September 2009


29 Sep 2009 01:34 am

twinsOne of the ultimate excuses for not measuring impact of Marketing campaigns is: "Oh, that's just a branding campaign."

Admit it, you've heard it.

I suspect you've even used it liberally!! : )

Before we go any further I must clarify that I love branding campaigns just as much as the next guy.

I love campaigns that Visa runs. I love watching the IBM ads (with the Linux kid perhaps the best of the lot). I loved the I'm a PC ads from Microsoft (and I am a proud PC!). I loved the Wario Land: Shake It ad from Nintendo on YouTube (now that's creative!). I love a good billboard ad, Budweiser does good ones. My absolute favorite branding campaign of all time: Think Different.

I could keep going on.

The common theme through the above campaigns is that their primary purpose is "branding". The hope is by connecting with you, or interrupting you, a lasting impressing, a feeling, might be left in you so when its time to get a credit card you think of Visa and not MasterCard, when it comes time for hiring consultants for a multi year project you'll choose IBM and so on and so forth.

All well and good.

Here is the minor problem.

There is a very tenuous connection between these campaigns and outcomes, they are for the most part faith based initiatives. If supported by "data" then it tends to be of the most fragile kind (usually the the fact that the CEO saw it during the Super Bowl and felt happy suffices as actionable data).

None the less they persist.

Online it does not have to be that way.

It is criminal not to measure your direct response campaigns online. I think we have established that. I also believe that a massively under appreciated opportunity exists to truly measure impact of branding campaigns online. Paid Search or affiliates or email or display or YouTube or whatever channel you end up choosing.

[Oh and don't tell me that your "branding" campaign is to increase "engagement"! Remember: Engagement is not a metric, its an excuse.]

dont tell

The Top Secret Hidden Never To Be Reveled Come Hell Or High Water Key To Measuring Branding Campaigns:

Answer this simple question: Why %&#$!^ are you doing the "branding campaign"?

Every campaign, and in turn website has a purpose. All you need to do is figure out what the purpose of your campaign is, no matter how outlandish (or childish) your goal.

The typical focus by companies, and the creative types in their employ, is to simply focus on figuring out what you are doing to do in the campaign.

I am recommending that they hold their horses / put their pants back on / slowly sit down in their over-stuffed chairs. You too!

Figuring our what you are going do do with your campaigns can come after you figure out why you are doing these campaigns. No, not just because you have money or because that is how things have always been and absolutely not because someone (a HiPPO!) asked you to.

Once you know the desired outcome you'll be surprised to learn all the wonderful measurement possibilities that await you online, things that would be nearly impossible offline. [The web rocks!]

Measurement Recommendations for Desired Branding Outcomes.

In order to help you make the leap in the rest of this post I want to share the most common outcomes I have heard associated with branding campaigns, and my recommendations as to what they should measure for their brand campaigns online.

My hope is that this bushel of ideas will spark your own creativity when it comes to measuring your campaigns.

Outcome #1: To attract "prospects" / new customers.

This is perhaps the most common desired outcome: "I am doing branding campaigns to attract new prospects to our website. They will come, they will be wowed by our glory, they will immediately convert."

It is not very hard to measure these campaigns.

all visits comparison with new visits

Measure the change in the percentage of New Visitors to your website, its the orange line in the graph above.

Ideally you'll measure the number prior to your branding campaign, say Feb 2009, and then you'll measure it again during your campaign, March 2009. See if you were able to get more traffic to arrive at your site, and if they were Existing Visitors or New Visitors (hopefully measured with a first party cookie in your website analytics tool).

For good measure, just to be extra sure you'll segment out the visitors who come by clicking on your campaigns (display/banner, YouTube, rich media, whatever), and see how many of them were truly new.

At this point there is no expectation that any other outcome was delivered, just a visit by someone who had never been to your site before. A fairly low bar.

Outcome #2: To share your business value proposition.

