Hiring Guidelines for Insights Talent

What do I tell HR when looking for people to join the insights team?

Sometimes analysis and insights are hard to quantify from a traditional job requisition perspective.  Here's a quick open letter. 

Hey HR:

Before we talk about experience and skills, let's outline the real requirements.

• We need a very scarce skill.  So, I need a compensation model that doesn't require senior level staff to manage people, products or places.
• We know we have the need, but we can't articulate the job details.  So, I need the flexibility to specify objectives, outcomes and success criteria in the broadest terms.
• We are going to ask big questions that cause all kinds of disruption.  So, I need to provide him with the freedom to work across P&Ls, business, and technology without fear or roadblocks.
• We are creating gold, not mining for it.  So, I need a culture that embraces experimentation; in return we will apply the learnings broadly.
• We need to support this role to provide job satisfaction.  So, I need an organization that asks "What do the data say?"  "What do they imply?" to give her a fighting chance of succeeding.

Now that we have the important bits out of the way, I just need an alchemist.  

cheers
anthony

Planning for the Adobe Digital Summit

What do I expect to be doing next week?

Last year was my first time attending so this year I should have a better sense of what to expect.

• By definition, the sessions will be very product centric.   Consultants will provide better context on the problems and solutions than product marketers.  So, I've skewed the sessions in that direction.
• There will lots of future CMOs and senior marketers in attendance since in the not too distant future the adjective 'digital' will be dropped and this will just be marketing.  I'll be talking to as many as I can introduce myself to.
• The discussions will be range from practical 'how do I implement evar' and the theoretical 'where is marketing headed?'.  I'm looking for something in between 'how can we migrate offline spend to digital and ensure a steady growth in sales?'
• There are always off-the schedule meetings to be had.  This year I want to talk about accessing audiences across touch points, figuring out how content works in the customer experience, and leveraging transaction data re-targeting. 

And of course there are the evenings...more meetings.

Implications of Big Data on Marketing

Can past advertising campaigns help explain what to do?

A colleague recently shared SapientNitro's "Insights 2013"; a collection of trends, discussions and case studies that center on their idea of connected thinking.   At 218 pages (and 40mb) it is a lot to digest at one time - but worth the read.  

The tome is organized around four trends:

1. Real Time Control: New Consumer-Oriented Devices and Data
2. Predicting Desire: Building the Infrastructure to Anticipate Consumer Needs in Real Time
3. Continuous Experiences: How Companies are Blurring the Online and Offline World
4. Globalization: The Global Marketer and the Rise of the Global Consumer 

All of this leaves me thinking about the implications for an organization, particularly around the use of data.   So, using some famous advertising taglines - here's what it means to marketing.

• Think different (Apple 1998) - it is not business as usual.   Applying approaches that worked in the past will not suffice in a consumer-connected world.  We should be asking different questions and to use Cloudera's tagline "Ask bigger questions."  How are decisions made? How do we facilitate choice?
• Don’t leave home without it. (American Express 1975)  This is an interconnected world - and not just among people, but devices as well.  The smartphone has made us first generation cyborgs; Google glasses and contact lenses are next.  The explosion in data exhaust is from sensors and events, not just from humans.  What is the context and intent surrounding the interaction?  Can we figure it out as it occurs (or even before)?  What could we do if we knew?
• Hey, Mikey...he likes it! (Life Cereal 1972)  Testing and experimentation allows for new services/plans to be created.  Google has around 500 experiments going on at anyone time to improve their products.  What things are we testing right now?  Why didn't that idea work?       
• We bring good things to life. (General Electric 1981) The shopping journey is often expressed as an experience.   This requires understanding what people want and how a brand could best satisfy those needs.  This turns the question on its head and inside out.  What would make our journey more interesting and useful?
• Reach out and touch someone. (ATT 1979)  Social sharing (or just plain links) changes the dynamics of the sales process.  It isn't just us and them as direct response teaches us and it certainly isn't a linear funnel - it could be flipped, a doughnut or any other shape you can imagine.   What content are consumers consuming? Where will they go next?  Who will they rely on for information?  How does content work?
• Where’s the beef? (Wendy's 1984) All this data requires new ways of thinking about and organizing analytics. Consider what New York City has done - they reduced ambulance response time by 1 minute by understanding how food/coffee/restrooms impacted response time.  They also reduced injuries to firefighters by 15% by predicting illegal apartment conversions thru the integration of disparate data sets. What can we learn from mash-ups?
• Just do it (Nike 1988) You can't put the right plan in place since there are so many unknowns and changes occur at the speed of thought.  Execute - Track - Adjust will replace Plan - Execute - Track.  What will we do next that we've never done before? 

