Non-Line Blogging

American author Will Rogers is attributed with this rather caustic one-liner from the 1930’s.

"When the Okies left Oklahoma and went to California, the average intelligence of both states went up."

It has since been used by many people, including  Robert Muldoon, the former New Zealand Prime Minister, about his countrymen migrating to Australia.  I’m proud to say that I am personally responsible for improving the average health of many cancer patients – another example of the “Will Rogers Phenomenon”.  A while ago I was at the un-healthy end of the “Stage 2” cancer segment, but after a couple of scans and no change in my overall condition I was popped into the “Stage 4” cancer segment along with some really ill people.  In that single action I had improved  the “average health” of both segments . How barmy is that?      

These examples show us that averages can be misleading and confusing. So, what do we get when we open up our web analytics reports or our email campaign dashboard?  Averages!  I’d like you to stay awake long enough to convince you that averages are a bad, bad thing in digital marketing and here’s why...

“Utterly Useless Average Metric” Number 1 – Site Conversion Rate

Let’s take an “average” site conversion rate.  It’s probably 3%.  That’s average.  But just a couple of simple steps gets us to a better place – how about conversion rate between new and repeat visitors, or conversion rate by source of traffic? A couple of clicks and you’re somewhere exciting.  As Avinash Kaushik would say “BAM!! Context, Baby!”  You find that repeat visitors are probably 3 times more likely to convert than first-timers, or that people coming from your Facebook invitation-only group convert 10 times better than people clicking on banners.  So get into some pretty basic segmentation to give some meaning to what you’re doing.  Here’s how Matthew Tod of Logan Tod demonstrates it.

See, conversions from banners are much worse than “average” and conversions from previous customers is much higher than “average”.  In short, it looks like people don’t buy on their first visit from any source, but will probably buy if they are returning visitors or past customers.  We could drill further and find out what products drove highest re-purchase rates, or what creative/media gave the best conversion rate....there may be some winners lost among the “average”.  More segmentation; more context; more correct actions taken!

“Utterly Useless Action” to improve average site conversion rates? Turn off your marketing.

I’d like to thank Matthew Tod for this one, too.  Promotional activity will drive prospects to your site and they are unlikely to buy...but some of them will.  Trouble is, they drag down the average conversion rate.  So switch off all your campaigns and watch those averages soar. The only people who will come to the site will be your family, close friends and past customers - and they love you.  Average conversion much better, sales volumes plummeting. 

Be careful what you wish for.

“Utterly Useless Average Metric” Number 2 – Email Click-through rate

Here is a list of average metrics from Mailchimp. 

Email marketers feel a sense of “market sector envy” creeping in as they look jealously at how big the averages are elsewhere.  Art has a 26% click-through rate and Arts_Music has a feeble 5%.  And what about Education getting a whopping 181% click-through rate?  Does the average relate to how good they are compared to other sectors, and if you have a low average are you a bad, bad marketer? 

The trouble here is that we are worrying about “soft” outcomes – even if we do improve our average open and click rates we will not necessarily improve our revenue, or net yield. 

If you want to improve your click-through rate you could probably make everything you wanted people to know about on a web page.

"Find out how many points you earned last month - click here"

Bingo!  A quick and easy way to improve average click-through rates, and here's another...

“Utterly Useless Action” to improve Average Click-Through rates?  Don’t mail non-openers.

So let’s really improve the average click-through rate.  Below is a graph showing what happens when you mail lots of "emotionally un-subscribed".  If the number of customers (who love you) is low, then the average click through rate will be, well, average!

But stop mailing people who have not opened for over a year as part of the same campaign.  Drop them into a different programme and the composition of the file will change.  You will have fewer non-clickers and your average click through rate will shoot up like this:

I have been banging on for years about “open” and “click” rates being “soft metrics” and I’m sure I’ll blog about that again soon.  In the meantime let’s all just agree that improving “average” click through rates is an utterly meaningless objective. Getting the “emotionally un-subscribed” to click on something is good, but judge yourself by what “outcomes” you deliver and then you’ll be adding value to your marketing activity.

We really need to segment out the non-openers and treat them differently (test From fields, Subject Lines, Preview Panes, time of day, day of week, text only messages etc etc) and you will be able to improve the open rate for non-openers.  Just don't mix them up with your best customers or you'll never be able to be relevant and engaging to all of them.

So there you are.  If your company's "bean counters" want you to improve averages, tell them that it may not be what they really want.  Increased clicks, sales, order values by all means.  But not averages, please. 

And how about this for one final silly average...Twitter users have more than the average number of ears.  That’s because nobody has 3 ears (apart from Davy Crockett with his “wild front ear”), but a few people will have only one or no ears.  So dividing the total number of ears by the total number of users we have a figure of 1.99 something, so most people have more than the average number of ears.  

As I said, Barmy, isn't it?

With Christmas nearly upon us I was wondering what presents digital marketers would like to find in their stockings.  A perennial favourite toy across the world is Lego and, whilst we’re probably a bit too old/busy/grown up to get any this year, the joy and wonder of the little bricks is brought to life with these fantastic advertisements from 2006.  

