Analytics-blog

The Mayor suggested we blog about the much ballyhooed “single version of the truth.”  To be clear, this is not a blog on theology.  So if you were getting nervous…it’s okay.  We are talking about business intelligence and analytics here.

Once again we will channel Wikipedia for a semi-formal definition of the the Single Version of the Truth:

Single Version of the Truth: In computerized business management, svot, or Single Version of the Truth, is a technical concept describing the data warehousing ideal of having either a single centralised database, or at least a distributed synchronised database, which stores all of an organisation’s data in a consistent and non-redundant form.

Swap out the “s” for “z” on some of those words if you are American and hopelessly confused. Apparently the author was taught British English.

So much like many of my college textbooks, the formal definition is abstract and semi-useless.  So let’s talk about some examples that dummies like myself can understand.

One of the first steps that must be taken is to create definitions that are the same across the enterprise.

For example, if we are looking to answer the question “What are the sales for the past week?”; then we must standardize the how we derive the sales numbers.

Various departments, such as finance and operations,  might view the calculation method from different viewpoints. For example, Operations might include shipping and handling in their sales calculation,  Finance might prefer inclusion of the taxes collected in their sales calculation, and Marketing most likely has a different definition of when the calculation should start and end.

More variables on that simple question include:   What is defined as a sale? Is it considered a sale when it leaves the fulfillment center? When the customer orders it? When the customers pays for it? Does it include tax, which tax? Is the tax different from state to state…

So what sounds simple on the surface is not.  Sort of like marriage.  .

Getting those definitions together will not be a small task in a large organization.  Just that process alone can take weeks or even months depending on how “controversial” those definitions may seem.  It may turn out that your sales organization wants the numbers to look as big as possible and your operations team disagrees.  Good luck.

Once you’ve addressed definitions, then you get to ponder the accuracy of the data itself.  Manual data entry, missing records, and allocations based on the time dimension.  This is where you really earn your salary.

After that, then you get to tackle interesting questions like “how many unique customers do you really have”?  Is clint@analytixondemand.com the same as clint.brauer@analytixondemand.com?  If there is yet another “Clint Brauer” at yet another shipping address….is that the same Clint?

All something to ponder.

Then there are all the technical implications of wanting to have “one version of the truth.”  Do you have multiple databases?  If so, how and when do they synch and what are the rules for updates to various fields?  How do you validate and continue to monitor your data for accuracy purposes?

As the Mayor called out to me, this is yet another case where it’s more about the journey and less about the destination.  Chances are you will constantly be making adjustments to your business rules and your technical systems.

As always, the key is to communicate and make sure everyone is drinking the same truth serum.  As long as everyone understands the definitions and the limitations at any given point, then your organization can go about making constructive decisions based on quantitative data instead of pure speculation.

I just started a discussion on the eMarketing Association group on LinkedIn.  Feel free to post there as well any thoughts you have on analytics.

First post is about call centers.  Before you can garner new customers, take care of your existing customers.  Right?

http://tinyurl.com/lex57s

Alright friends,

I’ve been speaking again with my co-worker, “The Mayor,” about list pulls.  He has a lot of experience in this realm and thought it would be worth mentioning on this analytics blog because analytics are essential to understanding what list to “pull.”

The approach we were discussing is based on a prospect showing great interest in making a purchase via their behavior.  You just simply remind them of their one-time intentions in a very “soft” manner.

Here are a few examples:

• Abandoned Shopping Cart – The potential customer added a product to their cart, but never completed checkout.
• Inquiring Prospect  - The potential customer called your team and gave them all the required info, but for some reason backed out of the transaction.

Obviously, you should be storing this data inside your analytics or business intelligence system.  And, obviously, your call center analytics or business intelligence system should be hooked up to your email and outbound call systems.

Assuming you are technically solvent…if you sent the customer either an email or called them with a special offer regarding the specific product they “almost bought,” there’s a good chance they will not be mortally offended and threaten, or abuse your team members with legal action.   And, there’s a decent chance they will finish the purchase.

Now we can take it a step further.

• Unidentified Interest – In this case someone contacts your call center team, shows a lot of interest in a product, but doesn’t give out much info.  Much like the author of this post, they are paranoid.

In this case hopefully you can understand why they are refusing to buy the item.  Then you can segment these folks out and decide which ones are worth calling back.  Now obviously…you don’t have much information beyond their phone number.

However, you can take that phone number and contract a firm to provide you with the name and address of the owner.  Then you can send some follow-up snail mail to them with an offer about that same product.

It’s a little sneaky for sure, but The Mayor tells me it works quite well.

After having this discussion, I wondered if some of my “deja vu” experiences can be attributed to the Mayor.  ”I know I’ve seen this product before…know I wanted it…did I dream this?….I think that I did!”

And if you are barely clinging to reality- like myself- you will take it as a signal from God that you are meant to purchase that item and will happily shell out the cash.  It was destiny.

And if you can get a customer to fulfill their destiny 10% of the time it was probably worth the effort.