You are a news site like The New York Times or you are a non-profit like Idealist or you are the team running Google Maps.

The goal of your campaigns is to simple share your unique value proposition with everyone. They'll be impressed enough to come visit your site and then do so repeatedly.

The ideal metrics for this desired outcome are Visitor Loyalty & Visitor Recency.

branding campaign visitor loyalty report sm

(Click on image for a higher resolution version.)

The data in the above report shows how frequently during a time period do the website's visitors visit the website. In the Before version you can see that most people, 69.79%, visited the website just once. In the After version, when the branding campaigns were running, only 63.25% of the visitors visited just once. Which means atleast 7% of the visitors shifted to visiting more than once.

You can credit the branding campaigns with that shift (if that is all you were doing). Better still you can segment the traffic from the campaigns and validate that hypothesis.

If people were impressed enough with your value proposition and visited more often the the brand campaign was a success.

Another good idea is to measure segmented Visitor Recency.

visitor recency segmented measurement

In this case the analysis will try to judge if the traffic acquired by paid search branding campaigns is visiting my website more frequently in any time period, when compared with other segments of traffic (in this case I am comparing it to All Visits).

Outcome #3: To impress people about your greatness and buy more.

I wanted to put this as #3 because if it were a "conversion" campaign then it would not be a "branding" / feel good campaign.

But there are certainly campaigns that you run to prop up your brand that will entice people to buy more from you. If they were only going to buy underwear then now they'll also buy a pair of shoes and headphones.

I recommend segmenting the traffic and measuring revenue lift but also measuring the average order size, if you did your job right then that latter number should be higher.

google analytics ecommerce report

In this case our Yahoo! display campaigns did wonderfully in terms of conversion rate, but not in terms of the major goal of the campaign – sell more stuff.

Another thing people forget is to measure the overall impact, beyond simple conversions. Sure measure it as above but it is also good to marry up the qualitative data and measure Task Completion Rate using a onexit survey tool (use a free one like 4q from iPerceptions).

website task completion rate

You can accomplish these goals:

    ~ Get an optimal understanding of what kind of people you ended up attracting to your website (look at primary purpose & distribution).

    ~ Were these people, even if all you wanted from them was to buy from you, able to complete their tasks.

In the above case that is clearly not true.

Perhaps your Average Order Size is not great because only 44% of the people who came to buy, as a result of your branding campaigns, were able to complete their task!

You would have fired your ad agency for a crappy campaign, turns out they did their job well but it was your website that stunk. Perhaps someone in your team needs to get fired? Perhaps you? (Just teasing!)

Note how marrying the Qualitative and Quantitative data can be helpful and identify true points of failure / success.

Outcome #4: To whisper sweet nothings to drive offline action!

Most commerce / love / stuff still happens in the real world and many many companies use various online marketing channels to drive people to take offline action (make purchases in stores or via their phone channel, show up for a woman's rights rally, meetup at a concert etc).

You can measure the impact of these campaigns right on your website, using any onexit survey tool and by applying some delightful regressions on your data. You can compute two important metrics:

Likelihood to Recommend / Brand Lift

likelihood to recommend brand lift

You can measure this at an aggregate level, or you can measure it just for your campaign traffic.

It helps you understand what was the brand lift, positive, as a result of the person's complete experience (your campaign, plus your website).

Likelihood to make a Offline Purchase / Action

likelihood to make a offline purchase

Simple right? Well it takes some planning but it is not that hard to measure (and if you get a decent sample size then you can also segment this data easily by source of traffic, brand campaigns, and show a causal relationship).

Phone Calls / Conversions Driven from Website

Another wonderful way to track offline impact of your campaigns is to use unique phone numbers with your campaigns (either on your display banner ads or on your website).

You can track the number of phone calls made to your call center, by campaign (or keyword or whatever) and if you have a integrated IVR then you can also track conversions / sign ups from those campaigns.

