The good news is the inbox is now more of a marketing channel than ever before. The bad news is there is more competition, and marketers need to work extra hard to stand out.

Luckily, measuring email campaign performance is moving beyond opens and clicks. New measurements are available that can tell you not only how your campaigns are faring but also what’s working for your competitors.

Email Forwards

When I talk about email forwards here, I mean when a subscriber literally opens the email and hits the “forward” button in the email message to send it off to a friend or family member. I do not mean the ability to forward or share through a Web form or through a social channel.

This forward measurement is not only a good indicator of interest from your subscribers, but also their friends and family. Not to mention those recipients are more likely to open the email since it’s being sent by someone they know.

You can use this metric to look at past performance of campaigns and, more importantly, you can look at the forward rate of your competitors and borrow their best tactics.

For example, during the Christmas season, Kohl’s and JCPenney sent campaigns at roughly the same time period.

Kohl’s subject line – “Take an Extra 20% Off Everything With Your Friends & Family Shopping Pass!” – presented a clear offer and gave a clear call to action.

Compare this with JCPenney’s bland subject line, “A Letter And Gift From Our CEO.” Even Kohl’s creative was much better, with the CTA above the fold offering a large discount. The JCPenney offer was all text and no images with the offer (a $10 gift) buried at the bottom of the email.

If, like me, you guessed that Kohl’s was forwarded more, you’d be wrong. The JCPenney email was forwarded over twice as much. Plain, direct and to the point won out over conventional best practices.

“This Is Not Spam” Rate

We’ve all had emails land in the spam folder; and once that happens, your open and conversion rates drop. At this point, you hope your subscribers actually go into their spam folders and mark the message as “not spam.” This is rarely the case, though.

A recent Coca-Cola MyPoints email campaign sent to the program’s entire list had less than 5% of the campaign delivered to spam folders. Ninety-three percent of those emails that landed in spam were marked as “not spam,” a remarkable feat for anyone.

In this case, Coca-Cola’s advantage was its strong, recognizable brand, which was used in both the subject line and the from address. Personalization and segmentation were also used, and clearly, the results paid off. The program’s successive campaigns, thanks to these high “not spam” rates, had inbox placement rates of 100% at Gmail.

Deleted Without Opening & Deleted After Opening

On the other end of the spectrum, knowing what people ignore or delete after reading can be just as useful. In the case of Coca-Cola, their most read campaigns were also their most ignored.

Subscribers read the MyPoints reminder campaigns more than other campaigns, but a high amount — almost a fifth of their list — deleted the emails after reading them. Email marketers can use this along with their conversion tracking to determine where creative isn’t performing.

Another Coca-Cola campaign for a sweepstakes had read rates that were similar to other campaigns, but 40% more subscribers deleted that campaign without ever opening it.

If we were only looking at read rates or open rates, we’d probably deem the campaign a success, or at the very least, acceptable. Looking at the rate of those that deleted it without ever opening it tells us that we need to further test our subject lines, further segment the list to appeal to the others who are active but just not interested, or both.

Beyond Opens & Clicks

These metrics move beyond the traditional opens-and-clicks and provide more insights into subscriber behavior. Testing is still a requirement, but now, you have more data to support faster, better decision making.

Now that the new market has had time to mature, we can take the opportunity to step back and review what we have learned about this new marketplace and how it interacts with traditional paid search.

Cannibalization Between PLAs And Text Ads

Cannibalization has always been a concern with PLAs, as they compete against traditional text ads for the same impression and click opportunity. We looked at a month’s worth of search query data across a number of retail advertisers, and found that, on average, 34% of all search queries that resulted in the display of a text ad, also had a PLA displayed. Of search results that had a PLA displayed, 49% also had a text ad displayed. This means that as much as half of volume gained on PLAs may be taken from text ads.

Unfortunately, there is little that can be done about this overlap. We can compare the parallel-auction system to a poker game that allows each person to play two hands at once: folding one hand decreases competition evenly across all players, so you end up holding the short end of the stick by having only one hand to compete with instead of two.

With this system, retail marketers have little choice but to accept that some cannibalization in inevitable, and continue to enter both auctions to maximize reach and volume. This is especially true since PLAs and text ads have fundamental differences in how they behave, and neither is a complete substitute for the other.

PLA Scalability Is Limited

One major difference is that the scalability of PLAs is limited in comparison to text ads. There are two main reasons for this:

• PLA impressions are dependent on the inventory you have in your merchant center feed.

