Efficient Frontier Insights

Facebook and Seafood in Seattle

    On the evening of April 27^th, Efficient Frontier hosted clients, prospects, and partners at Seattle’s iconic Canlis restaurant. The topic: Facebook.  After enjoying ample Dungeness crabs, the group heard from Forrester Research’s Elizabeth Shaw who referred to the channel as the “800 pound gorilla in marketing departments”.  Shaw, who covers Emerging Media at Forrester, spoke to a room of marketers about the importance of being agile and adaptive in the social space. She forecasted that within a year “one in six marketers will be absent from social channels”.  Shaw also emphasized that not every brand can and should use Facebook in the same way. Brands must have a strategy if they have a Facebook page. Social clutter (in this case when one’s news feed is bombarded by irrelevant noise from friends) is a real way for brands to miss the value of the channel. 

    Justin Merickel, Efficient Frontier’s VP of Marketing and New Product Development discussed three key insights on social marketing the medium. First, he reinforced Shaw’s perspective on the value of Facebook marketing. Justin also shared several insights on how advertisers have been using Facebook in their direct and brand marketing efforts. And last but not least, he wowed the audience by showcasing some key workflow and optimization of EF’s proprietary platform. 

Overall, the event fused thought leadership, excellent company, and exceptional Seattle seafood. Keep an eye out for other events brought to you by Efficient Frontier’s Performance Institute.

Attendees in the Canlis Penthouse

 

 

 

 

 

 

 

 

 

Julia Soffa, Marketing Coordinator

Facebook 101 – Images/Ads

In case you haven’t heard, you can advertise on Facebook now. Great! So how do you get started? Here are some things we have learned so far (through our own testing, as well as conversations with Facebook reps and engineers). It’s not an exact science yet, and I wouldn’t presume to define it as “best practices,” but it’s a good place to start. Enjoy!

Image Size
Our recommendation is to use a 110 X 80 or 85 X 62 image in your ads. Facebook accepts various sizes, but they are then resized down to the 110 X 80 size, and the image can become foggy or distorted. Therefore we recommend the 110 X 80 size so your image will appear crisp and clean.

Image Quality
There are three main recommendations in regards to which images to use. The first is to try not to use white space in the image. Since the background on Facebook is white, an image with a white background will blend into it, whereas an image with a dark background will be distinguished from the rest of the page. The second two recommendations go hand-in-hand: try to avoid using logos, and product action shots work best. Logos are great, but images where the user can click on the product action shot perform best. For example, if you are offering a travel deal, use a shot of a plane taking off or a picture of the beach. If you are selling laptops, show someone using one, instead of your logo. 

Ad Language
Facebook recommends the following as good drivers of click-through-rate (CTR):

• Ask questions in the ad
• Use third person
• Include a time prompt (now, today, limited time)
• Emphasize discounts & free offers

On the flip side, Facebook suggests avoiding the following pitfalls (so CTR is not diminished):

• Being too wordy
• Being too trendy in verbiage
• Using first person
• USING TOO MANY CAPS
• Percentage signs (it works better to include the actual dollar discount)

We hope these suggestions will help maximize your Facebook advertising campaigns. Happy Facebooking!

Karen Maciolek
Senior Account Manager/Master Social Media Strategizer

Tips For Managing Ad Fatigue on Facebook

In my latest Search Engine Land column, I take a look at the phenomenon known as “ad fatigue” and offer tips for how to combat it when managing your ad campaigns on Facebook.

Due to the nature of Facebook, it’s likely that users will see your ad repeatedly, which may potentially render them less effective over time. So what can you do to keep your Facebook ad campaigns fresh and effective and combat “ad fatigue”? Here’s a link to my analysis and several tips:

http://searchengineland.com/tips-for-managing-ad-fatigue-on-facebook-73554

Dr. Siddharth Shah
Sr. Director, Business Analytics

 

Facebook: Let’s Get Personal

Everett Rogers published the “Diffusion of Innovations” in 1962, which according to Wikipedia, is “a theory that seeks to explain how, why, and at what rate new ideas and technology spread through cultures.” In layman’s terms, it’s this classic graph that they show you on the first day of business school:

  Source: Wikipedia

The most interesting aspect of Facebook is that depending on how old you were between February 2004 and September 2006, as well as whether or not you had a .edu email address, there were restrictions on where one fell in the above graph. It wasn’t until September 2006 that anyone over the age of 13 with an email address could join. By that time you probably heard about Facebook from a friend, or a friend of a friend.

A friend of a friend is now an important metric on Facebook. It’s called Friends of Connections (FOC) and refers to the friends of users who like your page/brand. The Facebook reps I work with (who are fabulous, by the way) inform me that the FOCs are one of the most important segments you can target since people are highly influenced by their friends and networks.

