Learn

The Importance of Data Quality (Garbage In, Garbage Out)

Insights / 03.14.2018

Red Door


While there is still some uncertainty about who first coined the term, "garbage in, garbage out" (GIGO), the consequences of leveraging bad data can be found littered throughout history. One infamous example involved Lockheed Martin and NASA, setting them back years in their quest to learn more about Mars. In 1999, after traveling over 130 million miles, the Mars Climate Orbiter reached the red planet's orbit. But, unfortunately, due to a lack of proper data transformation between the Imperial units used by Mission Control and the Metric units used onboard the Orbiter, the craft dipped 105 miles closer to the planet than expected. The result was the total incineration of the Orbiter and $125M lost in space.

GIGO is a Big Problem for Business

With this one mistake costing over $100M, the claim that bad data cost U.S. companies over $3.1 trillion in 2016 alone is not all that far-fetched. This is especially true when you factor in inefficiencies caused by bad data, even when that data is corrected before it becomes an input. This is what Tom Redmond calls "the hidden data factory" in his book, Getting in Front on Data.  

dataerrors-(1).png

GIGO is Becoming More Important Than Ever in Marketing

Several trends in the marketing industry point to the need for companies to embrace data quality management (DQM) and prevent GIGO, particularly when it comes to marketing data. Read on for reasons why GIGO has become increasingly relevant.

  • The MarTech stack is getting ever more complicated. The marketing technology landscape has grown at a dizzying pace. The average enterprise company has 91 different cloud-based marketing technology investments in place, most of which are producing their own datasets. 
  • BI automation is creating efficiencies in reporting. Business Intelligence tools like Domo are creating great efficiencies in reporting by automating the connection to various MarTech platforms, as well as the ETL and data visualization processes. But if the data coming in lacks quality, then the reporting is going to be inaccurate and/or incomplete.
  • Data is driving more marketing decisions than ever. Everything from content strategy to media attribution modeling relies heavily on data that drives strategy and decision-making in marketing. If your reporting and analysis are wrong, then your decision-making will be flawed, too.
  • First-party data is providing smart marketers with a critical competitive advantage. In a world where marketers are swimming in potential data sources, many aren't even taking full advantage of their own customer data, or that of the prospects that interact with them. Doing so can reap huge rewards, whether through audience insights, segmentation, or targeting. But if you can't tie your cross-platform customer data together to create a single customer record, then you'll have an incomplete and siloed view of your prospects and customers. 
  • Personalization and marketing automation is exposing data inaccuracy to consumers. Dynamic website, email, and media creative relies on a system that brings together first, second, and third-party audience data to target individuals and deliver relevant, custom experiences. If your audience data is inaccurate, then the messages that you are customizing can be embarrassingly off-target. 

How to Prevent GIGO

DQM includes the standard operating procedures, documentation, collaboration, and best practices that help prevent missing, duplicated, outdated, corrupted, erroneous, incomplete, and/or hacked data from entering/exiting your MarTech stack (and beyond). While the work of DQM is wide-ranging and iterative, focusing on the implementation, configuration, and integration of MarTech platforms provides ample opportunity to improve your marketing data. Here are six things to keep in mind as you work to prevent GIGO:

1. Marketing channels are the delivery mechanism for serving impactful messaging to relevant audiences. They generate a high-volume of performance data and audience data. While some of this data is tracked through web analytics platforms, front-end metrics like impressions, likes, video views, email opens, and SEO rankings are not. Getting accurate data from channel platforms such as your ad server, bid management, email/marketing automation, SEO, and community management platforms requires a deliberate, structured approach to tagging campaign URLs, as well as advanced configuration within each channel platform. 