A real Tree of Discouragement. So sad.

Tell me if you’ve had this experience.  (actually, I already know you have)

You’ve given business for years without complaint to some firm or company.  You were not a coupon/discount shopper.  You rarely, if ever, called their support team.  You just forgave them for small annoyances.  You provided referrals through your social interactions.  You paid on time.

Then one day something goes awry.  So you pick up the phone and call support.

After you work your way through the Phone Tree of Discouragement, they put you on hold.

For a long time.

Then they re-route you because you couldn’t quite pick the right option from the Phone Tree of Discouragement.  You dummy.

Back on hold.

Re-routed.

Insulted.

Flipping out.

Frankly, this describes about 80% of my experiences.  It’s why I’ve made a rule to transact as little as possible.  It’s just not worth it.

So why does this happen to a valuable customer like yourself?  It’s certainly not a wise business decision to anger your best customers.

I’ll tell you exactly why.

Companies don’t know the value of their best customers.

And even if they do, they have no clue if a valuable one just bought something from them, if they are on the phone or just sent an email.

Honestly, there is no excuse for this gibberish.

What needs to happen is as follows in no particular order:

1.)  You dump your call center data into a database

2.)  You dump your ecommerce and retail data into that same database

3.)  You dump your loyalty data into that same database

4.)  You make sure that you can connect the data points at the customer level

5.)  You do some really simple segmentation using some nice business intelligence and analytics tools (or you can do complex segments..your choice)

6.)  You pipe those segments into your call center, commerce and retail systems

7.)  You program those systems to react accordingly based on the segments

Voila!  When your high value customer calls in…they go to the top of the queue.

Congrats!  You have a 360 degree view of your customer!   You are in the top 10% or all companies.  (just an estimate)

Honestly, this isn’t that tough.  I did this over five years ago with some bailing wire, conviction and multiple low-cost SaaS analytics, CRM and email providers and a very small team of infighters.

So put on your big boy pants and look out for your top customers…before they look for someone else.

One of the guy’s here at AOD said “Churn Rate” was one of his favorite metrics.  You know that you are serious about analytics when you have “favorite” metrics.  But we would expect nothing less from the guy we call “The Mayor.”

Churn Rate Metric Definition (Mayor’s version):  Measures customer attrition.  Refers to % of customers a business loses over a specific time period.

Churn Rate Metric Definition (Wikipedia version): Customer attrition, also known as customer churn, customer turnover, or customer defection, is a business term used to describe loss of clients or customers.

I like the Mayor’s definition better as it offers a little more in the way of detail.  Also, he might give me lunch for saying so.

So, here’s the magic formula.  (Our version)

Churn = Ct / Cat

Ct = number of customers a business loses over time period

Cat = number of active customers at the start of time period

Below is a sample chart:

So what conclusion can we draw from this chart above?  Frankly, nothing concrete if you view this chart only.

However, if I were in charge of the company above, my research would begin with the customers that have stayed over three and four years.  I would be curious to learn why they have stayed with the company, and try to find more customers like those loyal folks first and foremost.

In addition, I would take a close look at the chart above cross-referenced by my media channels and campaigns.  It might turn out that you have specific marketing campaigns that are contributing to high turnover.  Common suspects include any promotions where you discount your offering.

Then I would cross-reference the data above against the customer support information.  However, as you are probably well aware, most folks that “churn” don’t take the time to tell you why they are leaving or even that they are leaving.

One last cautionary tale about churn metrics.  The chart above is pretty simple by design, but I’ve watched churn rate discussions turn into spontaneous combustion at the executive management level.

It normally goes like this:  Your meeting is going along fine because of your painstaking preparation and then someone…with their chart upside down…says “I don’t get this.”  Then your head explodes.

So be very cautious about how you define your time periods and make sure everyone understands up front EXACTLY how this is calculated.  Some of your co-workers may have a difficult time understanding charts like those above.  For those people, tell them “The Mayor” is on your side.

Now that the total ownership costs of business intelligence systems are being driven down by SaaS, we’ll likely see the business intelligence market grow like a California cash crop over the next years; regardless of economic conditions.

In other words, the chart below might be about to “elbow” or “hockeystick”.

So braid my hair Mary Jane, it’s going to be a fun ride.

Source: The OLAP Report 2009

Hey you.  Yes You.  You with the small budgets but the big ambitions.

Now You can sit at the table with the adults.  You can afford a luxurious business intelligence system.

No doubt you’ve been enviously watching your large company executive peers dine with Cognos, Hyperion and MicroStrategy purveyors for years.  You’ve been watching them out-organize and out-analyze your smaller operation for a long time.

Of course, they probably didn’t mention how many people it took to manage the system, how difficult the system was to use for non-analysts, the ongoing expenses…

The good news is that you won’t have to worry about any of that.  You can eat like a king without needing to melt down the golden goose.

SaaS analytics providers, like AnalytixOnDemand, have emerged on the scene and they are driving down costs, compressing implementation timelines and extending reporting capabilties to non-analysts.