PLAs have less impression opportunity compared to text ads in that Google knows your inventory and pricing, which makes your ad unlikely to show for a search query describing an item that does not exist in your feed. For instance, the search query “red boots” is unlikely to show your ad if you only have black or brown shoes in stock, and similarly for “cheap boots” if your inventory is priced higher than competition.

• PLAs only show for product-related queries, and becomes increasingly likely to show as the product description in the search query becomes more specific.

For example, a search for “shoes” yields plenty of text ads but no PLAs due to the category being too broad. Similarly, “zappos”, an online shoe retailer, shows no PLAs, nor does a search for “nike”, a shoe manufacturer and brand. However, a more targeted query such as “nike shoes” starts to yield PLAs – but still only one or two for any given impression — while a specific product query like “nike running shoes” results in a full block of 5 or more PLAs.

The limited scalability of PLA campaigns is one reason why it is a bad idea to forego text ads entirely, even if PLAs are performing better. In terms of striking the ideal balance of spend between the different ad types, it becomes important to determine how much each PLA campaign can scale while maintaining efficiency.

Spend And CPC Inflation On PLAs

For text ads, marketers typically use average position (ad rank) as an indicator for how much room there is left to scale. However, there is currently no positional information available for PLA campaigns. Though one was temporarily in place last year, Google decided to retract this feature, claiming that they are working on fleshing out a better position calculation for PLAs.

Looking at how average spend share, click share, and impression ratio of PLAs vs. text ads has trended since last summer, we see that impression ratio and click share plateaued in January, while spend share on PLAs continued to rise. In aggregate, we see average CPCs on PLAs with respect to text ads rising almost 18% from December to March, despite there being little change in actual clicks gained. This means that advertisers are pushing spend onto PLA campaigns seeking additional volume that does not exist.

In search, it is common practice to shift spend towards campaigns that are more efficient, except when the high-efficiency campaign is already bid up to a very high position, in which case any marginal volume can only be gained at a hefty price. The lack of such an indicator for PLAs is likely playing a part in the CPC inflation we see currently.

PLA Impressions Are More Binary

So how do we prevent this inflation? There are several ways to do this, the simplest being to always make sure that increased investment always results in greater click and impression volume as well. If it does not, then you are trying to push more money into the campaign than is optimal.

If we lower the bid on a text ad while competition holds steady, the average position for the ad is almost certain to drop, which in turn decreases click-thru rate (CTR). However, this is not necessarily the case for PLAs: as all ads shown for a given impression is displayed within a block, the effect of display position on click-thru rates are likely to be much smaller. In this sense, PLAs function more like display ads than search ads.

It is possible for Google to influence CTR by varying the number of PLA ads displayed based on the bid distribution, but we do not see this happening at scale: a majority of PLA blocks on desktop devices seem to be delivered either as a 5-ad box in the left column above organic results, or as an 8-ad box in the top right.

This means that for any given product target, once a certain bid level is reached, there may be little benefit to raising the bid further. If you raise or lower your bids on a PLA target and see very little change to impression or click volume, this is a good indication that bids and CPCs can be lowered further to increase efficiency without sacrificing volume.

Takeaways:

1. Always play both hands, PLAs and text ads, even if one is performing considerably better with respect to the other. Spend can be shifted between the two ad types depending on relative performance, but never put all your eggs in one basket.
2. When pushing spend on PLAs, make sure to do so only to the extent where you are seeing an increase in impressions and clicks. Bidding up beyond that point will only result in unnecessary CPC inflation with very little incremental volume to be gained.
3. Be wary about applying the same set of rules or ROI goals to both PLAs and text ads. Scalablility differences mean that the point where their performance converges is not always the most efficient.

CPCs remained relatively flat month-over-month with little movement. Retail CPCs showed a slight decline for Google, showing relative softness compared to last year. It remains to be seen whether this is a temporary dip or a long-term indicator. Finance CPCs nudged up as the tax deadline approaches. Overall, Google CPCs are 1% up year-on-year, while Yahoo!-Bing is flat.

March 2013 Cost-per-Click (CPC) Update, US Search

And now, we’ll discuss ad scheduling strategies for Google Enhanced Campaigns, providing some revenue-per-click (RPC) trends due to ad scheduling. With a little knowledge and effort, these strategies can be used to improve your overall ROI.

Enhanced Campaigns

Google’s announcement of Enhanced Campaigns last month made large waves in the paid search space. Among other changes, the option of creating device-targeted campaigns will be deprecated in favor of device-unified campaigns with mobile bid modifiers. In the new campaigns, desktop and tablet devices will be grouped, and mobile devices bid to a percentage of the desktop and tablet bid.