2. Tags (i.e., pixels) allow marketers to track front-end user activities from marketing channel platforms. Tag management systems make it easy to centrally manage and configure MarTech tags on a website/app, without ongoing IT support. If a tag management system is not implemented, it’s likely that tags are missing, duplicated, or misplaced. Even if a tag management system is in place, expert implementation and configuration are required to get accurate and complete data. Additionally, putting data layers in place is extremely valuable in making an implementation more reliable, and can facilitate integrations with CRM and other platforms (passing user and order IDs, cart values, etc. in the background).

dataflow.jpg

3. Web analytics platforms provide user behavior data for websites and mobile applications, using site tags and URL tracking parameters. Analytics platforms allow marketers to know how a website is generally performing (total visits, load time, bounce rate, etc.), how users are engaging with the site (pathing, page views, etc.), what actions they are taking (conversions and micro-conversions), as well as how each traffic source (channel) and campaign is performing on-site. In order to provide accurate and complete data that is prepared for data warehousing (or a data visualization tool like Domo), web analytics platforms need to be implemented (site tagging) well and configured properly. 

4. Full-featured Data Management Platforms (DMPs) aggregate and segment anonymous audience data, feeding that data back to paid media, email, and content management platforms for targeting and personalization. These top-tier DMPs can provide de-duplication and merging of customer touchpoint data into a single customer record. On the other hand, more basic DMPs have a narrower focus, only segmenting data and pushing distinct audiences out to programmatic ad platforms. 

For reference, first-party data includes website cookies, email subscriber activity, POS transaction data, and digital ad impression data. Second-party data comes from partners who are willing to share their audience data. And third-party data includes demographic, psychographic, and behavioral data about audiences not previously touched by the company, or other contextual data (i.e. local weather), that can be used for targeting and personalization. 

Without a DMP platform, ad frequency can be unintentionally high, if the same person gets targeted across several different ad platforms that are all operating in silos. And, conversions/revenue reporting can be greatly exaggerated, because often times more than one channel contributes to a conversion, and all of them are taking credit for it. With a DMP, it is important to ensure that data is standardized (e.g., conversion types, customer IDs, etc.), so that fields and anonymous unique identifiers can be properly mapped, and duplication can be eliminated, at least within marketing channels.

5. Customer Data Platforms (CDPs) are primarily a system of record for managing disparate data about customers across the company. CDPs deal with your first-party data and are built to protect your customer’s Personally Identifiable Information (PII). CDPs differ in how they create audience segments from DMPs in the way they build customer profiles. DMPs using probabilistic algorithms and CDPs use deterministic algorithms. Probabilistic methods involve associations without known links. Deterministic connect direct links between attributes of your customer and your customer record like their email address. CDPs create audiences and segments which are persistent and useable across all your marketing stack, including ad servers and email. They are built from the ground up to take on customer data. 

6. Master Data Management (MDM) Platforms prepare and orchestrate the moving of data between point of sale (POS), CRM, back-office, enterprise resource planning (ERP), and the MarTech stack. Going well beyond marketing data alone, MDMs create a single 360 view of each customer (inside and outside of marketing), and acts as an authoritative source of other business and operations data across the enterprise. Without an MDM, a company will likely have duplicate records for a single customer (product, employee, or vendor) across different systems. With an MDM, its critical to standardize data (mass maintaining) and incorporate rules to eliminate incorrect data from entering the system. MDM operationalizes your data for all your business systems.

While every enterprise is drowning in data, marketers have a particularly herculean task in wrangling it, since we deal with data created by every consumer-brand interaction. This becomes even more complex as MarTech stack proliferation and fragmentation causes marketing data to get splintered into more and more silos. Often times, the result is garbage in, garbage out, which produces inaccurate reporting, poor decision-making, and even disastrous results. With big data, comes big responsibility. So, make sure that your data is up to par by taking at least a few of these steps toward comprehensive data quality management.

Interested in learning more about data quality management? Contact us today.

Data Analytics & Visualization,Insights

Subscribe to Our Blog

Be the first to get our expert tips, industry insights, Red Door happenings, and more.

Comments

Comments
Comments
Blog post currently doesn't have any comments.