Companies like Fresh Egg in the UK or Mongoose Metrics and ClickPath in the US, amongst many others, provide these phone call tracking solutions.

[Bonus reading material: Multichannel Analytics: Tracking Offline Conversions. 7 Best Practices]

Outcome #5: To break through the noise / make an introduction to your business.

Very often when you run branding campaigns your goal is simply to introduce your business (like we are trying to do with Market Motive, our start up that provides certification courses in Web Analytics, SEO, PPC, PR etc etc).

A common mistake in this case is to simply focus on one outcome. If you are running a branding campaign then it is likely that you either have a very soft call to action or, more likely, you have a very general "our business is magnificent" message.

My recommendation is to quantify the online impact of these campaigns by measuring both the Macro & Micro Conversions.

For example if I were to measure impact of branding campaigns for this blog (remember it has no ecommerce of any sort) then this is how the report would look:

micro conversions google analytics

My macro conversion is to add to my current total of 27,300 RSS feed subscribers.

The above report shows the overall add to that number in this month but by segmenting my Yahoo! "Avinash is awesome" display campaigns I can see how many "macro conversions" occured.

But that's just one part of the story.

I will also measure the "micro conversions" (goals 1, 3 & 4) to get a more complete picture (for example note the large percentage of "Loyalists" that ended up being from my brand campaigns!).

This methodology can be applied to any business.

For example if I were in-charge of campaigns for OfficeMax, I would measure ecommerce conversions but that's just a sideshow for these types of campaigns.

For full impact analysis I would measure:

~ # of leads received

~ # of requests for catalogs

~ applications for OfficeMax branded credit cards

~ increase in facebook fans and twitter followers (hopefully relevant followers!)

~ # of coupons printed

~ # of free downloads

And so on and so forth.

In the case of Market Motive for our branding campaigns we will measure outcomes by focusing on # of sign-ups for the Master Certification program but also the # of people who sign up for the free webinars we do all the time, the Ask Us inquiries, the trial memberships, # of sample tutorial videos viewed etc etc etc.

When you are trying to break through the noise you'll take any measure of success to detect a signal, use the macro & micro conversion mental model.

Outcome #6: To destroy your competition.

Very common goal for many marketing campaigns. Show how awesome your brand value is and directly or indirectly show your competition in poor light.

DirecTV does it well, though now Dish seems to be totally trashing them atleast during NFL games [Look at them go at it: DirecTV trashing Dish vs. Dish trashing DirecTV.]

Apple of course is a master at it (though sometimes they can be mean). CPG companies are perhaps a bit more subtle about it, but their goal is clear: get you to buy their brand of chicken bouillon or diaper or ageless blush.

There are a number of wonderful metrics you can use to measure online success of such marketing campaigns.

Share of Search.

The first thing you want to measure is how much "share of voice" you have "stolen" from your competitor. One great way to do this is to measure Share of Search.

If you have done a great job of branding then the number of people looking for you (searching for you) should go up. Oh and not in your Site Catalyst or WebTrends reports! Rather in the "ecosystem reports" you can get at a competitive intelligence tool…

share of search broad match compete

www.diapers.com has a 3.64% "share of search" prior to the campaign. What was it after?

Did they make a dent in the universe?

I am using Compete for the above report. When you use it you'll notice that Target is #10, Babycenter is at #15 (with a 0.77% share which seems looooow!). If you run branding campaigns for either company you now know how you'll measure success.

You can index your performance for your campaigns, and against your competitors.

If you use broad-match like I did above you get a "category" view of your performance. If you use the exact match report. . .

share of search exact match compete

. . .to get a better idea about about "brand" performance (and find new competitors – facebook anyone? :)).

You can do this for pantene, shampoos, styling treatments, pro-v, 2-in-1 shampoo + conditioner… the world is your oyster.

This analysis also helps you understand how well your offline branding campaigns are doing online.

