Why Some Analytics Initiatives Are Failing
"Without big data analytics, companies are blind and deaf, wandering out onto the Web like deer on a freeway." — Geoffrey Moore
Geoffrey Moore, author of Crossing the Chasm and Inside the Tornado, certainly got this quote right. And, he's not alone. The age of analytics is here and there's no turning back.
The ability to combine data and analytics to improve business results is the single most important competitive advantage in the world. Accordingly, companies everywhere are rushing to deploy armies of analysts who are cranking out predictive models that promise to catapult revenue and earnings to stratospheric levels. By combining behavioral data and appended data with machine learning and logistic regression, these competitors will engage the right customers with the right messages at the right times and over the right digital channels. It's a data-driven marketing gold rush.
"Somewhere along the way to this analytical nirvana an inconvenient truth emerged."
But wait – somewhere along the way to this analytical nirvana an inconvenient truth emerged. More and more companies are discovering this critical flaw that can derail hundreds of thousands of analytical initiatives every year. This one problem can lead to more marketing failures, more wasted dollars, and more career missteps than anything else in the analytical arena.
What is it?
It's your dirty data.
If you think you're immune to this dangerous data dilemma, you are almost certainly wrong. Years and years of data audits have empirically proven it.
Literally every single data audit we have performed over the last 25 years has revealed large and serious data problems. In addition to flawed data, every customer database suffers from missing or incomplete data.
"Every single data audit we have performed over the last 25 years has revealed large and serious data problems."
And, when millions of customers are involved, these data problems impact hundreds of thousands of customer records. Even worse, when defective and incomplete data is used for analytics and direct marketing, significant financial problems and risks always emerge. These include:
- Wasted marketing budgets
- Missed revenue and profit targets
- Frustrated customers and prospects
- Lawsuits and costly fines
- Lost reputation and brand equity
When dirty data is the foundation of your analytics and marketing, your problems never end.
"Garbage In, Garbage Out"
Computer scientists and coders across the globe have long held fast to this truism: "Garbage In, Garbage Out."
Often referred to as GIGO, this simple yet powerful concept means this: the quality of your output is determined by the quality of your input.
In the case of marketing analytics, the quality of your predictive models is determined by the quality of the data you use to develop and deploy those models.
"The quality of your output is determined by the quality of your input."
More importantly, the quality of your decisions and marketing results are determined by the quality of your predictive models.
Once again in English: If your analytical gurus are using dirty data to develop their models, you are going to get inferior business results at best, and disastrous business results at worst.
And, here's the most insidious aspect of your problem: it's easily overlooked and undetected. As a result, many companies continue to develop and deploy sub-optimal marketing analytics without any knowledge or awareness of their failings. Their marketing teams blindly execute sophisticated, data-driven, digital campaigns that are built from bad inputs.
These marketing campaigns are dead on arrival.
Predictably, these companies get very poor results.
Data You Can Trust
So what's the solution?
Let's discuss five concrete steps you can take to improve the quality of your data. These steps will give you cleaner data, stronger analytics, and better marketing results.
1 - Correct and Standardize Customer Addresses
Even in our digital world, your customer's physical address matters. It matters a lot.
The physical address is the touchstone for so many critical data elements and updates. Unfortunately, 15-20% of addresses can be incorrect on many customer databases. Make sure you are using the latest data hygiene algorithms to clean and update your customer addresses. Your data cleaning should deploy up to 16 different processes, including CASS, DPV, NCOA, occupancy based corrections, PCOA, deceased suppressions, zip and city name corrections, CMRA detection, de-duplication, and more.
"15-20% of addresses can be incorrect on many customer databases."
Once cleaned, your correct addresses will immediate improve the following downstream efforts:
- Data appending and enhancement match rates
- Ethnicity and language preference coding
- Email appending match rates
- Social media ID appending match rates
- Phone matching and correction (mobile, VOIP and landline)
- Calculations around proximity to locations (yours, competitors)
2 - Precisely Identify Unique Individuals and Households
It is critical that you accurately identify unique individuals and households from any source of data that contains a name and an address. Advanced analytics teams refer to this as "pinning." The algorithms that are used in your pinning process should go well beyond the use of simple match codes.
