Channel Intelligence Blog

Broad Outsourcing

In 1994, I was introduced to a new innovation called the “internet”.  I was in college and I distinctly remember watching a demo of some early version of Mozilla and thinking, “This will change the world.”  And, it has.  This technology has fundamentally changed the way we play, work, and communicate.  More recently, I’ve watched blogs, social networks, and web sites like Wikipedia start to do the same thing. 

Today, another world changing innovation is taking place: Broad Outsourcing.  Specifically, Amazon’s Mechanical Turk.  In short, put tasks out to 100’s of thousands of workers around the world and have it done faster and cheaper than you could ever imagine…with quality.  From the workers’ standpoint, they can work where they want, when they want, and how they want.

 

In one of our early tests, we sent out 73,000 tasks and had each task worked 3 times.  We defined success as having 2 out of the 3 answers agreeing with ours.  If we did this internally, we figured it would take us about 5 months and a cost of about $35,000.  Using Turk’s broad outsourcing, we were done in 4 days with an 87.3% success rate.  Better yet, we did it very economically.  Incredible results.

Now, the internet didn’t solve world hunger and I don’t expect this will either.  However, there is a fundamental shift that is going to take place.  The same way the internet made information globally accessible, broad outsourcing will make the global workforce accessible.  This will change the world.

Good Idea, Gus

One of the most common objections we get from manufacturers who are not ready to link to their retailer partners is the proverbial 'we are about to change our site.' Last week I met with Gus Brito of Tiger Direct and Gus had a good idea. 'Instead of trying to do product-specific links right away,' said Gus, 'manufacturers should do retailer-specific links.' Good idea Gus.

We call those Dealer Links and here is a good example. I know I talked about retailer-specific links a couple of weeks ago and I said I wasn't sure anyone shopped that way. But if a manufacturer is delaying an implementation then retailer-specific links are better than nothing. At least with these kinds of links the manufacturer gets reported sales and click through data.

Now, I'll always be a proponent of manufacturers getting product-specific links in place as soon as possible. A site re-design should not be a reason to delay product-specific links. In fact, it should be a reason to get them in place. If you get the links up before the re-design you can do some A/B testing and measure the effectiveness of your new design. Installing product-specific links is also a simple change. I know, everyone who wants you to update your site tells you it will be a simple change.  'Hey Bob, you should install all of those SAP modules this week. It's a simple change!' Eighteen months later...

However, a couple of lines of javascript placed in a template is pretty easy. So don't delay your implementation. But if you must, then consider linking from the single page where you have all of your retailers listed.

From Search to Sale: yesterday, today, and tomorrow (2 of 3)

.today (part 2)
From the brand perspective: 100,000 customers a day, about 4,000 an hour, what do you do?

The proliferation of the Web today has created very strange dynamics: consumers show up at manufacturer Web sites to find out more or initiate commerce. When consumers go to a retailer's site to make a purchase, the retailer's web-based product page is the modern functional equivalent to a "package."

So the retailer has become the packager and the manufacturer is forced to merchandise and become a larger part of the purchase process from search to sale.

This shift, created purely by online consumer behavior, has resulted in significant challenges. Retailers creating the primary form of 'packaging' at the point of sale is critically dangerous. Retailers are primarily motivated by closing sales, period. In the long term Brands invest in positioning, differentiation, messaging, and the definition and representation of their brand or brands.

With regards to Manufacturer Web site traffic, a Brand is no more innately capable of merchandising goods to consumers and closing sales (those activities that were granted to retailers to create expertise) than retailers at creating product differentiation and brand essence. Brands need to learn new tricks, hire new expertise, and learn to think like retailers. And retailers need to be conscious of clearly communicating brand-created differentiation and essence. Without these respective foci, differentiation will be lost and commoditization and erosion of meaningful consumer choice will ensue.

But could there be yet a stranger even more blurred world? YES!, the world of tomorrow...

Lies, Damned Lies and Statistics

An article in the Baltimore Sun released earlier this week stated that consumers look, but don't buy, online. I also heard this same story on my local radio station during my drive into work. As someone who has been closely tied to the online world for the last 10 years I know this isn't true.