This original study has been advanced to the realm of retail and buying patterns, where it’s become proven that the early and late majority are your potential customers, those most likely to buy the product. So what I wanted to look into were how many friends these early- and late-majority Facebook users have versus early adopters. I didn’t include innovators, since I only have two or three friends from Harvard on Facebook, or laggards because the laggards I know (hi, Dad!), still aren’t members of Facebook.

Here’s what I found by taking a random sample of 10 friends from each category:

According to the theory, the groups in blue (above) are your potential customers. The interesting finding is that if I look at the average number of friends in each of these categories, there is a definite trend that develops in the late majority based on age.

So is this good news? Yes. Focusing on the late majority group, I broke this up into the three age ranges listed above. What’s unique about the late majority (over 40) is that their friends on Facebook are actually their friends in real life, which brings more legitimacy to the power of suggestion. What’s unique about the late majority (under 20) is that they have a huge quantity of friends to advertise to, and since late majority users generally move in a pack mentality (eg, they only buy when someone in their circle does does), this is great news for advertisers. Also worth noting is that a good portion of the late majority (under 20) users haven’t even been to college yet, where their friends on Facebook will grow exponentially. Add these factors together, and the late majority user on Facebook looks like a promising target audience.

 

Karen Maciolek

Senior Account Manager

Master Social Media Strategizer

Best Practice on Google ACE

Google ACE is an excellent tool that came out in October last year, which allows for scientifically significant testing across many features within an Adwords Campaign. In this blog post we won’t discuss how to actually implement ACE tests (Google has that covered), but will focus more on the functionality and best practice when setting up one of these tests.

So what is ACE?
ACE stands for Adwords Campaign Experiments, and it is a tool that allows you to split potential Google traffic at its source. It is applied at the campaign level, and traffic can be split in a number of ways (usually we opt for 50:50) between a ‘control’ group and an ‘experiment’ group. You can also put parts of the campaign into both groups at the same time. Best practice here is to have everything you aren’t testing in both groups, the new part of the campaign you want to test in ‘experiment’ group, and the current part of the campaign (i.e. the part that you want to replace if the test goes well) in the ‘control’ group (this is explained in more detail further down).

What can you test?
You can currently test anything you can change within a campaign. You can add new keywords, negatives (Adgroup level), placements, different ad copy and alternative Adgroup structures. However there is currently no functionality to test anything applied at the campaign level. I have heard talk that the next thing to be added to ACE is the functionality to test site links, which would be nice, but our wishlist also includes GEO-targeting, campaign level negatives, networks, devices, location extensions, dayparting and campaign budgets.

How do you test?

We have found that the biggest limitation with ACE is its reporting.  Google was obviously keen to get this useful tool in the marketplace as soon as it could, so reporting seems to have been neglected. The way we get round this at EF is by building duplicate Adgroups (a generic example is below).

Campaign: Holiday Destinations:
Control \ Greece Experiment \ Greece
Control \ Portugal Experiment \ Portugal
Control \ Spain Experiment \ Spain
Control \ France Experiment \ France
Control \ Italy Experiment \ Italy

This is the starting point for an ACE test (the above experiment adgroups are exact replicas of the control groups). We would then make the changes we wanted to test (i.e. a new landing page URL) to the experiment groups only, and start the ACE test to split the traffic 50:50 between both groups. Many of our account managers now maintain a continuous ‘test’ campaign, purely for ACE testing as they always have several tests running at any given time. This ethos of continuous testing is important in PPC (and a policy that Google itself has always championed) because it is the only way to keep learning ways to increase the performance of your account / business in the long-term.

So what is the benefit of using ACE?
In short, flexibility and accuracy. Previous tests relied on the before and after technique of making a change and comparing last week’s data with this week’s. The problem with that approach is that no two weeks are the same. Anyone who has been running a PPC campaign for some time (particularly those working on large-scale, complex accounts) can tell you that performance changes because of  a variety of conflicting variables, including consumer demand, seasonality, the weather, even the quality of what’s on TV!  Using ACE means both the control and experiment group are being exposed to the same market conditions. You can write off all of those variables because they should affect both parts of the account equally. You can be more confident that any differences in performance are due to your changes, rather than a statistical fluke.

As well as being more accurate, ACE is more flexible. Previous ad copy tests relied on putting the whole campaign on even-rotate. This approach is inefficient as your worst performing ads in other ad groups are now being shown more often than they should, dropping quality score, inflating CPCs and decreasing conversion rates. Using ACE allows you to test just the two ads you want to test, leaving everything else in the campaign as it is.

Conclusion
Used correctly and extensively, ACE is a great way to discover actionable insights into what drives performance in your account. It doesn’t take long to set up, is fairly intuitive and can help you drive performance. I would highly recommend anyone not using ACE to start using it as much as possible.