For example I don't know of airline companies that run more television campaigns than Southwest Airlines.

Yet currently for queries like cheap tickets, cheap airline tickets, cheap flights etc www.southwest.com does not show up in the top 15 in "share of search" reports, in some cases not even in the top 25. And that has not changed in the last few months (even with the barrage of new TV ads).

The TV ads are perhaps super productive in driving people to the phone or perhaps directly to the site, both desirable outcomes. But they are certainly not working in getting people who are looking for airline deals and searching for them to go to Southwest.com.

Traffic Differentials

If your campaigns are successful you'll know it from your Site Analytics tools like Google Analytics or Yahoo! Web Analytics. But in the grand scheme of things did you have an impact?

Use tools like Google Trends for Websites (or Compete or HitWise) to analyze your performance…

google trends for websites hilton starwood hyatt

In the report above if you (Hilton Hotels) ran your campaigns in March 2008 (purple arrow) then you managed to accomplish nothing. Notice the competitive trends?

If you ran your magnificent branding campaigns in Jan 2009 then buy yourself some beer and dance around because you can see, against competition, you clearly narrowed the gap (black arrow).

Of course you would not stop at the simple analysis above, that's just a start. You can export data into excel, you can segment it by Geo and ensure the lift is where your campaigns were targeted, you can segment by demographic and psychographic visitor attributes to see if you got the right kind of people.

Using Share of Search or Traffic Differentials are just two of many ways in which you can measure if your branding campaigns are indeed crushing your competition. There are many other analyses you can do, there are many other tools you can use. Don't give up, look.

Outcome #7: To emboss your brand into someone's skull.

As Marketers we try and do this all the time as well.

You say Jeans, and people say Levis.

You say Jonas Brothers, and people say Pink by Victoria Secret (I kid you not, yes I was surprised, I have research to back this up!).

You say online search, and people say Bing! No, not so fast! Google!! Ok maybe it's just a matter of time. :)

You catch my drift.

Branding campaigns are particularly effective at "embossing" brands into your psyche with the goal of improving unaided brand recall.

The challenge of course is how do you measure this elusive, but very desirable, outcome.

I have two suggestions.

Primary Market Research

In the online world we don't make enough use of primary market research, and that's a shame.

Field surveys, focus groups, interviews etc can be used very effectively to gain a indepth understanding of your customers and their influences (hopefully channels and methods you use to influence them show up in the answers!).

There is a ton of math and rigor involved in these studies that helps you get a great understanding of your audiences, even with small enough sample sets.

Plug into the Database of Intentions

I love that term: "database of intentions".

For now atleast search is used by many people as they seek information online, and that allows for this data to reflect intent, rising and falling trends, preferences etc.

You can use Insights for Search for this type of analysis.

I am a Assistant Senior Vice President of Brand Marketing Campaigns at Orbitz. I have been spending my lovely marketing dollars on tons of TV campaigns (See: Orbitz Golfers). Oh and a smidgen on online campaigns as well.

So what was the impact, when people search for "hotels" do they think of Orbitz? Here's the data you are looking for…

related searches hotels

In the online database of intentions Orbitz is not on the horizon.

The data on the left is important because it tells you what people search for when they look for hotels.

The data on the right is killer. It shows which terms (hence brands, sites, properties) have risen the by the most statistically significant amounts. This is fantastic because it mines the data that is below the surface and brings the movers and shakers forward.

Some of Orbitz's competitors show up there, their marketing dollars seems to be working well in improving the likelihood that when people are doing category searches, hotels here, that they would look for expedia, priceline, hotwire, etc etc.

But I am not going to give up, I just started running these massive tv campaigns a few months back. Let's see the data for that!

relate rising searchs hotels us last 90 days

Hmm…. the "winner" here seems to be Marriott from the hotels category in the last 90 days. Good for them, sad for me.