According to the Pew Internet and American Life Project, 40 percent to 55 percent of shoppers surveyed said they looked online for information but they more often relied on brick-and-mortar stores - or human agents - when it came time to buy.

That may have been true 20 years ago when people would go into a retail store to talk to a well-informed salesperson, but the Internet has changed that. What was this hyperbole-touting reporter thinking?

As you read deeper into the article you find the categories the Pew survey was studying - cell phones, music and housing. Ah HA! Now it all makes sense. Cell Phones? People don't buy cell phones, they buy cell phone plans. They get the phone that comes with the plan. And they don't buy plans online. There are tens of thousands of verizon, AT&T, Sprint and T-Mobile stores and kiosks around the country to pick up these plans. Music? Do people go to Sony BMG or Warner Music Group to figure out the next CD you want to buy? No, you listen to the radio and you talk to people. Music is a commodity product that can be easily picked up in brick-and-mortar stores and usually isn't worth the online shipping costs you may incur to buy it online. Housing. Housing? Are you telling me people don't buy housing online? I'm shocked. They research online and then shop with a realtor. It makes sense to me.

So things that are not bought online are not bought online - a self-fulfilling prophecy. It's stories like this that give eCommerce a bad rap. Spouting statistics about services and commodity products and giving people the impression that these are indicators of how well eCommerce is doing overall, is misleading. From every indication I've seen in our business and every survey we've done in the past five years, eCommerce is alive and kicking and as strong as ever.

‘Fire and forget’…it ain’t

It’s funny how the old adage, ‘Nothing is constant, but change itself’, applies to everything.  Take building and sending data feeds as an example.   We use the concept of templates to build data feeds for our partners.  Every time we build a feed, we follow the most current template and tweak it from there.  This helps to ensure we’re following the partner's specifications and that we can create the feeds as efficiently and quickly as possible.  However, just because we’ve followed the specs today doesn’t mean the feeds are right tomorrow.

Recently Pronto added “Retail Price” and “Special Offers” to their specifications and Google added support for about a hundred new attributes.  Now, these new attributes aren’t always required, but the statement “recommened for effectiveness” is a pretty powerful one.  Even though we’ve had hundreds of feeds already built for these destinations through the templates, we still needed to go back and modify them in order to take advantage of the new data points.  While keeping up with all the changes from all the destinations can get a little monotonous, it’s pretty exciting to see traffic and sales increase because of it.

So, the next time you fire off one of your data feeds, be sure you don't just forget about it.  One little change could mean thousands of dollars in new revenue.  It’s certainly worth the time to make sure you’re leveraging every change and making your feed as effective as possible.

Feed the hot shooter

As the NBA playoff roles through the final round, I am reminded of the old basketball adage, “always feed the hot shooter.”  The idea is that when a player is shooting well, you will score more points by going to the hot hand than by arbitrarily passing scoring opportunities around the entire team.

This adage is also valuable advice for publishers as they consider a product linking service.  One thing a publisher should internalize is that not all retailers shoot the same percentage.  Measuring retailers by the sales dollars they generate per lead, the simple truth is that some retailers are many multiples more productive that even their closest competitors.   There are a number of issues that likely drive this difference; including price, availability, selection, and service but that is a subject for another post.  In short, the best retailers have created a formula that better meets the customer’s needs and customers respond with their wallet.

Regardless of the reason, this state of affairs has a number for implications for publishers:

·         For publishers with a vested interest in generating product sales, where you send your traffic impacts your top line.   Any product linking decision which does not use sales per lead as a key component of determining the order in which to render retailers, costs you sales.   If you purposely or accidently bury your hot merchants for too long, it could cost you the ballgame.   Organizations may choose to accept this cost but they should do so with eyes wide open as to the very real and immediate impact of such decisions. 

·         For publishers who focus on their end-user’s experience, the implication is similar.   Like fans attending a game, customers show support with their purchases.  A high sales value per lead over time reflects the retailer’s winning formula.  Rendering winning retailers higher on the page helps ensure that a publisher’s end-users will have a positive buying experience.

Feed the hot retailer or be ready to end the game with fewer sales and less satisfied users.

