Efficient Frontier Insights

Increase returns by tracking and optimizing the right web metrics

Metrics are important, both in online and traditional marketing but it's the trackability which makes online marketing so exciting. What's even more exciting is that the technological advancement in online marketing has enabled maximization of multiple success metrics simultaneously.

Track right metrics

Tracking provides visibility on business relevant metrics like return on investment (ROI), cost per order (CPO), return on ad spend (ROAS) and user lifetime value (LTV). You should try to track as many metrics as possible, even if they are not needed at the time the campaign is set up. It's frustrating not to have the data you need at a later time.

Conversion data on various steps in the process can provide useful insights on visitor bounce rates, page views, links clicked and actions taken. Having this information can help understand how users navigate through the site and the pages where users drop off.

Optimize to success metrics

If the conversion is a single step process, it's easy to choose the optimization metric. However, often the conversion cycle involves multiple steps. Marketers are faced with the challenge of selecting the best optimization metric in such situations.

For example, for an online dating advertiser it could be confusing whether the optimization metric should be registrations or subscriptions. By trying to maximize only subscriptions, the advertiser runs the risk of ignoring registrations which could potentially bring subscriptions. On the other hand, by trying to maximize only registrations, the advertiser might end up with too many registrations which may never turn into subscriptions.

These metrics are interdependent and for optimization purposes one can not be chosen over the other. This is when multi-metric or "blended" optimization is necessary.

Blended optimization

At Efficient Frontier we use a blended optimization approach which allows our clients to optimize their ad campaigns concurrently to multiple business metrics. For example, the marketing objective for one of our banking clients is to maximize new customer sign ups. The sales process involves three steps: application form completion, approval and final sign up. The campaigns or keywords driving final sign ups are more valuable than the ones driving approvals and form completions. With Efficient Frontier's tracking technology, the client is able to track entire sales funnel and the user drop off at each stage in the funnel.

Efficient Frontier's optimization technology allows our clients to assign different weights to different stages in the sales funnel based on the value of each stage. Using this technology, the banking client is able to maximize customer sign ups while maintaining the constant inflow of the leads or application form completions.

Web metrics can tell the advertisers exactly where the marketing budget is being spent and what they are getting out of every marketing dollar. If implemented, tracked and optimized right, these strategies can certainly help increase the return on investment.

Google Becoming More Efficient: Improved CTR

The recently reported decline in Google paid clicks caused what may be an unnecessarily strong reaction in the investment community this past week. Efficient Frontier's US Client Index, which covers more than 20 million paid clicks each month, did indeed see a 5% drop in click volume on Google when comparing January 2008 to January 2007. But same advertiser spend was up on Google by 7% in the same period, and click through rates improved by 10%. CPCs were up by 13%. This data suggests, as Comscore posited, that Google is becoming more efficient at serving ads by delivering more relevant clicks to advertisers. Thus it is able to charge a higher premium for those ads, hence the increase in CPCs.

Our Search Engine Performance Report: Q4 2007 showed that ROI on Google was up 7.5% from Q4 2006 to Q4 07, which does not spell disaster in my mind. Next week we'll take a look at overall statistics for February on this blog. Stay tuned!

SEM Tools: Risks, Limitations & Best Practices - Part III: Demographic Targeting

In the previous two posts in this series, I discussed key elements to pay attention to in geo targeting and content targeting. In the final post in the series, I will discuss the risks, limitations and best practices in using demographic targeting.

Part III: Demographic Targeting

Demographic targeting is a relatively new tool currently offered by MSN AdCenter and Google Adwords (for limited site targeted content campaigns). Through demographic targeting you can target your audience based on gender, age, ethnicity, family income and even whether they have children at home or not. This would be one of the greatest tools for search engine marketing if completely accurate, since you will be able to show your ads to the preferred demographic. However, there is still room for improvement.

1. Inaccurate User Information

Search engines base demographic targeting on the information users enter at the time of signing up for any of their products. MSN AdCenter, for example gets this information from MSN Hotmail and Microsoft Windows signups. It’s a common knowledge that people give inaccurate information during signups and there is no shortage of fake Hotmail and Gmail profiles on the internet. Search engines base their trends on this information which is perhaps not entirely accurate.