Michael Taylor, Associate Account Manager, Efficient Frontier

When the numbers don’t add up: Our analysis of the current Facebook CPC debate

Last week, Efficient Frontier reported a 40% QoQ increase in Facebook CPCs, and this week TBG reported an approximately 30% drop in their U.S. campaigns. As Twitter chatter and the blogosphere debate the results, there are a few key differences in the numbers that have been missing from the discussion.

First, we reported on the actual CPCs that marketers paid while TBGs numbers reported trends in the suggested min and max CPCs on Facebook.  Therefore, the difference in percentage growth and decline  are an apples to oranges comparison.  Suggested CPCs are a function of Facebook’s algorithms and less a function of what advertisers are willing to pay for a click. In our opinion, CPCs are a better measure of the market because they reflect what people are ACTUALLY paying Facebook. Suggested CPCs are at best an inferential measure.

Second, we compared data from Q4 2010 to Q1 2011 while TBGs numbers trend from March 19^th to April 11th, a period of about three weeks so it is a much shorter time frame in which to draw conclusions.

Third, the methodology of the analyses has not been mentioned. I would like to point out that for Efficient Frontier’s analysis, we examined a broad range of clients from a variety of sectors including retail, financial services, travel, and automotive.  Our index includes more than 10 enterprise class customers that have been Efficient Frontier clients on Facebook for more than 2 consecutive quarters. While we feel that our dataset is fairly robust and directionally correct, we are aware that there would be some variance in the numbers. This is typical of an early stage, rapidly growing channel where the marketplace is very dynamic. However, a +/-40% discrepancy is startling. Which brings us to the crux of the debate. Which number reports the correct trend?

Mathematically CPCs, click volume and Facebook’s revenue from Marketplace ads have a very simple relationship. Facebook’s marketplace revenue is the product of the CPCs that Facebook charges advertisers and the clicks that advertisers get .

If CPCs are truly down 30% the percentage of paid clicks in a quarter would have to increase by 30% for Fsacebook’s revenue to remain flat. Assuming that audience reach is relatively stable, the ad serving population would have to almost triple in a year in the U.S. for Facebook to record the same revenue as last year. Anecdotal evidence shows that such an increase in inventory is unlikely. As one commentor on AllFacebook points out, the roll out of the new profile in January would have actually decreased inventory as the “More Ads” link was removed.

This is contrary to all the trends and reports that predict that Facebook will double revenues in 2011. A look at the early days of search suggests that there will be some of both CPC increases and paid click increases going on. On the demand side, advertisers are rapidly leveraging the platform and are willing to pay more for their ads while on the supply side, Facebook is growing its audience reach so that there is an ever greater number of ads that Facebook can serve its ads. However, to record a doubling of revenues we believe that the bulk of the growth will come from the increase in CPCs at this stage.

In summary, there are 3 reasons why we think our numbers are directionally correct. First and foremost, we report on the actual CPC metric and not an inferential, indirect number that does not necessarily reflect the market. Second, we have a robust methodology that looks at longer term data. And finally, our numbers are consistent with the anecdotal evidence and reports about Facebook’s growth in 2011.

I hope this helps clear up the discussion.

Dr. Siddharth Shah
Sr. Director, Business Analytics

New guide to performance marketing for retailers

As we’re working closely with a number of retailers to integrate search, display and Facebook, we thought it would be useful to share some of our best practices in a free guide available to download here.

We recently presented this paper at the Internet Retailing Expo and also at the inaugural Efficient Frontier Performance Institute Breakfast Seminar called “Top Tips for Retailers” at the Covent Garden Hotel in London.

The guide highlights six trends that retailers need to be a part of if they want to maximise the results they can generate via performance marketing. These include: utilising campaigns which link real time stock information to advertising; ensuring that the adverts are not making promises that the retailer can no longer fulfil; and tailoring results to customer intent and making an effort to understand the journey that the customer makes prior to purchase. (The customer may make several generic searches via search engines before they even search for a specific brand name, and by that stage their research may have led them to the same brand at each stage. How many ads had the customer seen before finally clicking on the one that led them to that purchase? What can that information reveal to the retailer?)

It’s also important for retailers to optimise cross-channel campaigns effectively. Our research has shown that TV ad campaigns can have a significant impact on search term performance, for example. Retailers have to abandon a silo mentality. If a customer sees a good deal online, they want to be able to walk into a store and get that deal. Using unique identification numbers on online coupons that the customer can redeem in store can make tracking conversion much simpler for the retailer. It can also encourage customers to make a purchase, especially if they were reluctant to shop online.

Facebook is a really interesting channel to watch. Facebook Credits are being sold offline in stores such as Tesco and Game, encouraging customers who are reluctant to part with their payment details to spend money via Facebook. With such a large number of users providing a rich source of data about themselves there is no doubt that Facebook is becoming a major consideration for retailers undertaking a performance marketing campaign and something that retailers should keep an eye on.