Let me hasten to add that it is quite possible that the desired result of these offline (and online) branding campaigns was to get people to go to the site directly or call Orbitz on the phone.

The first premise you can measure easily, how is our website traffic doing when we are running all these campaigns…

expedia orbitz priceline compete visitor data

Two years of data. Expedia is green, Orbitz is blue, Priceline is orange (query on compete). It does not look too good for me (and if that was not enough Priceline just crossed me for the first time in my history, boo! boo!).

[It is important to point out I am simply doing outside-in analysis, a sport always fraught with risk. You on the other hand work at Orbitz and will have the tribal knowledge to make sense of this data better.]

For the last piece of analysis to measure unaided brand recall analysis, I'll try is to correlate my brand marketing spend with the number of phone calls to 1-888-656-4546 (and keep my fingers crossed that I'll see a massive spike in phone calls from the campaigns because, on the surface, it is hard to detect a impact directly on the site).

You can do this type of analysis for anything.

Let's say Victoria's Secret has indeed been heavily spending on branding campaigns for Pink by Victoria's Secret in the last 90 days. I can look at this data to see if the brand Pink is amongst the fastest rising by those people who look for Victoria's Secret "stuff"…

victoria secret unaided brand recall

Not so much. Why?

Now if I were the Brand Manager for the Jessica White line then I would rejoice! While Jessica White is not in the "top related searches" category yet, it is rising very very fast as a result of my campaigns! Yes!

(Mental note to check later: What the heck is "victoria secret application"? Sounds dicey!! :)).

So there you go. A portfolio of seven strategies that you can use in the ultimate quest for any online marketer / analyst: measuring branding campaigns.

I hope they spark your creativity and lead you to finding even more innovative solutions to your unique challenges.

Buena Suerte!

Ok your turn now.

Are there other outcomes you can think of for your branding campaigns? Can you think of other ways to measure the seven outcomes mentioned above? If you have tried one of the above strategies did it work? If it did not, why not?

Please share your feedback / learnings / critique / kudos.

PS: Like this post? Perhaps you'll consider ordering my * new * book: Web Analytics 2.0.

PPS:
Couple other related posts you might find interesting:

14 Sep 2009 02:07 am

Yes, books with a s. : )

It is with immense excitement that I am sharing the news that I have just finished writing my second book!

Web Analytics 2.0:
The Art of Online Accountability & The Science of Customer Centricity

web analytics 2.0 online accountability customer centricity

It is a long title ain't it? The good news is we are going to refer to it simply as Web Analytics 2.0.

In this post I wanted to share thoughts about the book, the process of writing it (and doing three rounds of edits!) and outcomes.

The Background

Since mid-2008 Willem Knibbe, my wonderful Acquisition Editor at Wiley, was very kindly encouraging me to update my (best selling!) first book, Web Analytics: An Hour a Day.

The "problem" was the book continued to sell at a nice rate and I was not sure what to update because 90% of the content was still current and relevant.

Still there was a lot of new stuff I had written, new models I had developed, new and more advanced techniques, new problems we were dealing with in the world and so on and so forth.

That lead to my proposal to Willem to write a new book that would use Web Analytics: An Hour a Day as a starting point. The second book would be an advanced book that would allow the first book's readers to truly become Super Analysis Ninjas, and for those that had not read the first book to have the finest possible immersion in web analytics.

And that's just what Web Analytics 2.0 is.

The 2.0 Book

The book's core philosophy is based on the framework you have seen me talk about on this blog. . . the quest to answer four key questions: the What, How Much, Why, and What Else. . .

web analytics 2.0 3

The awesome thing about writing a advanced book is that I can start with a bang! No history and what not. It starts with: Here is how your world should look like and this is why its important, now let's get down to business.

That's by page 9. : )

And then it just keeps kicking it up a notch. Bam! Bam! Bam!

Like the first book this is not a book about Omniture or Xiti or Google Analytics. It is not a "press this button in the tool and then press that one" book.