For example, a 50 year old man in Canada can easily create a user profile of a 27 year old woman in San Diego, California. Thus a cosmetics ad intended for women in their 20s would not reach its intended audience in this case.

2. Again, know your business

Just like , demographic targeting would be applicable only for certain businesses and not for others. If your product is targeted towards a particular demographic, it may make sense to use this tool.

When I looked at the demographic report for the keywords "evening gown" and "tuxedo" from MSN AdCenter’s new MS Excel Add-in "Ad Intelligence", I could see very clear gender-based search trends. If you are a business which aims to sell evening gowns and tuxedos, you should keep two sets of keywords, one targeted towards men and the other towards women. 

Similarly, if you look at the demographics for the keyword "baby clothes," the majority of searches are from women between the ages of 18 and 50. 

On the other hand, if you look for demographic trending of electronics related keywords like "hdtv" and "digital camera," the search volume doesn't skew strongly towards one gender or age group. In this case demographic targeting would limit your reach more than necessary and is not recommended.   

All said, you should have enough data and proof to determine which demographic group is more likely to buy your product before you use this feature.

Unlike geo and content targeting, demographic targeting is relatively new. I am still not completely confident of its accuracy as many web users likely enter incorrect information. I am positive though, that the tool will become more robust in future. If you decide on using this tool, I would strongly recommend conducting research on demographic trends in your business from both internal and external sources.

SEM Tools: Risks, Limitations & Best Practices – Part II: Content Targeting

In the second post of the series, I will focus on another popular search engine marketing tool: content targeting. This post mainly covers various setting options in content targeting to be aware of and ways to avoid common mistakes.

Part II: Content Targeting

Content targeting implies the use of the SEM platform to place bids and advertise on websites which are part of the search engine’s content network. There are few points to keep in mind while using content targeting for SEM campaigns.

1. Do not turn on search and content for the same campaigns

Search engines provide the option to have the same campaign targeted to both, the search and the content network. Creating a single search+content targeted campaign seems like an easy choice, but in order to be more effective, it is better to create two separate campaigns, one search targeted and the other content targeted.

Content advertising is quite different from search advertising. Through content targeting your ads show up when someone is browsing a web page/website which contains keywords related to your campaign. In search advertising your ads only show up when a user enters a keyword in the search box that you have placed a bid on. The intent of the searcher is much stronger than the intent of the user browsing a website, where they may already be viewing the content they were searching for. Thus, content advertising has much lower click through rates and lower CPCs, but can give your campaign broader exposure outside of search.

In order to show ads relevant to a web page, the search engine algorithm needs to match the keyword with web page content. Since it is not possible to find relevant web pages simply by looking at one keyword, the search engines look at all the keywords in an ad group and identify the dominant theme represented by that ad group. Once they have identified the theme, they find relevant web pages and place ads on them. In search advertising the algorithm just looks at the keyword to show the ads.

Therefore, it’s a good practice to have separate campaigns targeted only to the content network. Within these content campaigns group similar keywords together in ad groups so that each ad group has its own theme.

2. Be careful what you bid

As mentioned above, content CPCs are much lower than search CPCs, because of less competition and lower conversion rates. There is no reason you should bid same amount for search and content placements, especially when search engines provide options for separate bidding. As recommended earlier, always try to have separate campaigns for search and content but in case you have the same campaigns targeted to both the search and the content network, you should separate bids for search and content clicks. For example, to separate bids in Google AdWords you can check the box "Content bids" during campaign set up.

3. Understand Billing Options

You should understand and if required, use the various billing options available for content targeting. For example, Google currently offers several billing methods such as CPC (Cost per click), CPM (Cost per 1000 impressions) and PPA (Pay per action).

In CPM you pay for 1000 impressions whether or not anyone clicks on your ad. CPM is applicable to the advertisers who want to reach customers early in the advertising cycle, with ads designed to increase awareness but not necessarily generate clicks or traffic.

In PPA, you don’t pay for the clicks but pay only when certain action (for example, subscription or lead capture) is completed by the users on your website. You can set this action and the price you are willing to pay for it. PPA billing is useful when click-to-conversion rate is extremely poor and when the advertisers know exactly how much each conversion is worth to be able to set its value.