The implications of this change in terms of losing the ability to set device-specific bids at the keyword level has seen plenty of commentary, but one aspect of this change has not been discussed: its impact on ad scheduling.

Scheduling On Enhanced Campaigns Apply Across Devices

With Enhanced Campaigns, you can still adjust bids by day-of-week or time-of-day. The problem is that the same schedule is applied across all devices, even if RPC (revenue per click) trends are different. Whereas before, you could make separate campaigns for desktop and mobile, setting different schedules. This option is no longer available short of creating separate campaigns targeting separate times of the day, which is impractical in most cases.

Day-of-week and dayparting strategies are informed by two factors, RPC and CPC patterns, where the latter represents market competition. Allocating more budget toward days or times when CPCs are cheaper and RPCs are higher lead to better average ROI.

A basic ad scheduling strategy is to adjust the bids in proportion to the change in RPC. For example, if RPC is 20% lower on weekends, one might also shift bids lower by 20%. A similar strategy applies to dayparting, where bids can be raised or lowered according to the RPC by time-of-day.

For Enhanced Campaigns, the simplest solution is to set a static mobile adjustment factor to the average RPC ratio between devices: [mobile RPC] / [desktop + tablet RPC]. However, sophisticated marketers may want to consider changing the mobile adjustment factor on a daily basis, if significant differences in RPC trend occur across devices.

For example, below is a real-world example of day-of-week RPC trend by device. Desktop and tablet trends similarly, while smartphone trends are quite different. This causes the RPC ratio to vary by day-of-week.

RPC Trends by Device

In a simple strategy for the above scenario, you can set a static mobile bid modifier of 66% for the entire week, as this is the average difference in conversion rate between smartphones and desktops + tablets. A more complex strategy would change the mobile bid modifier on a daily basis according to the RPC ratio for each day of week.

Mobile Bid Adjustment

This method also applies to dayparting, though this means more time-intensive campaign management. Using a platform or tool that allows bulk changes and advance scheduling on bid modifiers will help to implement such a strategy.

CPC Trends Across Devices Will Become Similar

Marketing competition tends to rise and fall as advertisers shift bids or run out of budget through the day and week, which causes average CPCs to shift accordingly. Currently, these patterns vary by device, but as the Enhanced Campaigns migration completes, I predict that these differences will diminish.

Though CPCs for smartphones will remain lower with respect to desktop and tablet, the relative trend on smartphones will end up being directionally similar to desktop and tablet counterparts, as the number of advertisers making frequent changes to bid modifiers will be limited. Also, since campaign budgets are shared across devices for Enhanced Campaigns, automatic pacing and budget depletion will also occur simultaneously.

Tips For Marketers On Ad Scheduling For Enhanced Campaigns

• If you are currently using or considering ad scheduling for your campaigns, this is a good time to revisit your strategy, as Enhanced Campaigns will affect your available options.
• A basic strategy is to set to the mobile bid modifier to the RPC difference between mobiles and desktops/tablets: [mobile RPC] / [desktop + tablet RPC].
• For campaigns where RPC trends vary by device, consider making intra-week or intra-day adjustments to mobile bid modifiers in order to improve overall ROI.

The good news for marketers is that people consume more email now. The bad news is that without the right analytics, most marketers are in real danger of being left behind.

Top 100 Retailers’ Email Opens By Platform

Return Path looked at the mobile email patterns of the top 100 Internet retailers in the last quarter of 2012 and found mobile, for the first time, to be the dominant platform. In fact, more people were opening emails on smartphones than on desktop clients (like Outlook) and webmail providers (like Gmail and Hotmail) combined.

The trend shows that mobile email usage is still increasing, making it a hugely important channel for marketers. There are three things every marker should track to identify fundamental shifts in their customers’ behavior.

1. Platform – Know what device your subscribers are reading your emails on. This matters because it will determine your mobile strategy and how your email looks; and, it will influence whether and how subscribers engage.

For example, Return Path looked at email opens by platform across a variety of industries. The banking industry had extremely low mobile open rates and high opens on desktop email clients like Outlook. One can assume that people prefer to do their banking on their desktops during banking hours, and/or they prefer the extra level of security and transparency on their desktops. (It’s easier to tell on a desktop than on a mobile device whether an email is phishing.)

Additionally, people will read emails on multiple devices throughout the day. Knowing your subscribers’ preferences for various platforms can inform whether you should design for mobile first or adopt responsive email design, for instance.

