Which of your retail marketing efforts are driving sales? 

How confident are you in your answer?

Our work with leading national retailers and restaurants shows that 47% of digital marketing spend is wasted — as in the ad impressions drove no incremental sales for the business. This research includes all types of large chains such as big box, department stores, specialty, discounters, QSRs and full service restaurants. Ironically, these retailers and their agencies were already using some form of marketing measurement to monitor performance. So what gives? The root of the problem is flawed metrics; you can only manage what you can measure. 

For example, if your team is using click-through rate (CTR) to track engagement with your digital marketing, they will be optimizing for clicks. Unfortunately, those clicks are not typically correlated with sales that occur offline. No matter how you analyze the data — first or last click attribution, weighted, time decay, or machine learning — using CTR can lead you to de-optimize. The problem is that few people ever intentionally click on advertisements. Moreover, CTR ignores all historical and non-digital data such as brand equity, brand preference, previous purchase experience, and word-of-mouth referrals.

Location Data Isn’t Great Either

Many retailer marketing analytics leaders have taken the next step and moved beyond CTR by using location data to measure their campaigns. Essentially, they try to track if advertising results in increased store visits. While this approach seems more logically sound, it can have accuracy issues due to lack of scale.

A location data seller may accurately claim to have data on 150 million phones, but most of those phones are not tracked on a continuous basis. Typically, the customer has to be surfing the web while in the store or have an app enabled to share location even when it is not in use. The impact of these constraints is significant and varies by vendor. There is one thing you can be sure of: If your location measurement company isn’t clearly reporting sample sizes and match rates, you know they are probably too small. Insignificant samples are often accompanied by opaque modeling and extrapolation practices.

Small sample sizes just compound the fact that using location data as a proxy for sales impact makes two big assumptions: (1) anyone near my store buys (especially challenging in malls and urban areas), and (2) every purchase is average. The latter is particularly flawed. The standard deviation sales/ticket or basket size is likely much larger for your chain than you would guess.

6 Ways to Improve Accuracy and Usability

So, how can you improve the accuracy and usability of marketing measurements?

1. Increase sample sizes with broader data sets.

Increasing sample sizes allows you to drill into the data with more confidence. Knowing that a recent campaign drove a 2% sales lift is nice, but the information isn’t actionable. If you could further see that one audience had a 4% sales lift but two other audiences showed zero increase, that becomes highly actionable information. Just like other forms of market research, the larger the sample, the greater your ability to drill into the insights.

2. Test for statistical significance.

Significance testing is required to know if the performance differences in your measurements are real or just noise in the data. The last thing you want to do is take on a bunch of work to measure, analyze and adjust a campaign only to negatively impact the sales of your brand.

3. Obtain results ASAP.

The sooner you know what’s working and what isn’t, the faster you can make adjustments. Slow, post-campaign measurement can make for good resume bullet points. It can help you plan for next year if competitive activity is similar. But in-flight measurement is actionable. With real time insights, retail marketing spend can be shifted to what’s working and improve results now.

4. Check for control group fit.

Control groups are used to measure the impact of something, like a drug or an advertisement. In retail marketing, the control group is the audience that doesn’t see the ad campaign, while the test group does (and hence is often referred to as the exposed group).

A high-quality control group should have be as similar to the exposed group as possible. Often this is defined in terms of demographics and location, but you also need to ensure that the test and control groups had the same likelihood of purchase before the advertising period. A good fitting control group ensures you make apples to apples comparisons. Poor fit can easily lead to accuracy issues and taking action that actually hurts the sales of your business. Ask your measurement vendor to see a comparison of pre-campaign and post-campaign spend for your test and control groups to check fit.

5. Ensure data controls and consumer privacy protection are in place.

The California Consumer Privacy Act will take effect in January 2020, but it’s still far from settled law. There are many amendments still under consideration. Additionally, several other states are debating how their own future laws will be similar or different. No one can predict exactly what future regulation will bring. To evaluate whether the consumer data you use for measurement will continue to be available in the future, ask yourself or your measurement provider if you have the right to collect the data, the right to use the data and the right to share the data.

6. Use a neutral party. 

Marketers too often accept inherent bias in their KPIs because campaign measurement is thrown in as a value add by the the same companies that were hired to do the ad targeting. This is particularly common with location based targeted ads. But would an ad exchange tell you that 47% of your spend with them was wasted?

In summary

A retail marketing campaign can only be successfully optimized if the measured key performance indicators reliably represent business sales. With large sample sizes, quality control groups and statistical significance testing, you can drill down into the details to see what’s working and what isn’t. Armed with that insight while your campaign is still running, your optimization efforts will be rewarded with higher sales.

If you have questions about this or the ways in which Commerce Signals can help your business, please contact us.

Share This