Time to hack something...

Metrics can be Technically Correct and Precisely Wrong

Why do we have issues with data and metrics?

Data and metrics are unique in their ability to be both technically correct yet precisely wrong at the same time.   It is like they exist in a quantum, parallel universe.

Consider the case of Unique Visitors and Publication Opens.  A recent report showed 9,931 publication opens as well as 13,410 Unique Visitors.   And to help provide insights, the ratio of the two is computed as 'publications/visitor'.  

But 0.74 is less than one publication open per visitor.  

Something, that on the surface seems odd, if not outright wrong.  If publication opens happen by visitors then shouldn't the average be at least one?

The answer depends on how those two numbers are defined.

• Publication Opens: a fairly precise technical event logged by the system based on an individual opening the index page (usually) of a document.  Given that definition it is easy to count.

• Unique Visitors: a business definition is required to answer the question:  What set of events do we want to include in the definition of 'unique'?  If I open two publications;  I am counted once in Unique Visitors and twice in Publication Opens.   If I start a session, but don't open a publication the numbers are 0 Opens, 1 Visitor.   Hmmm….

All is revealed by looking at the way it is implemented, ie at the business logic used in the code and not the raw data.   The definition of "Unique" according to the business requirements, and hence code, covers not only "Publication Opens" but five other events as well.   So, it is quite possible that there are visitors who do NOT open a publication.   And in fact that is exactly the case. 

So, the implication is that the 0.74 is technically correct.    There are 9,931 events logged as an open; and there are 13,410 unique visitors to the site based on six events.  

However, the ratio and thus the business interpretation are clouded because the number doesn't answer the question the client is probably asking:  Among people who open publications, how many do they open?   

Language may be powerful, but it is imprecise.

We (myself included) often rush to say the data are bad…what we are really saying is:  We don't really understand the business requirements we conveyed nor do we have the skills to ensure what we meant actually shows up in the results.  It is experience with clients' business and marketing that allows us to at least apply a sniff test.   Does less than one publication open per visitor make sense?  What is my client going to think?  While I got what I asked for, is it what I need?

The data are what the data are. 

It is the logic conveyed by the business and the language used to implement where most issues arise.

Innovator's DNA

What does it take to disrupt the status quo?

I'm currently reading "The Innovator's DNA" - the most recent in the series on disruptive innovation (co)authored by Clay Christensen.   This one looks at the skills of disruptive leaders and tries to figure out what's different - how did they discover their idea?  According to their research, it comes down to five traits. 

1. Associating - the ability to look at a variety of things and connect the dots in new and novel ways.  
2. Questioning - poking holes in the status quo by asking "why does it have to be that way?"
3. Observing - the art of seeing is grossly under-appreciated
4. Networking - getting ideas and feedback from as a diverse source of inputs as possible
5. Experimenting - since these ideas haven't been done before, learn by doing, discarding and defining

These are not the typical traits we look for in managers, in fact they are the polar opposite since we want experience with delivering results within a defined business model - not a new model altogether.

In the context of marketing, and digital in particular, these traits make sense when trying to deliver a seamless experience across the touch points.  

There are some interesting self-assessments on the website as well.  I lean toward the 'discovery' vs. 'delivery' side of things.

360 Degree Views and Reality

Is there one 360-degree view of consumers?

As a consumer, I'm viewed thru one of three lenses from each marketer that has an interest in a baby boomer with a penchant for fly fishing, being a foodie, and travel (at least for work). Each and every marketer looks at me in one of three ways:

• I'm a customer for the few I buy from, so they know I exist
• I'm in a segment for others who have a vague idea that I exist
• I'm hiding in the dark for those who know I should exist, but can't prove it

The result is that I'm completely different in the eyes of each firm's desire to establish a relationship.  In media planning rooms all across the country I'm on the board in one of those three buckets.

So I'm bombarded with all types of marketing campaigns based on different points of view. The mix includes broadcast, targeted, and personalized 1:1 each attempting to influence a very different "person."

And to reduce cognitive dissonance I tend to filter them out because I'm still just me - one person.

Now it is unreasonable to assume that companies will each take turns marketing to me - they have their objectives to meet.   So, how should you gain access and get thru the bubble?

1. Recognize that reach, targeting and CRM are simply different layers of the same objective - influencing my choices.  I exist in your three media plans, please don't think of them as separate functions.  Package them up.
2. Surround me with interesting content along the journey.  At any point in time I will be 'in different buying stages' for a lot of different things.   Since the brain is hard wired to filter out the irrelevant you need to find a way to fire those neurons. Don't be boring and predictable.  And by all means be consistent across the touch points. 
3. Since we're not talking about the base of Maslow's need hierarchy it is the experience and the resulting emotions that really matter.  It takes more than a coupon.