2. Day of Week – Many benchmark studies show that the best time to send an email is midweek: Tuesday, Wednesday, or Thursday.

If you look at retailers’ opening patterns in the chart, you’ll notice the mobile peaks and troughs: The peaks are Saturdays. One can assume, then, that people are checking email more frequently on their mobile phones — especially in the weeks leading up to Christmas — to see which stores have the best deals.

So, we then looked to see when retailers were sending their emails and found that Fridays were the most popular; few campaigns were actually being sent on Saturdays.

3. Time of Day – Time of day is important because it can place the context of your subscriber. According to Google research, people mainly use their tablets while on the couch or in bed (35% for both).

What was surprising to me was that more people (6%) used their tablets to shop in bed than to shop from the couch (3%). So, more people are likely to be surfing and researching while on the couch than they are to actually buy anything. Smart email marketers could try sending product information in the evening, and then try to close the sale in the morning with a different email and an offer.

Your customers are unique, though, so it’s vital to look at these patterns of email and website usage for your own program.

This is really only the beginning for mobile email. As smartphones become even more widely adopted, consumer behaviors can shift in ways we probably wouldn’t imagine. But, having the right mobile email analytics and competitive intelligence at your fingertips is the only way to adapt and survive.

Google’s CPC drop has finally bottomed out in January, our analysis finds, supported in part by increasing CPCs and click share on tablet devices.

As is typical for January, this month is seeing a month-on-month CPC drop in the retail sector from the highs of December’s holiday shopping season, and a slight increase in CPCs in the finance vertical as advertisers start ramping up spend on income tax related products and services.

Overall CPC Trends

On a search engine level, most notably we see a 7% year-on-year rise in Google CPCs from 2012, the first strong lift in for Google CPCs that we have seen in some time after several consecutive quarters of falling CPCs. Bing, on the other hand, showed relative softness in Q4 – ending flat year-on-year, lower than the 5% – 10% annual inflation that has become the market standard.

The reason for the Google CPC drop as seen in the past several quarters has been the subject of much speculation in the industry. Google’s official statement was that a combination of foreign exchange rates, new ad types, and growth of cheaper mobile traffic contributed to the CPC decline.

We attribute the trend reversal seen in Q4 of 2012 two main factors:

1. Increase in both CPCs and click share on tablet devices
2. Increase in CPCs and click share for Product Listing Ads in the retail vertical.

This article focuses on the former trend, while I will cover the latter in one of my future posts.

Market Trends By Device

Looking at the retail sector in particular, CPC trending had been directionally similar on all device types up until June of last year, but smartphone CPCs have now started to diverge from that of PC and tablets. This trend is particularly apparent in Q4, where desktop and tablet CPCs rose — as is typical for the retail sector — but smartphone CPCs fell, ending with a 11% drop year-on-year.

Another major market trend we are seeing for devices is changes in search click share. While click share for both tablets and smartphones have been rising rapidly for the past several quarters, it appears that smartphone share has capped out around Q3 of last year, while tablet share continues to gain ground.

This is consistent with the fact that tablets are a younger segment than smartphones, so there is further room to scale. Also, extensive industry research has shown the conversion rates and overall efficiency of smartphone traffic is considerably lower than that of desktops and tablets, and so market competitiveness as driven by ROI will be limited as well.

While CPCs may continue to fall for smartphones on the short term, this is unlikely to have any further directional effect on overall CPC trends, as advertising click share on smartphones seems to have reached a steady state for the time being and will be outweighed by rising CPCs and market share on tablets.

Two factors may affect mobile and tablet CPC trends in the future. One is when major advances are made on mobile advertising technology that significantly improve conversion rates on smartphones, but this will probably take some time to materialize. Another is if wider adoption of tablets results in a user segment shift, causing significant changes to conversion rate.

To Summarize

• Google CPCs have bottomed out and are returning to pre-decline inflation levels for some verticals. Marketers need to counter this effect on ROI performance by continually optimizing their search campaigns.

• Tablet CPC trends are now decoupled from smartphones, as marketers increasingly split out device targets into separate campaigns. Given conversion rate differences, we expect that CPCs and market share will continue to diverge across different devices. Investing in device-specific targeting will become increasingly important in 2013 as tablets gain greater market share.

 Adobe’s director of business analytics, Sid Shah, contributed to this column. 