As a result, for the first time the “hard part” of business intelligence is not about systems, it’s about having the discipline to use the system properly to improve your business.  More on that in a separate post(s) in the future..

I’ve been walking the halls at AOD speaking with folks about “what it is we do.” At the end of the day, what we do is give folks tools to optimize their business decisions.  But that phrase “optimize business decisions” is still meaningless to the average bear (or CEO).

What I’m hoping to do is get into some details, and expose folks to some of the metrics we present to help organizations optimize their business.  Please let me know if this is helpful or not.

So here goes our first metric, Cost Per Lead (CPL) by Media Channel:

The metric is calculated by dividing the Total Advertising Cost (TAC)  for a Media Channel by the Total Leads Generated (TLG).

So CPL = TAC / TLG

This chart helps you understand the spend required to acquire a somewhat qualified potential customer.  For those of you with a CRM or sales background, you know that a lead is nothing more than someone who has responded to your messaging, or someone that you SUSPECT might be interested in your services.

Here are the decisions you could potentially make using this report:

• Decide to further research your efficiency by looking at a Cost of Conversion by Media Channel report.
• Decide to allocate additional funds toward lower cost lead sources with the goal of understanding your spend limit before diminishing returns occur.
• Decide to further drill down into each Media Channel to understand why some costs are higher than others.  You might find that a few specific media vehicles are costing more than they should to generate leads.
• Decide to further drill down and research the differences in the creative across those Media Channels.  You might find that that the creative is driving your efficiency more than your media channel.

Obviously, what you don’t want to do is make a knee-jerk decision based on this one marketing metric alone.

Given that fact, just like virtually any other measurement criteria on the planet, it is only useful if compared against other metrics.  In this case your would want to understand the conversion rate by Media Channel as well.  It might turn out that while you are paying more for leads in certain channels, those leads are better qualified and thus convert at a higher rate.

Hope this made sense.  I look forward to any and all comments.

So it’s not quite like Spy vs. Spy.  In fact, to some extent it’s a complimentary relationship.

For example, I’m a big fan of Omniture and their SaaS online marketing analytics system.  While at Sony I did my best to push adoption across multiple marketing and website groups…and in the end they were adopted across most (if not all) Sony divisions.  They have a great user interface and their support system was excellent.  They are the best in online marketing & analytics tools in my opinion.

At AnalytixOnDemand we also talk about marketing analytics.  However, when we talk about it, online marketing analytics are just a subset of a much larger picture.  We also support statistics such as sales per square foot and return on equity that have very little to do with online marketing.

So the real difference between an Omniture and an AnalytixOnDemand is pedigree.

Omniture has a pedigree in online marketing analytics.

AnalytixOnDemand has a pedigree in business intelligence.

Business Intelligence encompasses online marketing analytics + call center + finance + logistics etc.

This means AOD must be conversant in hundreds of reports and KPI’s across industries.  This is not easily attained.  It requires years of experience.

In addition, companies with a business intelligence pedigree like AnalytixOnDemand expect data to appear from MANY sources.  Not just a browser.  As such, our systems and processes are built around the concept that we will be dealing with messy data.  We’ve built tools and processes specifically to identify and report these issues.  And it’s just not for online data…it’s ANY data.

If you read this blog you probably are thinking…well why in the heck would I even consider Omniture?  Well…they have a great system as pertains to anything you are doing online.  And also, Omniture combines those analytics with various optimization tools that help folks real-time organize and organize their website, campaigns etc.

Our application helps you look at performance across all channels or even across your entire enterprise….and also drill down to figure out why a given trend or pattern exists.

There is certainly some overlap between the two systems.

If you are a Director of Online Marketing, I would not advise you to drop Omniture in favor of AOD.  However, if you are VP of Marketing with a multi-channel marketing effort and managing the call center, I would advise you to hire AOD in addition to retaining Omniture.

Hope that makes sense.

Nor is he a young Harry Potter.

You should know better…but maybe you don’t. Depending on the very fact that you call someone your “reporting guy” might signify you are wasting your resources.

It’s one thing to have a reporting analyst or several analysts to help you dive deeper into the numbers.

But if your analyst is spending a lot of time gathering data, putting it in a system and then massaging it for “executive consumption”; you are lost.  There’s also a good chance that the data you are receiving contains errors.

If there are errors, it doesn’t mean your reporting guy is incompetent or that the unicorns are acting unruly.  It means that he’s human and it’s too much to manually create.

There’s nothing worse than being the reporting guy, and realizing in mid-meeting that you made a mistake.  I’ve been there…and it ranks right up there with being kicked by a unicorn in the cajones’ in terms of sinking feelings.

The truth is you might not even know the “how” of what occurs in the Reportland.

You should find out.

Your young wizard might be nervous at first from all the attention, but once he realizes that this is a chance for him to unload the repetitive portion of his job in return for more time finding “truth” among the analytics, you’ll likely be rewarded in terms of increased enthusiasm and even better analysis than previously experienced.