Then I began thinking that, just like some children these days, digital marketers have too many toys to play with.  We should play more often with the fantastic tools we have or we should be using our imagination to make the most of them. 

Display Advertising - Just add imagination

What do you see when you look at a 728 x 90 Leaderboard display ad - just some pixels (a digital Lego Brick?) or a wonderfully flexible and creative marketing format?  

Here are some games you can play long into the New Year with your digital ad inventory:

• Frequency capping – it’s amazing how many campaigns still get deployed in a wasteful, sub-optimal way...work out how many times people need to see an add before it wears out, and make sure your ad budget goes further by serving it to somebody else
• Format testing – so they didn’t click on a leaderboard, so re-assemble the pixels into a sky or an MPU and measure the impact of format on response (interactions or clicks...you choose).
• Contextual targeting – find out from publishers how they can deliver ads based on page content and do some playing (testing).  The click-through rate uplift should pay for the incremental costs and your conversion rates should grow too.  
• Behavioural targeting – make 2010 the year that you unwrap individually targeted ad deployment.  It will mean that you can alter your creative to suit people who are in different stages of the consideration and purchase funnel, based on what they have been viewing and clicking recently.

Email recipients are like snowflakes - every one is different

What do you see when you look at an email address?  Just another name to blast a standard message to?  The wide-eyed marketing child will see some wonderfully exciting opportunities to create imaginative messages that make email more interesting for the customers and more successful for you. 

Here are a few traditional email marketing games to get you started...fun for all the marketing family:

• Acquisition tools – if you’re renting data, ask the list owner what variables they hold and then deliver different versions of the same message.  For BtoB that should be different subject lines, opening paragraphs and calls to action based on “job function” or “industry sector”.  For consumers you may know their lifestyle and affluence from geo-demographic variables that list owners like Acxiom hold...have different propositions for less affluent and older prospects or use a different creative for young professionals.  Let you imagination run free!
• Conversion - as digital marketers we sit on the most valuable real-time prospect data so let’s get it out of the toybox and play with it.  Who clicked on an email link but did not complete a successful outcome?  Who looked at deep product pages on your site but did not buy?  These re-marketing campaigns should deliver 4-16 better conversion rates than one-size-fits-all messages so start building them.
• Retention – send different message programmes to your newer customers, or have a different tone of voice for purchasers of specific products.  Build “personas” (imaginary friends?!) to help with your tone of voice, imagery and calls to action.
• Re-activation – how do you know when you’ve lost a customer?  Probably when they’ve not bought for a specific time period.  So develop a “win back” programme with the first message triggered by a “date of last purchase is more than 60 days".  And be relevant...”we’ve noticed that you have not bought from us for a little while....” is a good start.

Fuel your imagination – get reading!

It’s good to know that even digital marketing kids can find pleasure in moveable type. When I was growing up the “must have” book was the BBC’s “Blue Peter” annual.    This year you ought to be asking Santa to bring you the wonderfully comprehensive “Web Analytics 2.0” by Avinash Kaushik.  Weighing in at more than your festive turkey, with 450 pages that gives you just over a page a day for all of 2010! 

But aside from being a real “value for money” present, this book encourages us to explore our world of data.  It’s partly a “how do they do that” book (go on, admit you’re not really sure how “multi-tabbed time on site is calculated, are you?), but it’s also an activity book along the lines of “what shall we do today to make sense of our marketing”.  With information covering pure web analytics, analytics for search, email and social media and links to further reading it is as near to a “Boys Own Annual” that digital marketers can get.

So there you have it.  Some ideas for kindling your imagination in 2010.  And we’ve not even touched on multi-variate landing page testing, search marketing or social media experimentation...better leave some of those for your birthday!

Here's wishing you a peaceful Christmas and an imaginative New Year. 

Avinash Kaushik has been de-mistyfying the world of web analytics for years through his excellent blog Occam's Razor.  He took us deeper into the murky world of Java-script tagging and standard reports with his comprehensive "Web Analytics:  An Hour a Day" in 2007.  Now he has written a book that all digital marketers should buy, read and leave on their desk to refer back to on a regular basis. 

Web Analytics 2.0 shows us how to move from shovelling  buckets of meaningless "clickstream" data around our organisations to developing a love for true insight.

In short he encourages us to move towards adding qualitative data to our limitless supply of quantitative data in order to really understand what people are doing on our sites.  We must learn to use our hearts as well as our minds.

Lets take a simple example - that old favourite of "Engagement".  Marketers run so many analytics reports to get a fix on engagement that the lights in most offices regularly dim.  And the bad news according to Avinash is that you will NEVER be able to measure how much people are enjoying themselves on your site just with the click-stream data.  For instance, to paraphrase Avinash, 2 people visit your site and spend 10 minutes looking at 12 pages.  Both  happy right?  One loved your site, but the other was frantically trying to find some content and gave up after 10 fruitless minutes - you will never ever ever know this just from your data.  By adding some qualitative data (an on-site survey?) we have more chance of finding out how satisfied, not engaged, these 2 visitors were.

