand | Sub-Category | Navigational
6. Non-Brand | Sub-Category | Transactional
7. Non-Brand | Product | Informational
8. Non-Brand | Product | Navigational
9. Non-Brand | Product | Transactional
10. Brand | Category | Informational
11. Brand | Category | Navigational
12. Brand | Category | Transactional
13. Brand | Sub-Category | Informational
14. Brand | Sub-Category | Navigational
15. Brand | Sub-Category | Transactional
16. Brand | Product | Informational
17. Brand | Product | Navigational
18. Brand | Product | Transactional

Why such a complex structure? What are the benefits of such a structure?

Clarity

The above structure is pretty self-explanatory – you can guess where a given keyword should reside by answering three simple questions: Does it contain a brand term? Does it contain a category or sub-category or product name? Does it contain an informational or navigational or transactional term?

Quality Score Optimization

You can write more specific ad copy based on user intent. It is a valid test to rotate informational ad copy for informational searches, and so on. It can only help the CTR and the Quality Score – while it might hurt the conversion rate. The right balance can be determined via a proper A/B testing procedure.

Regardless, this structure should help determine relevant keywords/ad copy combinations, landing pages at every stage of the search process/buying cycle.

Search Funnel Analysis

A clear structure helps effectively measure all interaction at the campaign level – making it much easier for all major tracking solutions to provide granular insights.

More specifically, we expect informational searches – where most of the search volume sits – to be entry points in the conversion funnel, while navigational and transactional searches are more likely to convert.

Whatever you are expecting or assuming, this logic will help measure what is actually happening in your account.

Budget Allocation & Impression Share Optimization

A clean structure makes it easy to ensure brand campaigns have unlimited budgets and a high impression share – or at least a high exact match impression share, see this article for more information about impression share data.

Also, you can temporarily limit the daily budgets or even pause “Non-Brand | Informational” campaigns if your budget is limited, or if these campaigns are not meeting efficiency goals.

Site Link Ad Extensions

Relevant sitelink extensions can be built across all category and sub-category campaigns (not so much for product campaigns) based on top-selling products, top-rated products, current promotions, and so on.

This structure is theoretical and should be made more specific to every advertiser’s strategy / website / product offering.

For example, you might want to break down this structure even further by platform (desktop, mobile, tablet), or by geo, or both. The point is that your paid search program should be organized in a systematic manner allowing clear and concise insight into the consumer buying cycle and allow easy optimization based on consumer behavior.

I like thinking of a SEM program as a whole, or more specifically as a chain. Just like a computer is as fast as its weakest component, a SEM program is as strong as its weakest link. This holistic approach is actually pretty straight-forward and originates from the user experience itself – from the search query to the landing page, see graph #1 below.

As opposed to analyzing multiple indicators independently, the purpose of this article is to emphasize the relationships between those indicators.

While a process is typically a chronological series of steps, a chain is rather a set of links which simultaneously interact with each other and where every link matters. The point here is that search marketers should manage their SEM programs as a chain, and not as a process.

Also, one can measure those interactions within the chain – as defined above – using different performance indicators, see graph #2 below.

Query vs. Keyword Indicators

While the total number of keywords in your SEM account is a fairly good indicator of the development of your account, a more granular way to evaluate your keyword list is to look at the percentage of impressions in exact match type. Basically ,you want your keyword list to effectively match relevant search queries which are related to your offering.

In the long run, most impressions should occur in exact match type for maximum control and relevance. Other match types such as phrase, modified broad, and broad should be active only to explore new queries and feed your list of exact and negative keywords.

Also, impression share (IS) data at the campaign/ad group levels can be compared with the exact match impression share (Exact IS). While it is practically impossible to reach an IS of 100% due to poor ranks in phrase/broad match types, it is definitely possible to reach an Exact IS of 100%.

This means that search marketers should look at this indicator to ensure maximum visibility for their exact keyword list, as opposed to less qualitative queries associated with phrase/broad match types.

Keyword vs. Ad Copy Indicators

Most advertisers think of the Quality Score (QS) as a keyword level only metric, while the QS (in both AdWords and adCenter) actually reflects the whole user experience from the user query to the landing page – with a strong focus on the click through rate (CTR).