From my perspective, there is no singular 360-degree view but rather a set of parallel universes.  Your job is to make it easy to decide which one I want to exist in.

Dear Vendor: Listen First, Discuss Second

How would I act if I were on the client side?

Having spent my career on the vendor side of the table talking with marketers and technologists about their needs I recently sat on the other side. And it was quite the eye-opening experience.

We were listening to pitches to develop research around shopper marketing and the path-to-purchase and here's what I noticed.

1. While there was a project brief and opportunity to discuss in detail well before presenting, the opening gambit consisted of canned scripts.  "We're this; we've worked with them; we do the stuff you asked for."  Boring and not relevant; we did our homework - that's why you were invited in the first place. Cut to the chase.
2. Our needs don't fit a standard model - in fact that's precisely why we're not doing the work ourselves - so presenting a canned approach doesn't help your cause much.  Teasing this out of a written brief might be difficult, but a little research on our company - and who would be in the meeting - would have shed some light. Do your homework.
3. Style and personality matters a lot more than I thought.  Just like an ad agency, we know that whatever is pitched now won't ever see the light of day - there's work to be done to get it right.  We're looking at you asking the question:  'would you deliver the results and make our decision look good?' Project passion, not a project.
4. The level of interest in the work ranged from:  "We don't do precisely that, we'll get back to you." to "That would be really different and cool."   I have respect for both answers, be honest with yourselves.  It is OK to not win our business if it isn't your business.  Fold or go all in.
5. While we asked for pricing, it wasn't really meant to be the final number but rather a means of testing the assumptions - both yours and ours.  So, don't get upset that the rules of the game changed (usually by adding stuff we hadn't considered.) Custom doesn't come from a price list.

So, don't present at all.  Come in, sit down and ask questions and listen.

3 Rules for Presenting Analytics

How should we present the results of analyses?

Three rules to keep in your mind when sharing the results of analyses throughout the business and with clients in particular. 

• First, admit it:  We don't have (nor can we get) all the answers.

• Second, speak with conviction and as realistically as possible. Remember, this is not the time for hyperbole, beliefs or a vision of a new-world order. 

• Third, gain a high-level understanding of the 'how' before sharing the 'what'.

When presenting any analysis based on data mining, digital exhaust, or raw transaction data it is helpful to literally go thru a set sequence to avoid disillusionment after the fog wears off and reality sets in.

1. Define the business objective and what you trying to achieve   
2. Outline the methods used to collect, manipulate and process the data.
3. Articulate the limitations, assumptions, and known unknowns.
4. Show the key results of the analysis, backed up by an appendix of details.
5. Discuss the business implications as well as ways to improve the measurement plan itself.

These will help you prevent Garbage In; Gospel Out.

The corollary to this: unless you understand how the results were prepared and the attending limitations don't present them.

Data Mining isn't Like Looking for Gold

What's wrong with comparing data mining to gold mining?

Fire River Gold in Alaska goes through about a half a ton of dirt at the Nixon Fork mine to produce enough gold for just one ring.  The productivity of gold mines is measured in GPT - Grams per Tonne - and the whole process is focused on maximizing productivity as efficiently as possible because gold mining comes with a very high fixed-cost business model.  

Working with data, either digital or transactional, has been related to gold mining since a lot of raw material has to be processed and we're looking for nuggets.  But the similarity stops there - in mining and prospecting for gold, we know precisely what we're looking for and are trying to eliminate areas with low value or probability as quickly as possible.  In data mining the converse is true, we're trying to find the smaller pockets of insights that can be leveraged in our relationship with consumers.

Data mining creates gold, it does not find it. 

(Actually it creates understanding, and that is much more scalable than creating a mountain.)

And the bigger difference is not in the volume or even variety of data, but rather in the process that turns data into marketing recommendations.   In gold mining the excavation, smelting processes removes all the impurities and uncertainty until we are left with a known result -- Au.

Marketing analytics is quite different since we are adding uncertainty along the way to create those useful nuggets of understanding.  Using digital data as an example, we have...

• the logs of actions and we need to classify them into relevant business events, e.g. turn a click into a publication view.  But there remain anomalies like bounces, uncompleted events, etc. that require business rules to clean up.
• a variety of methods that allows us to guess at a visitor; certainly credentials (email) or recognition (cookies) makes this easier but there remains a gap between inference and knowledge.
• uncertainty about the value of what we may find.  The ROI of gold can be computed before hand (grams produced * $pg - less cost of production).  Data mining has no spreadsheet or planning formula.

So, data-driven marketers should think of data mining as an on-going business process rather than as a discrete manufacturing site.