Category Archives: Lab Store

Lab Store: AdSense Finally Works!

DataBase Marketing, Lab Store, Web Analytics

I have a very long history with PPC marketing.  When GoTo finally syndicated their ads to Yahoo, I thought, man, this is brilliant, this is what will make web advertising work.  Advertising at the point of demand (the search), with scale because syndication removes the site-centric orientation.  A direct marketer’s dream, let the testing begin!

AdSense, however, has always been a poor performer for the Lab Store (and many other commerce businesses).  AdSense is really a banner ad with lipstick on it, only marginally attached to the point of demand by context, and not by action.

We’ve tested it many times and always with the same result – unbelievable numbers of impressions and no sales.

In the last test cycle, we finally stuck pay dirt.  Behold, AdSense finally delivering value to the Lab Store:

(click to enlarge)

As you can see, we started out with a Cost / Conversion of about $155.  This ain’t going to work, because the net value of the customer generated from “context ads” is only about $65.  We’re literally losing money every time we make a sale.  But this time around, Google finally provided all the tools we needed to optimize this system, namely:

1.  More advanced control over site selection. 

In the earlier stages of this capability, we could not kill some sites without killing others, so it was not as flexible as we needed.  The most I could find out about this issue was it had something to do with the nature of the AdSense distribution agreements.  As of this summer, something had changed and we had finer levels of control.

The first thing we did was to kill any sites in Social Media land (primarily MySpace), which is where you get the first dogleg down in Cost per Conversion.  Then we continued to selectively prune more sites with the same profile as MySpace - tons of impressions, worthless traffic.

2.  Ability to change pricing models.  

Google allowed us to go from impression pricing (CPM) to click pricing (CPC).  Since we now had pruned the site list down to the most relevant, we were able to jam the price we would pay per click way above the equivalent CPM Google was getting from us before this change so we could own the real estate.

Look what happened – the next dogleg down in Cost / Conversion, finally stabilizing between $4 – $5 per new customer. 

This cost we can live with versus the net customer value of $65.

I know some folks may concern themselves with advertising higher up in the “funnel”.  Gee Jim, what about “awareness”?  How do you reach folks that don’t know about you, folks not searching for your products?

The answer is in the chart above.  I can’t afford to generate awareness, it costs too much.  Generating 400,000 impressions on MySpace basically creates awareness among people who are not interested.  I’m sorry, but I simply don’t see any value in that.

That’s what the mass media are for.

We have tons of high-ranking site content that generally addresses the awareness issue, if a person is casually interested enough in our category to do a simple search.  If they are not interested enough to even search, then why would I want to advertise to them?

I only want to advertise to people who are interested in our category.  Pull, not Push, if you know what I mean.  Relationship Marketing.

Just a beautiful business model I tell ya, this advertising at the point of demand.  As long as you only have to pay for the real demand, that is.

Thanks for making the changes, Google.  Some implications for display ads as a result of this test are discussed here.

How are your AdSense ads doing, have you been able to optimize them to profitability?  Or have you abandoned them completely?

The next post in this series is here.

Interview-Podcast w/ Jim Novo

Analytical Culture, Analytics Education, Brand Management, DataBase Marketing, Lab Store, Marketing / Tech Interface, Marketing Research, Marketing thru Operations, Measuring Engagement, Web Analytics

Friend and fellow blogger Alan Rimm-Kaufman spent some of his valuable time asking my opinion on various online marketing issues in a far-ranging interview and podcast.

We met in person for the first time doing a presentation together at the DMA show in Chicago this fall, and because he used to work at Crutchfield – a truly customer-driven remote retailer – we share some experiences and beliefs.

For those of you who might be wondering where a lot of the Marketing Productivity ideas I post here come from, this interview-podcast is probably a pretty good backgrounder.  We talk about a lot of stuff, including:

Monetizing customer experience

Importance of Control Groups / Source Attribution

Multichannel Marketing Strategy

LifeCycle Contact Strategy versus Calendar-based

Retail Business Models / Lab Store

Search box or not? / Serendipity

How to tell if online customers are really engaged – without web analytics

Here’s another link to the Interview-Podcast.  Enjoy! 

That was lots of fun, thanks Allen!

Lab Store: Panic Pack!

Lab Store, Marketing / Tech Interface

The average ecommerce / catalog packer can pack about 22 – 25 boxes an hour, depending on the level of automation.  Of course, we don’t have anything like traditional mail order automation in the Lab Store; just a computer, a printer and my wife Barb as Chief Packer & Chief Customer Officer.  I pitch in by assembling products and generally keeping the “warehouse” shelves stocked so the Chief can do her thing with the packages and customers.

For a variety of reasons we got a late start today packing the orders from over the weekend and New Year’s Eve / Day, and had to invoke a “Panic Pack” on these 131 orders.  A Panic Pack is a high speed affair where Barb picks all the orders but leaves some for me to pack (I suck at packing compared with her) while she packs the others.  We packed all 131 orders in 4 hours, just barely in time for UPS pickup.

UPS Receipt

That might not sound like a big deal versus the 25 boxes an hour, but remember we are also picking the orders and dealing with all the customer service – can you add this to my order, can you change my shipping method, etc.  Most packers simply pack an order that has already been picked for them, and don’t do any customer service.

131 packages

My point with this post is not that we know how to pack like banshees, but the enabling technology behind this capability.

It would have been impossible to pack and manage the service with this many orders in such a short time without a proper backend order management system – something I see many ecommerce folks go without.  Most web-based cart back-ends are incredibly difficult to deal with, especially on order changes.

In many web-based order processing systems, it can take multiple steps to make simple changes rather than just a few clicks – add another product to an order, run another credit card charge, reprint the packing slips, etc.  This is because once an order is processed, it’s not meant to be changed; order changes were not taken into account when these systems were designed.  Nobody talked to customer service to get specs, I guess…

“You mean customers might want to change an order they already placed?  Why?”  ‘Cause most of them are not geeks.  They make mistakes.  They forget stuff.

Often, when you call companies using these systems to add products to your order, they tell you “we can’t” and to go online and place another order.  Nice.  Great service.

We actually don’t mind if customers want to add to orders they have already placed with us – silly, huh?  Gee, you want to spend more money with us?  Sure, bring it on!  By the way, Flat Rate shipping encourages this behavior.

If you have a good backend system, you can just add the product and the software does the rest, because the order has not been “processed” yet as it has with web-based systems – you process the order right before you print the packing slips, including the credit card capture.  And, you can do all kinds of customization on the packing slip, like messaging for new customers, repeat customers, and so forth, and automatically interface with the shipping manifest system.

The labor cost savings alone when using these order management systems is huge.  When we moved from web-based “copy & paste” order management to local software, our time spent per order on customer service dropped by 50%.  This kind of gain in productivity is common, as you can see here.

And when you have more time to service each customer, you  can provide better, more customized service.  Simple as that.

Plus, our backend system creates one heck of a customer database, automatically consolidating orders at the customer level and providing one-click access to customer service history, cumulative sales, and so forth.  Whenever we are faced with a complex service issue, the first thing we do is look at the cumulative sales of the customer, and then we act accordingly.  In other words, for proven good customers, we bend the rules.  That’s how you build loyalty.

So you need a customer database to provide great service.  As far as Marketing goes, you need a customer database to measure the success of customer-centric programs like this one and this one.

If you don’t have a flexible and marketing friendly order management system, you really should consider getting one.  We use Stone Edge.