Continuing with our earlier list, here are four more ways a marketer might self-deceive:

#5: Confuse Accuracy With Precision

Example: “Our Average Customer is 32.47 years old”

Apparently, your average customer was born June 21. Ok, and your point is….? Do you honestly think you can measure average age of your customers to within three days (i.e., hundredths of a year)? And is there any difference between your customers that are 28 or 37? Maybe, for actuarial reasons, if you’re in the insurance business. Maybe not, if you sell sweaters.

Precision — particularly for those who are also subject to Self-Deception #1 (“Innumeracy”. See previous article.) — is often a fetish, giving a false sense of accuracy. In other words, if you can measure it as 32.47, then by golly, it must not be 31 or 33 so it must be right!

Instead, consider the difference in the meanings of the words:  Accuracy measures how close the data are to the true number (in this example, it would be the average age of your customer, rounded to the nearest year). Precision, however, measures how close the data are to each other, without regard to how close they are to the truth.

If you’re more of a visual person, imagine a typical bell curve. In this case, a bell curve of the frequency of customer ages. Accuracy is a measure of how close a number is to the top of the bell curve; precision is a measure of how spread out the bell curve is.

While insisting on what the words mean may seem a bit too prissy, self-deception #5 is a fairly common mistake, even among people who are pretty smart! So the better you get at avoiding #1-#4, the more likely you are to end up committing #5. Oh, we humans are an amusing lot!

General rule of thumb: you can’t be more precise with your calculated metric than you are with the least precise data. If you are measuring customer age in years (not months), then your average age will be in years. In fact, your math teacher likely told you to quote, instead, a median value, rather than the average, in such cases specifically to avoid this sort of self-delusion. (That was probably the time you cut class and got a detention for smoking under the bleachers.)

#6: Confuse A Local Maximum For A Global Maximum

Example: “We’ve tested zillions of variations on product images for our shoes — everyone prefers a human model wearing them! Everything else we tried lowered conversion!”

It’s easy to get caught up testing too narrow a set of variations once one factor displays its importance. What about multiple images? What about shoe treads/soles (often not seen in primary images)? Call-to-Action buttons in proximity to the image? Ancillary data (returned shoe rate) that has little to do with imagery but imbues trust? One can go on and on.

Testing is about continuous improvement; you’re not guaranteed perpetual improvement. And you’re not even guaranteed to come up with the important factors (though, I hope you do!). Get over it, and get on with it.

And, although the following comment has little to do with self-deception, I think it bears repeating since I first said it 2004: when you engage in continuous testing and you win (translation: you improve the desired outcome), then you make money. When you test and you lose (translation: you don’t improve the desired outcome), then you learn something, and that something will give you insight into testing more efficiently in the future. So you are ahead either way, and most especially ahead of competitors who do nothing.

#7: Intermix False Positives With False Negatives

A false positive: you are looking for a particular type of outcome and you identify one falsely. A false negative: you are looking for a particular type of outcome and you identify one as not occuring when, in fact, it really does. To be sure, this is less about self-deception and more about experimental bias risk — but the outcomes are usually so important to the company that not knowing about false positives or false negatives can lead you down a bad path.

This time, I’ll use a non-business example to get the point across and then circle back to a business situation. (No doubt, this will roll the comments in from both sides.)

Example: You’re in charge of security at  your airport. You’re trying to determine when one of the crazy dangerous people are trying to get on the aircraft. You’ve decided for whatever reason that you have this power to tell who is the crazy, dangerous person just by looking, so you decide to cavity search anyone at the airport wearing, say, a turban or burqa. I didn’t say it was a great idea, I’m just setting up an example to make a point.

So, a false positive is: someone who appears crazy dangerous but turns out to be just someone who got a stylish burqa from LandsEnd. The rule we used to identify dangerous turns out to not apply to this person; thus, a false positive.

A false negative is someone who appears completely safe — in this case all one has to do is wear jeans and a t-shirt to satisfy the cavity search rule — and then promptly hijacks the plane at 30,000 ft.

In both cases, we got it wrong. One was in to incorrectly find something we were looking for; the other was to incorrectly not find something we were looking for. (By the way, for those readers who did not skip out to go smoke under the bleachers that day, the false positives are referred to by the quants as Type II errors and the false negatives as Type I errors. Just FYI. )

Now, back to a business example so you can be ready for false positives and false negatives at your job:

Example: We’re running a test. Do Green Add-to-Cart buttons convert better than Blue Add-To-Cart buttons? The false positive: We run the test and it apparently shows that Green converts better; but in reality, the Blue really is the winner. The false negative: we run the test and it apparently shows our original Blue button remained the winner; but in reality, the Green really is the winner.