Download the full report here or to request the PowerPoint presented at the two events, please get in touch via @efrontier.

Merinda Peppard
Global Marketing Manager

Facebook: True vs. Unique CTR

In the search world, we calculate click-through-rate (CTR) as clicks divided by impressions, and it measures the number of times an ad was clicked versus the number of times it was viewed. This works a bit differently for Facebook. Here, a user can see the same ad multiple times per day, which leads to both a true versus unique CTR for that ad.

For example, assume you and I see the same ad on Facebook 10 times a piece, and at the end of the day, you click on it and I don’t. This would equate to one click on 20 impressions, or a CTR of 5%. But, if we were to look at unique CTR, we each saw one ad throughout the day, you clicked, and I didn’t. This would equate to one click on two unique impressions, or a CTR of 50%, which is obviously quite different.

Looking at this on a larger scale over the course of a week, we see the following trend:

If we analyze this campaign based on true CTR, we might assume a steady trend of clicks to impressions. But, if we include the unique CTR in the analysis, we see an increasing trend in the proportion of clicks to impressions (or user interest), which starts to decrease after the third day.

This brings up the next line of questions in regards to ads on Facebook. When does ad fatigue kick in? At what point do we need to refresh ads or images to interest the user again? At Efficient Frontier, we’re currently testing these questions and uncovering some interesting findings, so stay tuned for more on this topic.

Karen Maciolek

Senior Account Manager

 

 

 

Q1 2011 Display Publisher CPMs: Advertisers Clamor for Video Inventory

In recent months we have been releasing monthly Display publisher performance trends, and we noted that some interesting ones have emerged. Today we take a look at Q1 2011, and as we highlighted in December 2010, Video Publishers’ Inventory has clearly been the most sought-after placements for U.S. advertisers.

The story of February 2011 was that social media publishers had proportionally one-third of all Display impressions. However, looking across Q1, that was certainly a spike, because looking at Q1 as a whole, the overall impression share was about half that.

Figure 1 shows the different publisher types compared against each other, and Figure 2 shows actual numerical performance per publisher type and how they performed in Q1.

Figure 1: U.S. Display Publisher CPMs (cost per thousand impressions) – Q1 2011

Figure 2: U.S. Display Publisher Type Trends

In addition to Video inventory being almost twice as expensive as the next highest publisher types -- News & Sport and Shopping sites -- it also had the highest rate of user interaction (i.e. Click Through Rate – CTR) of any group. Gaming sites also recorded strong CTRs but proportionally were only 1% of impressions and 4% of the sites in Q1. In contrast, Video Publishers, while having fewer sites than Gaming Publishers, had nine times the amount of inventory volume, possibly indicating that users are engaging more with those sites across multiple pages.

Advertisers should use this data to get a gauge on the differing volume levels that are available in the market via the Display Exchanges and the comparative levels of CTR and CPM. This data could be a leading indicator as to how much time users tend to spend on those types of Publisher sites, which could dictate the types of creative messaging that would be most effective.

Ultimately, these trends and insights are only as useful as the sophistication of your marketing funnel and the platform that is able to optimize your Display campaigns. Finally, when managing a Display campaign, ensure that you are not looking at performance in isolation, as it important to fully comprehend the interaction between all your digital channels to achieve optimum results.

Chris Jacob

SMX München 2011

SMX (Search Marketing Expo) took place in Munich last week, this time in the Hilton Hotel Munich near Englischer Garten. Niels Jensen, Efficient Frontier’s own Account Director for Germany presented alongside Phillip Arnold from TUI Interactive on the panel "Search & Beyond". TUI is one of the largest travel brands in Europe with more than 10,000 hotels in its portfolio. The topic of the presentation was "The Adwords Strategy at TUI" and both presented insights and best practices from TUI Interactive’s Hotel Names Campaign. Efficient Frontier’s objective was to build up a search campaign for hotel names only, which are part of TUI’s hotel portfolio, without being over-reliant on TUI’s brand. 

Niels and Phillip discussed how the set up of the long tail was constructed and how the bid management for nearly 100,000 keywords was put into place. In addition, they discussed an A/B landing page test set up and how the dynamic ad copy was handled by Efficient Frontier. A key strategic decision by TUI Interactive was to replace the hotel pages of “Traveltainment” they used previously, with their own landing pages, in order to get better information to users.

Later in the presentation Niels and Phillip uncovered the results of the campaign. The overall Quality Score was increased to factor 7 and the key figures for the click through rate, conversions and return on marketing investment were increased substantially. On the whole, TUI Interactive is excited about the results they achieved with Efficient Frontier. Both Niels and Phillip were mobbed by the 80 strong audience after the presentation with many interesting questions which tells us that the topic was greatly received.