It hopes to be brain food.

Here is how you should think. Here are the traps to avoid when picking key performance indicators. Here are the core analytical techniques you should apply. Here are a bunch of reality checks. Here is how to embrace outcomes, regardless of the size of business you have. Here is how to achieve higher highs with testing and by listening to customers (literally). Here is how you leverage your competitor's data. Here is how you becoming a true Analysis Ninja (step, by step, by step).

And none of that is even close to the coolest part of the book (see why I am so darn excited?).

There are so many topics I deal with each day that I have not had time to write about on the blog, all the things I practice all day long in the five jobs I hold.

light bulb 1

The book gave me the impetus to write all that down.

So there are complete sections in the book that teach:

Why tracking the social web is such a massive problem.

How to measure success of blogs.

Meaningful non-crappy twitter analytics.

Mobile analytics! This was so much fun to write about.

Measuring rich applications whose primary usage happens with no internet connection.

And more such things.

But you might end up buying the book simply for Chapter 12, it covers two things that I think will rock your world:

1. Multi-touch campaign attribution analysis (dissected and presented in a way like you have not seen it any where, I think)

2. Multi-channel non-line analytics (practical tips, best practices, unique stories to inspire you)

Even after all that I was not completely satisfied. : ) There are two more new things to end the book. A complete chapter on how to start, nurture and advance a career in web analytics. The last chapter of the book is on how to overcome the hardest challenge of it all: creating a data driven organization!

The Writing Experience

This was a very hard book to write, in many ways harder than Web Analytics: An Hour a Day.

That's partly because this time around I had my full time job, my work with my start-up Market Motive, my advisory roles in three companies, my world travel to support my professional speaking career, my blogging (the only thing that suffered), and of course my family.

It is difficult to find time and energy to write a book with all that (and impossible without a magnificent wife who takes on three times a normal human's load to support you!). Especially to pull the writing and three rounds of edits in four months!

It was also hard because this is a much more advanced book with so many topics on the bleeding edge. It is hard to make sense of it all and understand it enough to apply a reality filter and then write something that people can apply today, and use for a very long time.

And yet it was a lot of fun to write this book.

web analytics an hou a day photos

I think that's primarily because with the first book I had no real sense for what the book would become, who it would impact, how far it would go.

This time around I have a much better sense for all that.

So many of you have written to me about all the ways the book has touched your lives. As I wrote this book that was constantly at the back of my mind. It pushed me to work harder and do better because I realized all the places it would go, all the people who will crack it open, all the expectations it had to meet.

I had this visual of all the people who might buy this book and how in some way something I wrote could have an impact on them. That was pressure, but it was also fun.

The Second Little Book That Could

Some of you know that my wife Jennie and I had decided that we would donate all the proceeds from the first book to charity. We had chosen The Smile Train and Doctor's Without Borders and split 50% of the proceeds between each.

My hope was that Web Analytics: An Hour a Day would sell enough for us to donate the $10,000 advanced we had received.

We have thus far received, and donated, 18 months worth of royalties from the book, approximately $70,000 (!!).

Not in my wildest dreams had I imagined that! And there is no way that we could have afforded to donate that much money.

In a very small way this blog and the book have helped other people in our lovely world. It has been an extremely gratifying experience for us.

With Web Analytics 2.0 we have decided to do the same again.

charity logos

100% of my author proceeds from the book (and all the amazon affiliate sales) will be donated to The Smile Train and Ekal Vidyalaya.

Ekal Vidyalaya runs schools in remote locations that reach the poorest of the poor children in India. Without Ekal these children would have a very limited set of opportunities in life, if any.

When the going got really tough with this book the thing that kept me going was to produce a book that would have a big impact on people who buy it and a small impact on the charities Jennie and I choose.

The 411

The book can be pre ordered on amazon now, if you are so inclined.

It will be released mid-October 2009.

Wish me luck.