Two different risks that have to be handled separately. Some readers will like to know what to do in such a case, to which I’d advise: use my earlier rule of do the opposite and look for evidence against what your wished-for outcome is.

In this case, whichever test result you end up declaring the winner, consider re-running the test again. You may not need to run nearly as much traffic through it, though you might run it longer just to make up for the lessened traffic.

See if this second test confirms the first test’s results. If you’re really keen, don’t run one test across 10,000 customers, but run 10 different test across 1,000 customers. And then try a lesser test a month or a quarter from now. Always be on the look out for how can I challenge my presumptions — especially those I got from testing a while ago. This is often a great exercise in July or August when you’re not quite ready for the Christmas season to start, but you’ve got extra cycles to try a few extra tests.

#8: Confuse Correlation With Causation

First off, a quick little illustration that hilariously spoofs this point.

Example: “I changed our Green Add-to-Cart Buttons to Blue in November. Out December sales figures were way up. Ahha! Blue converts better than Green!”

Not necessarily. Maybe sales were just up because of Christmas. Maybe our new marketing guy is doing miracles on Facebook. Maybe the new Button color actually decreased sales but the Facebook efforts more than made up for that, leading to a net plus for us. Maybe a lot of things. We just don’t know enough to be assigning credit.

Causation is about one thing causing another. And since our everyday experience of time flows in only one direction, that means the causer happened before the, hmm, causee. In the example above, if the changing of the Button color really caused sales to go up, then the change must’ve occured first. In fact, there’s a old logic fallacy, post hoc ergo proptor hoc — so old that they came up with it in Latin when Latin was a live language — which means roughly after, therefore because of. Just because something comes after something else, doesn’t mean the first caused the second.

Correlation is much looser. Two things are correlated when they tend to move in the same direction of change somewhat in accordance with each other. This doesn’t mean that one causes the other. Nor does it means that it doesn’t. It’s just that there’s some sort of relationship in how they tend to move. For example: Rock Music Quality correlates well with US Crude Oil Production. Or US Highway Fatalities correlate well with Fresh Lemon Imports from Mexico. These are correlated, but there’s no causation. (Unless you think the highway deaths are caused by slippage on lemon juice!)

But here’s another one: Men who smoke cigarettes and the incidence of lung cancer among men. Turns out these are fairly well correlated, with a lag of about 20 years from when the guy starts smoking to when the cancer rates start going up. Do you think there’s possibly a causation here? Yeah, probably.

Now, it’s true that the phrase, correlation doesn’t imply causation, is oft-bandied about. In fact, it’s over-used in the extreme. But an important point to keep in mind is that all causation implies some sort of decent correlation (if you’re measuring what counts); whereas, high correlation in and of itself just means, well, that there’s a high level of correlation.

Causation and correlation are not opposites; causation simply includes a way to give credit to a causal event. If you keep the distinction in mind, you’ve got a powerful way to come up with new ideas for testing more broadly or more deeply. Ask yourself: is there anything (other than the correlation) that causes me to believe that A caused B? This gives you a powerful way to come up with new ideas for testing more broadly or more deeply.

Well, that’s it. It’s been a long two articles! But I hope you’ve learned something that will help you keep sharp in your testing efforts: the metrics you use to measure success, and your self-assessment of what you think you know for sure.

Which ones of the self-delusions do you feel you’ve perpetuated on yourself in the last 90 days?

With apologies to Sir Walter Scott for my mangling of his most popular turn of phrase, I’d like to take some time now to discuss how people −and particularly people involved in metrics and evaluation of “how did we do this week/month/quarter?” − are prone to self-delusion.

Along the way, I run the risk of insulting you, but I see that as a necessary risk when confronting a topic like this. It takes a healthy sense of self-worth to say “holy crap! I’m a dufus!”

Now, let’s start off with some assumptions. First, that (most) people are honest, and don’t really intend to deceive themselves or others. (If you’re doing it intentionally, then bad karma on you, but I leave your fate to another writer.) This leaves with us two groups: the honest self-deceivers who do so consciously and those who do it unconsciously.

Conscious self-deception? Surely not!  Yet, how else do we explain common phrases for just this very thing:

• “I don’t want to know”
• “It’s got nothing to do with me”
• “Don’t make waves”
• “Look the other way”
• “Nothing I can do about it!”
• “Let sleeping dogs lie” (personally I love the double entendre here)
• “Ignorance is bliss”
• “Brush it under the carpet”
• “Why didn’t I listen to my intuition?”