From Search to Sale: yesterday, today, and tomorrow (1 of 3)

The rapid transparency and availability of information has clearly changed the game for brand managers, retail merchants, and consumers whether they recognize it or not!

.yesterday (Part 1)
The old paradigm (before the web) was from a simpler time. Brands bought raw material, manufactured product, and packaged it and shipped it to the retailer. They were concerned with the continual value in the eyes of consumers with a strong going concern and long term view. Advertising concentrated on communicating brand “essence.” Retailers became experts at the tactical consumer experience: driving traffic and converting sales. Retailers owned one key event: exposure to products or the initiation point of commerce. This first door the consumer goes through is a critical step and successful retailers were those better at merchandising/presenting and explaining new products and needs.

Until: The Big Bang! The rapid proliferation of the Web has resulted in a remarkable role reversal!

If you were a top brand, imagine 100,000 of your customers showing up at your corporate headquarters doorstep every single day. What would you do? Do you have any inventory, someone to speak with each of these customers, or even a cash register to make the sale?

The Delivery Person...

Imagine a hypothetical delivery man. He has a parcel he has to deliver and a document that specifies how it is to be delivered. The document describes where to delivery it, who to deliver it to, whether it has to be signed for or not and so on.

Our hypothetical deliveryman arrives at the address specified on the document. He sees that the name on the office door corresponds to the person who is supposed to be receiving the parcel. The person isn’t there. What does the deliver man do? He looks at the document and sees that it says the parcel has to be signed for but doesn’t say who has to sign it, so he goes to the office next door and says, “Do you mind signing for this? It’s for so-and-so next door. Great thanks. I’ll just leave it on his chair.”

The point of the story is that the deliveryman did not literally do what the document, said which was to deliver the parcel to so-and-so at some address. Instead, he left it for so-and-so and got someone else to sign for it. This is a perfectly normal human process. Everything we do by way of following an instruction of some sort involves interpretation. We mediate between the abstractions represented by the instructions and what actually happens in the world around us.

What is on the deliveryman’s document is data, the mediation process is information. If you prefer, information is interpreted data, or putting it another way, information is data that has consequences.

Let’s look at this in the context of people buying and selling goods. In a traditional store, the customer can see and touch the goods being bought. There is eye contact, speech, and tangible, physical presence between the customer and the merchant. This presence provides a context for the experience that results in an immediate understanding of what is being offered and what is being asked. It is the basis for a rich information exchange involving the customer and the merchant, and the customer and the general environment in which the interaction occurs. There are many dimensions to this but, for example, recent studies have indicated there are regions in the brain that react to the presence of others by an internal mimicking process that involves a literal replaying of the other’s actions involving the same neural processes as though the observer were doing the action themselves. This profound influence between physically present participants in an exchange is very difficult to replace in situations involving distance learning, remote working and contractual interactions carried out where the participants are not physically present.

Three Reasons Consumers Visit a Manufacturer's Site

Just like Family Feud. Name three reasons consumers come to a manufacturer's Web site...

...Product Information! Ding. Correct. That is the number one reason consumers come to a manufacturer's site. 60-70% of the time they are looking to learn more about the product.

...Find a place to buy! Ding. Correct. That is the number two reason consumers come to a manufacturer's site. 25-30% of the time they already know what they want, they just want to know where they can get it.

...Support! Ding. We were looking for Support and Service, but we'll accept that answer. Around 5% of the time consumers have already bought a product from you and they are visiting your site post sale. Percentages vary of course according to product type and category.

In each instance you have an opportunity to convert that consumer to a sale. Tell them where the product is in stock (online and locally). Even if they are looking at product information you don't know when they will flip from being a browser to a buyer. Support and service consumers are great candidates for your accessory products. Don't let them leave empty handed.

From our last survey of browsers on manufacturer Web sites we found a staggering 40% of visitors to the manufacturer's Web site actually bought either online, locally or direct from the manufacturer within 7 days of their initial visit.

If You've Got Two Watches....

Ever heard the phrase "If you've got two watches, you never know what time it is."?  I hadn't until last year at Shop.org (Vegas baby! VEGAS!).  A guy who wandered into our booth used it to cleverly describe a problem I was blabbing about.  I was explaining how I am often pulled into conversations with clients who are questioning the validity of the data we provide in our reporting.