In conclusion, for a campaign to be effective on content network ensure that similar keywords are grouped together (as discussed in my post on campaign organization) for the search engines to identify the dominant theme, campaigns are set up for the most efficient bidding and you are using the billing method most appropriate for your business.

 

SEM Tools: Risks, Limitations & Best Practices – Part I: Geo Targeting

Should you use all the search engine marketing tools currently being offered by the search engines? Maybe yes, maybe no. If you do use them it is very important that you understand them correctly and use them in the most effective manner. As the competition among search engines increases and advertisers have more options, search engines have to offer more and more SEM tools to attract and retain advertisers. These tools are definitely useful and tempting, but you must understand their limitations and applicability to your business before using them for your campaigns.

This 3 part series will explore risks, limitations and best practices in using some of the currently popular tools: geo targeting, content targeting and demographic targeting.

Part I: Geographical Targeting 

Through geographical or "geo targeting", you can target your advertising to any US or international region(s). Your ad will be shown only in the selected regions. However, before using this tool for your SEM campaigns you must understand:

1. Your Business type

Geo targeting is useful for businesses whose offering or product prices vary by location, such as online home listings, online dating and web-based job listings. In all these cases there is strong likelihood of the searcher having a location preference.

It is also useful for small businesses who can only serve a local market. Obviously, the neighborhood pizza place in San Jose, CA would not want to show ads in New York and pay for the non-converting clicks.

Geo targeting is most likely not useful for businesses like online retail and enterprise software, since the product offering and prices do not usually vary from one location to the other. You can avoid spending resources in setting up geo targeting if it does not make sense for your business.

2. Bidding limitations

Without careful planning, you will end up placing the same bid for keywords across different locations. This is dangerous and ineffective because keyword CPCs vary across locations based on what competitors are bidding for the keyword in that location. You should create separate campaigns for each geographical region you are targeting, in order to avoid placing the same bid for a keyword across multiple locations.

In the Google AdWords example shown, the campaign “Online Dating” is targeted to Australia, Germany, United States etc. The problem with this structure is that the keywords in this campaign will have the same bid across multiple countries. 

 

 

If there is a keyword “singles dating” in this campaign and if we bid $5 for it based on its performance in the US, we will end up placing the same $5 bid in Germany, in Australia and in China. While its performance might justify the bid in United States, $5 might be very high for China and very low for the Germany. With this bid you might end up at the top position in China, which may not be the most optimal position there. On the contrary, in Germany you might lose out on good traffic because of the low bid and hence low position. 

3. Technical limitations

Geo targeting works by the search engine’s ability to capture a user's IP address or other identifying information, check that information against a database that suggests the likely geographic location of that user, and then serve up targeted ads based on this geographical location. The database is not 100% accurate and there could be IP addresses which are missing from this database. No geo targeted ads are served to the users with missing IP addresses.

Secondly, company firewalls and large internet service providers (which use several IP addresses) tend to skew geographical location. This could result in ads to show up at unintended locations. Geo targeting therefore, is not 100% accurate and sufficient to saturate a regional market with search engine advertising.

To conclude, before you set up any kind of regional or international geo targeting for your campaigns, it is important to ensure that the targeting applies to your business and it is set up for the most effective bidding. In addition, you must understand that you might lose out some traffic volume because of the technical limitations.

Algorithms White Paper

Today we launched our new website, which introduces the new Efficient Frontier logo and corporate identity. As part of the launch, I wrote a white paper about the algorithms behind our technology. The paper can be downloaded here.

Efficient Frontier's algorithmic approach involves click modeling and revenue modeling for every keyword in a portfolio, and then chooses the optimal bidding strategy based upon those models. Without such models, it would be impossible to know what level of performance your campaign could achieve just by changing your bids.

Researching the white paper was a fascinating process - I looked into the history of mathematics, computer science, and economics in order to explain the application of modern portfolio theory to search engine marketing. Our founder, Anil Kamath, had a unique vision that pioneered much of today's thinking behind search marketing optimization. If the white paper whets your appetite for more, read this interview with Anil by Eric Enge.