All of these phrases are used in situations where the facts are uncomfortable and we consciously choose to ignore them. We make believe the situation is different, all the while knowing it isn’t so. I like to call this “quantum shifting” into an alternate universe where the unpleasant is no longer so.

Many of the situations where you’d use a phrase like the above are ones in which you feel you may have little or no control over things. So the natural human reaction is to ignore and hope that things change.

Is there a way to minimize self-delusion? Actually, yes, there is somewhat of a process you can use, though with varying levels of success.

• Admit that it occurs; realize, as a human, you will do it again
• Recognize when it’s happening
• Look for the opposite
• Friends don’t let friends self-delude: have a buddy who is willing to level with you

Again, I’d stress that the more compelling the bad news is, the more likely one is to steadfastly hold onto one’s convenient alternate reality. Let’s all group-self-deceive right now and chalk that up as “part of the charm of being human.”

Eight Common Self-Deceptions

Now, I can’t help you with that last buddy part. Nevertheless, here are eight common ways we unconsciously self-deceive using stats, ignorance, bad logic, or other wily human tricks. I’ll present them as encouragements − makes them funnier that way − so you can recognize when they might occur and be prepared to challenge yourself.

To encourage your own self-deception, it helps if you let yourself:

#1: Be Innumerate

Example: “what’s the big deal going from 2% conversion to 3%?? That’s only a 1% lift! We want more!”

Uh, no. Going from 2% to 3% is a 50% lift.

As the saying goes, “Innumeracy is a serious problem which affects 8 out of every 5 people.” I’m listing this as #1 on the self-delusion hit parade, because I see it the most often, and the people who practice innumeracy really do believe they “get math.” In some ways, you can argue this is easy to fix. In other ways, you can argue “you can’t fix stupid.” You decide which group you see yourself in.

#2: Start With your Wished-For Outcome… Then Look For Supporting Proof

Example: “Our Customers love us! We had 317 positive reviews on Yelp last month” (conveniently ignoring the 2,182 negative reviews and a poor Net-Promoter score)

I call this one the “Creationist Ostrich” technique. (Oh boy, I can see the comment box filling up now!) You start off in a comfy zone, and you really, really, really want to stay there, so you ignore all the evidence that contrasts with your view. Usually, this is because the alternative suggested by the evidence is bothersome in some way, and the more deeply bothersome, the easier it is to just dismiss, hold out with your head in the sand till 5 p.m., then go grab dinner. Problem solved!

Instead, under the part of the process mentioned above as “look for the opposite,” consider when you have a wished-for outcome that the best thing you can do for yourself is to try to find evidence against what you’d like the outcome to be. At least, if your goal is to uncover truth.

Virtually everyone involved in metrics, analytics, and testing-driven improvement online does, in fact, have a wished-for outcome: “get higher conversion,” “increase revenue,” “decrease cart abandonment.” Every metric you have for success implies you’d like more or less of that metric, so you always have a wished-for outcome at hand.

So, you’re subjecting yourself regularly to this sort of self-deception. Cut it off from the start by always trying to prove yourself wrong. Give yourself permission to be wrong, by constructing your tests in such a way that before you even run the test you can say, “heck if I see such and such from the test results, we know we’ve completely missed the mark” − and then look for such-and-such.

#3: Take Credit For Forces Beyond Your Control

#4 Settle for Industry Averages

More To Come

But what can you do to get your fans to share more? Based on research conducted at Buddy Media between December 18, 2011 and February 11, 2012, of more than 76,000 posts from 1,200+ pages, we have some tips to increase the virality of your Facebook content.

Post Visual Content

Photos and videos are shared more often than status updates and links, yet brands are missing a tremendous opportunity to improve sharing of their content. In fact, videos are shared most often, yet brands post videos the least. The moral of this story: post more visually appealing content and see the number of shares increase.

Make It “Cute”

Of the posts we analyzed, 50% of the top shared posts contained a picture of a cute animal. Content self-described with keywords such as pretty, beautiful, cute and adorable led to a share rate 49% above average.

The moral of this story is to post visually stimulating content that appeals to the emotional sensibilities of your fans. Not every brand can post a picture of a dog licking a baby’s face. But the more content you can create that evokes an emotional response from your fans, the more your fans will share.

Use A Call-To-Action

Using calls-to-action prompting users to share leads to 77% more sharing than average. Despite the overwhelming evidence for using share calls-to-action, we found fewer than one percent of brands are actually prompting their fans to share.