Interestingly enough, the Yahoo Quality Index was an ad copy level metric, which makes a good case for analyze keyword/ad copy performance as a pair. The bottom line is that it is quite irrelevant to analyze keyword performance without looking at the associated ad copies.

Ad Copy vs. Landing Page Indicators

The one metric search marketers usually look at is the conversion rate, i.e. the number of conversions divided by the number of clicks. While this metric makes sense in most cases, it is even more relevant to analyze the impact of the CTR on the conversion rate, and vice versa.

To clarify, you might have ad copies with a high CTR/QS and a low conversion rate, or ad copies with a low CTR/QS and a high conversion rate. Again, it really is a question of balance – either you optimize the CTR/QS without looking at the conversion rate, or you optimize the conversion rate by sacrificing the CTR/QS.

Whatever your strategy is, you might want to look at other ratios which take into account both metrics: either the number of conversions by impression, or the revenue by impression.

In a nuthsell, a successful SEM program is not necessarily a program with a strong Quality Score. What matters most is the overall profitability and revenue volume levels, which is a combination of all indicators mentioned above.

Beyond that, search marketers should have a good understanding of the whole conversion chain so that paid results are even more relevant than organic results – which is obviously advantageous for everyone: publishers, advertisers, and most especially web users.

Like vs. Google +1: What Are The Main Differences?

As Google product manager Daniel Dulitz mentioned, Google +1 is an annotation system which reinforces the interactions between organic and paid results, web pages, and your Google contacts.

That being said, your Google contacts are not necessarily your friends… and, you need to have a Google account to fully utilize the +1 feature. As a result, Google +1 acts as a collective rating system helping results be more trustworthy, but not adding the social infrastructure.

While Google’s +1 feature does draw upon a user specific network, it does not have the same social connotations of Facebook’s network. The +1 feature does function as a validity creating mechanism but does not contribute to the same social infrastructure that accompanies Facebook.

The Facebook Like is different in practice on several levels. Any website that is liked on a Bing search result is propagated into Facebook and vice versa. This means that something liked in Facebook will be shown on Bing despite any user interaction with the search platform.

This gives Bing/Facebook two avenues in which likes can be accrued as well as two ways to view the information. Additionally, Bing’s system is already privy to a year’s worth of information stemming from Facebook Like data giving Bing/Facebook a significant jump over Google.

Bing also brings forth popular pages based on global anonymous Facebook Like data – meaning that some anonymous Likes show even though searchers are not connected to Facebook, and makes it easier to go to Facebook after seeing a search result to have a “conversation” about that search query.

Note that you need a Facebook account to be able to see all these interactions; but unlike Google +1, you can create a Facebook account using any kind of email address.

Impact On Organic & Paid Search Rankings

Google +1 does influence search results, Google has said. It may also indirectly affect CTR, as it’s built to draw users in and encourage clicking.

In paid search, Google +1 will impact at least three Quality Score ranking factors over time: historical CTR, the quality of a landing page and account history.

As Bing Sr. product manager Duane Forrester explained at the Yahoo! Search and Bing customer forum, the Facebook Like button does not alter search results either, at least not generally. However, it does alter personalized results.

What Does This Mean For Search Marketers?

Bing’s market share could grow thanks to Facebook integration into the Bing search results because the Like button for websites has been available for more than a year, while the +1 button for websites was released just a couple of weeks ago. This should help Bing leverage the Facebook Like button as the Like button is already implemented across the web.

The truly personalized way Bing is starting to leverage Facebook data is promising from a search experience perspective. It could make lots of searchers – especially those with an intense Facebook activity – switch from Google to Bing.

For these two reasons and because it’s early days and it’s all about speculation right now, I would not be surprised if Bing gained some market share thanks to this integration with Facebook, especially if the Bing search engine becomes a bigger part of the Facebook user interface.

Social Search Is A SEM Game Changer

Google collects (by far) the most search queries, IP addresses, web history, and so on. Theoretically, Google should be able to determine every user profile based on search queries and web history. These queries somehow express interest, intentions, age, gender, and lots of other attributes and potentially make sense if they are well categorized.

If Google succeeds to convert all these queries to an audience-based database, and then leverage this for advertising purposes, Google could offer some ultra-targeted options without an actual social network behind it.

On the other hand, since Facebook does not directly collect queries (yet), but rather user profiles, Bing then needs to make the link between user profiles and search queries to serve the right ads, which seems to be more accurate and targeted.