Your households should be assigned a unique household identifier (HHID) and each individual should be assigned a unique individual identifier (INDID). These identifiers are invaluable for linking all sources of data together and providing a consistent single view of your customer across all channels (online and offline).
"Your pinning system becomes the foundation of your entire data structure."
Your pinning system becomes the foundation of your entire data structure. This system maintains the integrity and usefulness of your data going forward, including all downstream data enhancements, analytics, scoring, omnichannel execution, and tracking and attribution.
Do not skip this step.
3 - Enhance Your Data with Consumer Profiles, Behaviors & Propensities
You are now in position to accurately enhance your data with a wealth of consumer information. Data appending will correct data errors and fill in missing data. This data significantly improves your predictive models and and marketing results.
"This data significantly improves your predictive models and marketing results."
Here are some of the most common categories of consumer data that should be evaluated and appended to your customer and prospect databases:
- Financial data
- New movers
- Purchase propensities
- Purchase triggers
- Lifestage clusters or segments
- Shopping behaviors
- Brand preferences
- Retail preferences
- Channel preferences
- Ethnicity and language preference
4 - Append Omnichannel Contact Information and Preferences
If you don't know how to reach your customers and prospects, even your best analytics are worthless. Make sure you accurately and frequently append (and correct) the following omnichannel data:
- Email address
- Social media ID: Facebook, Twitter, Pinterest, Google +, Youtube, LinkedIN, Instagram, etc.
- Phone numbers: mobile, VOIP and landline
- IP address
- Channel preferences
"You are now able to leverage your analytics within a true omnichannel marketing strategy."
With this critical information on your database, you are now able to leverage your analytics within a true omnichannel marketing strategy, including:
- Email marketing
- Display advertising
- Social media marketing
- Customized website offers / experiences
- Mobile marketing (location-sensitive and beacon-triggered)
- Outbound and inbound telemarketing
- Direct mail and catalog marketing
5 - Standardize Your Data Inputs, Processes & Outputs
Now that you have expended time, energy and budget to clean and enhance your data, you must take care to consistently use your data. This means standardizing how you receive, transform, store, analyze, report, score and act-on your data.
While a full exploration of this step is beyond the scope of this blog post, here are some key ideas to get you on the right path:
- Naming: when consolidating and integrating your data feeds and silos, make sure you standardize data names and dictionaries
- Transforming: when transforming data, make sure to fully document each transformation step, along with an accurate description of the final data element to be used by analysts
- Modeling: use standard code for developing and evaluating model fit and effectiveness
- Scoring: use standard methods for scoring customers, ranking them, bucketing them into deciles, and otherwise deploying your analytical decision tools
- Executing: use standard processes to QC your analytical outputs and leverage them in your omnichannel communications platforms
"Take full advantage of your clean data by applying standards."
To summarize: what you do not want is ten analysts transforming data in ten different ways with ten different data dictionaries and ten different model scoring and ranking protocols. That's a recipe for analytical anarchy and direct marketing failure.
Take full advantage of your clean data by applying standards.
"War is Ninety Percent Information" — Napoleon
Long before the age of big data and analytics, Napoleon Bonaparte made this astute observation: "War is ninety percent information."
Clearly, the famous French general attributed the majority of his military success to having the right information.
I would only slightly modify his quote like this: "Winning a war is ninety percent accurate information."
So, where are you when it comes to competing on analytics? Are you replete with armies of analysts but dangerously uncertain about the quality of the data they are using? Are you spending millions on marketing that might be missing the mark? Are your analytics failing?
"It's time to clean up the data that's driving your analytics. Now."
If this sounds like your situation, commit to improving your data quality this year. Now. Roll up your sleeves and do the hard work required to create an analytical foundation built on high-quality data. Allocate the right budget. Make it an enterprise priority. Don't bury it and don't delegate it.
Here are my three closing thoughts. If you only remember three things from this post, make it these:
- It's a war out there. It's a daily battle for the hearts, minds and dollars of your customers.
- You must leverage analytics or you will be "wandering out onto the Web like deer on a freeway."
- But, first, it's time to clean up the data that's driving your analytics. Now.
Download "10 Data Hygiene Mistakes to Avoid"
Are you making data hygiene mistakes that are wasting your marketing dollars, limiting your revenue growth, and squeezing your profits?
Find out. Download this valuable resource now.