This is a huge missed opportunity for brands. Most brands are simply not giving their fans enough direction to increase the number of shares.

It seems obvious: if you tell people to share your content, they will listen and act. Yet most brands are not taking advantage of the enormous opportunity to spread content beyond their fan base.

So if you’re looking to increase sharing on Facebook, here’s a recap:

1. Post visual content. Photos and videos are shared more than status updates and links.
2. Make it “cute.” Aesthetically pleasing content is shared far more than the average post.
3. Use a call to action. Ask your users to share and they will listen.

For social media marketers, this means that it is more important than ever to understand what content will resonate with target groups and which will fall flat.   Without good ways to identify the content preferences of target groups,  the only option is to throw marketing mud against the wall and hope that it sticks — an approach that is costly, more likely to fail, and may damage the credibility of budding social media programs in the eyes of senior management.

Measuring the social engagement elicited by different types of content can help social media marketers pinpoint what is effective — and what isn’t worth precious time and budget.  Here are three ways to do it:

1.) Track Cross-Channel Social Interactions

Some social sites, like Facebook, provide engagement data for each content item posted. This is important information because it helps you determine what content works best within the closed walls of a specific site or community.  But how can you determine which content works best overall, across the multiple social channels in your mix?

First, develop a system for keeping track of content.  You may already have this if you are using a social media management tool (some provide a content ID number), but an Excel spreadsheet will also work.  At a minimum, your spreadsheet should include an ID number for each content item.  You can also include additional data, such as date released, post title, and content type (link, photo, video, blog post, etc.).

Next, create a rubric.  A total count of engagement actions (sharing, commenting , liking, etc.) may be enough for your purposes, but if you want to distinguish between different types of engagement, you’ll need to create a rubric for organizing metrics into consistent categories.  For example, metrics like Facebook comments, Blog comments, and Twitter @replies can be grouped into a category called “communications”.   Metrics that represent different types of sharing activities can be grouped into a “distributions” category.  You can come up with your own categories and assign metrics in way that makes sense for you.

Your spreadsheet will look something like this:

This DIY method of tracking engagement takes time and elbow grease (so you may want to limit the analysis to a specific time frame), but it can go a long way toward helping you determine which content topics and formats to focus on.

One word of caution:  This will work best if the content you are using is highly targeted to the audience you are trying to reach.  If it isn’t, engagement levels may not be a good indicator of what your audience wants.  For example, a funny picture posted to your Facebook page might be widely shared by people who are not interested in your product or services.   Consequently, high engagement numbers for this post would not necessarily indicate that this type of content works well with your targeted groups.

If you question whether engagement is occurring among the right people, web analytics data can provide important clues…

2.) Analyze Traffic And Site Behavior of Visitors To Shared Pages

If engagement and sharing in social channels drive traffic and high levels of interaction with your website, it’s a good sign that the right folks are engaging.

True, not all of the content you share in social communities links back to a page on your site, but if it does, measuring traffic to — and interaction with — shared pages can yield valuable insights into which content is worthwhile.

The new Google Analytics Social Reports makes this much easier by providing data on traffic to shared pages, visit duration, and even the number of times the page was interacted with out in the social world (“Data Hub Activities”):

Two important notes about the new social reports: First, Facebook and Twitter are not included in the Data Hubs tracked by GA, so for many social media marketers, the total number of interactions, or Data Hub activities shown will be under-reported.

Second, don’t focus solely on traffic to shared pages.  Even if traffic numbers to shared pages seems low, remember that quality can trump quantity in the ROI equation.  One well-qualified visitor who is eager to consume your content is worth far more than ten unqualified visitors who leave the site immediately.

Bottom line: content that leads to deeper engagement within the site is successful content, so pay attention to visit duration and pages visited.  (You may also want to set up Advanced Segments in GA to further analyze the behavior patterns of visitors to shared pages.)

3.) Calculate Interactions Per Referred Visit

So far, we have looked at two ways to judge content performance:  engagement activity in social channels and website activity generated by shared content.

These two methods can be combined by calculating “Interactions Per Referred Visit”, a metric that captures social engagement relative to traffic.  Interaction Rate is calculated by dividing total interactions or engagement actions in social channels by visits to the shared page.   You can add this metric to your spreadsheet:

Ultimately, how people respond to your content can’t be fully captured by any single number (or even a set of numbers).  However, as is often the case, analyzing the data is an excellent place to start!

Are you using engagement tracking to identify high value content?  If so, what techniques and tools do you use?