While some web users might find personalized results invasive (like those ads related to the content of your personal emails in Gmail) there is no doubt search marketers already are looking forward to targeting searchers not only based on geo/language/device (that is so 2010)  but also based on more granular demographic targeting settings, such as gender, age, marital status, etc.. as well as interest.

The question is not if, but when is it going to happen?

Whenever it is going to happen, advertisers should definitely implement the Like and +1 button onto their websites in order to be part of this significant shift in the SEO/SEM space.

In regards to the question: “I already have plenty of buttons (Facebook Like, MySpace, email, Tweet, Follow, etc…) don’t you think all these buttons are going to become a usability issue at some point?”

I would answer: Yes, but test different setups to get more insight into the buttons your audience actually uses, and only keep those buttons that make sense to your audience.

Methodology

We studied an aggregated data set on Google US from the period prior to the launch of Google Instant and compared it to the week following the launch for a wide range of clients. Our client base is mainly retail, so other sectors may experience different results.

Also, this analysis is about triggered keywords, not user queries. To clarify, we are looking at the impact on the keywords we are bidding on, and not the actual search queries. This is a clear limitation as several factors can interfere, including but not limited to how many keywords are active, which match types are active, etc. We constantly expand long tail keyword sets for our clients based on search query reports and other proprietary mechanisms. By expanding keyword inventory, we always bid on the right keywords in exact match type.

Even though keyword data might be less pertinent than user query data, the week over week evolution does make sense, as long as most impressions, clicks and conversions are triggered by exact match type.

Findings

Finding #1: More generic trigger keywords. The average number of terms in trigger keywords has decreased by 1.64% overall. More specifically, there have been more 1-term keywords and fewer 2-term keywords. No significant change has been seen from 3+ terms, which means a shift is happening from 2-term keywords to 1-term keywords, without affecting the long tail.

Finding #2: Shorter trigger keywords. The average keyword length has decreased by 1.65%. More specifically, there have been a higher proportion of short (3 to 9 characters) keywords and a lower proportion of mid-long (10 to 20 characters) keywords.

Given the nature of Google Instant, our expectation that searchers will stop typing queries as they see relevant results appear is true. This phenomenon leads to more short keywords (fewer terms, less characters) and fewer long keywords being triggered.

Our gut reaction is that Instant is Google’s attempt to create a smaller pool of high volume, more generic keywords to drive up CPCs. Either way, generic terms have never been more important. Because a higher percentage of impressions, clicks, cost and subsequent conversions are occurring on these terms, it’s crucial to measure the effectiveness of generic terms on the sales funnel.

Similar to conversion funnels in AdWords, we developed a proprietary conversion path report that helps us optimize the whole search funnel—not just terms that receive the last click. Without looking at the entire funnel we wouldn’t be able to associate revenue back to those generic terms boasting relatively high CPCs and low conversion rates.

Finding #3: Higher CTR. While we have seen an overall dip in impressions of -6.75% from the period prior to the launch of Google Instant compared to the week following the launch, the amount of clicks has increased by 4.54%. This is due to an amazingly improved CTR: up 12.14%.

Rank also improved by 2.21% and CPC increased by 1.11%. We believe the jump in CTR is statistically significant whereas the CPC increase is not due to seasonality.

This jump in CTR shows how relevant real-time results are.

Finding #4: More conversion volume, though at a higher cost per acquisition. Unfortunately for advertisers, the average conversion rate has dropped slightly by 1.03%. However, this drop is actually not that bad relative to the increase in click traffic.

Despite a dip in impressions, we have seen an overall growth of +3.49% in conversion volume, while the CPA has increased by 2.16%:

Conclusions

Google Instant represents a significant shift for search marketers. Initial results show that searchers are letting themselves be guided by Google Instant’s suggestions, leading to more short, generic trigger keywords.

More importantly, overall relevancy from the keyword to the ad copy seems to be improved. As a consequence, more traffic is being driven through paid listing versus natural listings, which is undoubtedly in Google’s favor.

As for overall efficiency and conversion volume, we are observing a slightly lower conversion rate—yielding a higher CPA—but a significant increase in conversion volume. For advertisers looking for improved efficiency, this is bad news—whereas for those looking for incremental revenue through search marketing, this is definitely good news.