Top reasons consumer data quality is failing retailers
A survey from Lotame finds 60 percent of marketers are concerned with the quality of data available from data providers/sellers. The survey featured 300 marketers in the U.S. who purchase and/or use audience data (first-, second- or third-party) in their roles.
The top reason marketers are concerned with data quality from providers/sellers was found to be the refresh rate. The data-management platform provider noted as an example that someone in the market for a new car in January might have purchased it in February and would no longer be in market in March.
Other concerns were bots, fake social profiles and other nefarious activity that can lead to bogus comments, fake clicks, incorrect bounce rates and wasted bandwidth. Marketers are further troubled by incorrect linking between a pageview and intent or motivation, which leads to broad or inaccurate categorization.
To improve data quality, marketers wanted to see:
- Transparency on data sources;
- Consistent data labeling across providers;
- The creation of universal industry standards.
Those actions were selected above changes or disclosure to refresh rate and consumer opt-out of data sharing.
Other findings:
- The majority of respondents, 65 percent, would only be willing to spend an extra five percent or more than what they’re currently spending on data buys for higher data quality. Only 25 percent would be willing to spend 10 percent or more.
- A quarter are not sure how to measure the success of their audience data purchases.
- While 57 percent of marketers say audience data is very valuable to them, only 14 percent said their audience data buys were very successful. Imprecise targeting is the number one challenge.
“It is this trend toward ‘all things data’ that has made scale a key industry driver for quite some time,” wrote Jason Downie, chief strategy officer at Lotame, in the study. “However, as the data market has reached maturity, the shortcuts that were taken to achieve scale are now undermining the quality of data available.”
Discussion Questions
DISCUSSION QUESTIONS: What advice would you have for purchasing and/or using audience data? Will the arrival of universal industry standards, advanced machine learning technologies or some other actions be necessary to improve the accuracy of purchased demographic and other consumer data?
Poll
BrainTrust
Jennifer McDermott
Consumer Advocate, finder.com
Ralph Jacobson
Global Retail & CPG Sales Strategist, IBM
Marketers want better data quality, but aren’t prepared to pay much more … unfortunately as with most things in life, you get what you pay for and this is particularly true with cheap “dirty data.” There also needs to be a shift away from the more is more approach to contacts. With so many highly-targeted products available, marketers should scale back and pay more for a smaller, more targeted list of a better quality, likely paying the same or marginally more than what they are now but for data that is significantly more effective.
Contextual evidence validated by a universal system imbued with human intelligence at the macro level would be a start. From there filter it into contextual data packets of specific micro-audience data, again validated by the universal system. The universal system is a language in itself, constantly evolving over time improved by ML and AI and millions of human inputs. First things first, to teach a machine to process billions of audience data points requires filtering for human insights and nuances to succeed and be valuable in marketing human audience data to marketers, publishers and agencies.
It is interesting, the percentage of people willing to invest more to solve their acknowledged challenges. Well, you get what you pay for. If you have data quality concerns, you must address them. Otherwise, what’s the point of any marketing campaign, let alone data integrity issues throughout the enterprise and its ecosystem, for that matter?
Data capture, refresh, analytics and insights require the best augmented intelligence available today. And the good news is that these capabilities can be achieved in small, edible chunks, rather than with a “Big Bang” investment.
This is basic stuff today, and the first step in transforming your business for the true digital age.
Nearly everyone claims that their data is clean. The reality is that 90 percent of solid, insightful data analysis time is spent cleaning and organizing source data. The new data paradigm reminds me of what has been said about industry standards. “Standards are a very good thing — everyone should have one.” While everyone has been discussing the merits of Big Data — I try to design and measure the efficacy and viability of shopper experiences using “small data.” Small data being defined as very specific data points that relate to a clearly defined business workflow associated with a key performance indicator/metric. It is near impossible to translate Big Data into a relevant, valued and personal shopper experience. Keeping it focused and small often leads to greater efficacy and viability — as well as more “AH-HA” moments and insights!
When purchasing audience data, retailers need to be cautious. Basing communication to customers on faulty or old data is very risky, as if the communication is not relevant based on inaccurate customer context, consumers may have a negative perception of your brand.
There will always be risks with purchased data. Similarity, B2B purchased data is often very inaccurate. There is no good substitute to curating and cleaning your own data. It isn’t easy, but it is usually worth the effort.
The focus should no longer be on scale, it should be on high quality, high integrity targeted data. If you have a smaller total spend for less data that is “dirty” anyway, then you can apply that savings to buying a smaller more focused but better quality data. Of course, nothing beats having more and better quality first party data. That investment should be the prioritized investment over any purchase.
The best data is always in-house. You know where it came from, you know how old it is, and you know how clean it is. Unfortunately, 3P data is part of the push to acquisition and usually for those just starting out, those with limited transaction volume, or those who haven’t figured out how to capture data from their businesses. The reason retail marketers don’t trust it, even at higher costs and decent results, is that it’s difficult to figure out reliability and results are rarely consistent. The data gets stale very quickly, and whatever the source, it’s even more challenging to go deeper than basic demographics to get to real consumer intent. Not surprised by the 14% success rate reported.
The solution is to test a list before you buy it. Got 100,000 customers on the list you want to buy? Get a sample of a few hundred and try it out. Anyone who won’t offer a sample knows their list is dirty.
My own experience echoes the results of this study. 14% isn’t a surprise to me either. There’s always a risk for acquiring 3rd party data, and I suspect most, like me, have not found a direct correlation between paying more for this data and getting higher quality.
I’d say that’s why you don’t see a willingness of the survey respondents to pay more for better quality — there just hasn’t been enough proof points of this yet.
One reason for this not discussed yet is that there are many not-so-reputable sources for buying such data. Those providers give the whole segment a bad reputation and raise suspicions of data quality more than you would expect. Marketers need to focus on ways of generating this data firsthand as that’s the best way to exert control over the quality.
I agree with many of the comments here. It’s important for marketers to evaluate shopper data, whether it’s “home grown” from a retailer’s FSP program or purchased from data providers. There are four primary criteria that will go along way to ensuring a retailer is building programs on clean, high-quality data:
Source and Collection Techniques – Marketers must ask questions such as, “Is the data provider passively or actively collecting the data from consumers?” and “Is the data survey, demongraphic or contextual data?” The answers to these questions will enable marketers to determine the value of the data set.
Data Quality Methodology – Marketers must also determine internally and/or ask data providers how the data set is regularly cleansed, distributed and protected.
Recency, Frequency, Consistency – These qualities are especially important when purchasing data from outside data providers, who often don’t proactively communicate aging, refreshment and frequency information.
Validated versus Modeled Data – Purchase–based shopper data are a big plus, as these shoppers are known buyers. Marketers should ask data providers what percentage of their data set is validated versus acquired through lookalike modeling.
It can be tempting when eager to get a campaign off the ground to cut corners on evaluating the data set, but as many of the other comments here demonstrate, it is well worth the time and resource investment.