So Avinash takes us on a journey to show us where we should be using our hearts to make sense of data.  On our way we look at the need to move away from once a week reports to continuous streams of  meaningful data;  we are constantly reminded that customers, not marketers, are the best people to inform us what our site should look and feel like; and we are taken on a guided tour of the mountainous areas of competitive insight and told how to mine it profitably.

But this book does so much more than just change the way you think, critical though that is.  It shows you what buttons to press to make your reports more actionable, tells you what sites to look at when considering additional solutions and gives clarity to virtually all the web analytics jargon terms.  Some of the content will be familiar to regular readers of his blog (like the excellent explanation of multiple-tab time on site calculations!) but that makes this even more of a reference book for all our analytics needs.

As you may know I am a huge fan of testing everything that we do in digital marketing and so the chapter titled "Failing Faster:  Unleashing the Power of Testing and Experimentation" took me around all my favourite sites in the digital marketing landscape:  A/B testing, Multi-variate testing and some really sound advice about where to start and a few quick wins to get you in the mood!  Here is my favourite slide that I use to introduce the issue of testing in my courses...I'm sure Avinash would not disagree!

Avinash shows us that web analytics is woven into all our digital marketing activity - from search to site usability and email campaign analysis to off-line integration.  I even spent a rewarding few minutes simply reading the sub-heads and being reminded of things we ought to be doing all the time: 

• Segment or go home
• Five Rules for creating a Data-Driven Boss
• The Key to Glory - Measuring Success 
• Context is Queen
• Failing faster - unleashing the power of Testing and Experimentation

So there we are.  Web Analytics 2.0 is a digital marketing book that takes you from thinking differently to doing better, packed with explanations about the things we ought to know about (or showing us how wrong we have been!).  It comes with a CD brimming with Podcasts, Video and Powerpoint material as well as lists of additional resources.  He even finds time on page 400 to mention Non-line Blogging as a resource people may want to use!  It's taken me 2 weeks to work from the start of the book to the end but it's been a fantastic journey...and at over 450 pages you may want to pack a lunch before you set off! 

We've come a long way with on-line analytics in a short period of time.  A couple of years ago we were all relatively happy with the "last click wins" referrer model and merrily shovelled money into Google's bank account.  Today we are more likely to obsess about attributing a fair percentage of a sale to the efforts of a number of digital activities;  as we always knew, somebody may have seen a banner and clicked on an affiliate link BEFORE using a search engine to find your site, so we'd better juggle our advertising spend accordingly.  But how far should we go down this attribution path, and if you give up too soon what effect will this have on how we judge success? 

Let's walk through the process.  Using a simple analytics funnel we can see what source generates a "successful outcome" - could be a sales lead for a BtoB organisation or a booking for a holiday company.  I've not included any time scales in these examples as the period from trigger to successful outcome will vary from one product to the next.

As I've already mentioned, we've probably come to terms with the over-simple view of the "last click wins" attribution model.  We may even be able to link together all our on-line advertising activity and identify dates of key events like "viewing one of our display ads on-line" or "clicking on an advert on a partner site".  We can then get a feel for the on-line customer journey (as we feel we have influenced it)

Most marketers seem to be at this point and we can get really excited about deciding what is a fair way to atribute the relative importance of all this channel activity.  However, let's not forget the importance of social media at this point - does somebody who goes to a LinkedIn discussion forum or plays with a Facebook widget mean we should give them some credit?  If we're not careful we overlook some of the digital experiences people may have, simply because we may not have tracked them in the past.

One that is relatively easy to track is the presence of email in the overall journey, but as this is often not included as part of the "advertising tools" we may under-represent the role email plays.  It does not usually sit with the "acquisition tool" family and is often overlooked when implementing multi-source tracking like DoubleClick's Floodlight or the newer offering from TagMan.   Matching an email file to a list of "successful outcome" email addresses may yield this multi-channel impact, but this is a manual, somewhat "clunky" production process. 

However, the big problem with our lovely "closed" view of our customers' mind is that it is frequently polluted by mucky, grubby off-line advertising.  Maybe it was a print ad that stimulated the click on a banner, or perhaps a direct mail pack thumping onto somebody's door mat that promped a branded search, as we can see below..

So the conclusion is a bit worrying.  No matter how much we obsess about our digital customer journey and no matter how clever we get with our attribution algorithm, we may have got the "demand generator" completely wrong.  As there is no way to isolate all off-line noise from our customers' minds, and whilst it is still rare (but not impossible!) to track off-to-on-line conversions we are happy to pretend that we have got a really accurate fix on how we get our sales.  Maybe we are not making as many strides in the world of attribution analytics as we think, and are happy to stumble on with our tracking tags.

And if you think I'm being a bit harsh, ask yourself this question:  

If one of your prospects is on the cusp of becoming a customer, will sending them a direct mail pack or making an out-bound phone call help to nudge them towards a branded search on Google?  

If your answer is, at worst, a reluctant "probably", maybe we need to re-think how much value single-channel attribution modeling really delivers and worry less about divvying up our on-line budget between a few suppliers.  Perhaps we should even be trying to invest more in reaching people off-line at the right stage of the digital process? So there.