"My <insert reporting package name here> numbers are different than your numbers!" is typically how the conversation starts.  The rest of the conversation typically takes one of two paths….

Path #1: Apples and Oranges

Let's take clicks for example.  Our original reporting goal was to do everything we could to make sure "Clicks" was representative of what the client would see if they logged into the respective marketing program's portal (Search, Comparison Shopping Engine, Email, etc.).  Said another way, "Clicks" is defined on our side as the number of clicks the respective program is going to charge you for, assuming a CPC model.

I can assure you that "Clicks" within <insert reporting package name here> is not defined the same way.  The definition is typically the total amount of inbound clicks that came from a specific lead source, typically tracked by a URL query argument.  This number will include test clicks, fraudulent clicks, and other clicks the marketing program most likely will not ultimately charge for.

So, when I hear "Your click counts are too low in your reporting...", I smell Apples and Oranges.  This situation is not limited to clicks.  There are all kinds of metrics that have this issue when "comparing" two or more reporting systems.

Moral of the story? 

1. Users shouldn’t assume "Clicks" = "Clicks".
2. Programs should do what they can to provide metrics matching all possible definitions.

On our side, we started to provide two different types of “click” values.

Raw Clicks: The number of leads that Channel Intelligence logged prior to any exclusions, such as click fraud rules or traffic from certain IP addresses. This number is designed to reflect the same number you would expect to see within any web site traffic reporting system you may employ.

Clicks: Clicks is intended to represent, as close as possible, the number of clicks you would expect to see within the respective marketing program's reporting portal. This number is achieved by filtering out raw clicks from sources known to also be filtered out of the respective marketing program's own reporting/billing, using actual clicks/cost data acquired from the marketing program directly, or a combination of both. Data acquired from the respective program will always supersede estimations made by Channel Intelligence. This level of combination is referred to as Blended Data within the reporting system.

This has helped tremendously.  Clients can now line up their package’s version of "Clicks" to our "Raw Clicks" since they are designed to be the same thing.  With those numbers being close to dead on (which in the reporting world means "equal"), this establishes the confidence necessary to not only get behind, but get excited about our version of "Clicks", which their package typically doesn’t have.

Path #2: Blinders are for Horses

Another common issue is that the sales data we associate with specific marketing programs is higher than <insert reporting package name here>.  "My system says 129 orders and yours says 152" is a common concern I used to hear a lot.

Welcome to the "Blinders" problem.

What is occurring in this situation is that their internal reporting package is monitoring not only the CI managed programs, but every other marketing program driving traffic and sales.  If CI is only watching traffic from the CI managed programs, the lack of peripheral vision results in the association of an order with a CI managed program when in actuality, a separate non-CI program led the purchaser back to the site in a subsequent click resulting in the sale. 

What I've just described is a common problem with every marketing program out there that is only able (or willing) to watch their own program(s) for reporting purposes.

Moral of the Story?

The only way to accurately report on any ONE thing is to watch EVERYTHING.

So, a while back we started monitoring all marketing programs for our clients even though we only report numbers on the specific programs we manage.  Cool thing is this has evolved into much more than we originally thought.  This level of data collection allows us to not only provide accurate metrics for the programs we provide, but show the relationship these programs have with other marketing programs, CI managed or otherwise.  Which programs are often working together to close sales?  Which programs are influencing sales, but not getting credit because of a last in wins setup.  And on and on.  Pretty cool stuff, especially when put into the context of specific products, categories, keywords, etc.

So, in conclusion, you may be asking  yourself “Why have multiple watches at all Dave?  One is all I need right?”. 

Great question.  The way I have always answered this is that no one reporting package provides everything to everyone.  Some are wonderfully flexible yet barely usable.  Some are easy to use but limited in scope.  I could go on.  My point is that with more and more service providers offering analytics that are typically very specific to the service provided, it won’t be uncommon for a clients to have many “watches” to look at.

My advice is to make sure the minute hands are not being compared to hour or second hands, each watch tracks all 24 hours of the day, and every now and then synch them up. ;)