Simply put, Penfield makes analytics Work.
A client recommended this blog post summarizing the main components of price promotion measurement. Across industries, our experience is that lift and halo almost never offset the markdown, cannibalization, and pull-forward effects. That said, promotion is a critical marketing lever, and avoiding it entirely is a pretty bad idea.
As companies continue to try to actually deliver on the promise of “Big Data,” we’ve seen a number of them fall flat on their faces in sometimes very public ways.
- In this article, a business school professor and Big Data expert has a spooky experience firsthand.
- And when firms sell their customer data (or even appear to do so), the response can be immediate, vicious, and public. An online poster to the huge online community Reddit posted this:
-I used a throw away hotmail that averaged maybe 10 spams a day that I’ve had for 5 or so years.
-I went to Geico.com and used their “Get Free Rate Quote.”
-The quote was $854/6 months.
-I said damn, that’s a lot. I had 1 speeding ticket when I was 19, -I’m currently 31 and drive a Nissan.
-I didn’t further my search at that time and put it on the back burner.
-I checked my hotmail later, 1,019 emails in 4 days, a letter in the mail from Gieco, an advertisement in the mail from Geico, other car insurance mail magically appeared in the mail.
-I called Gieco and asked them to chill the fuck out. They said they don’t do that. I said that’s funny you do.
-I went to Esurance. $92 a month and no increase in spam mail. Just the constant barrage of shit from Geico.
- In the same vein as Prof. Johnson above, I paid the fee to get a look at what Acxiom knows about me after this article in the NYTimes. The process was so complex that it became clear that the entire process was simply for PR purposes. It required mailed physical checks (what year is it? no credit cards, no online system at all?), email confirmations to a person rather than a system, and locked, encrypted PDFs. And after coming out the far side of the process, I received only the bare basics- the last 5 addresses I lived at, family members, etc.
So what’s the bottom line? Companies leveraging big data must:
- Drive the biggest insights from the aggregated data. (Trust us, there’s gold in that data)
- Be overly cautious about how it’s being used whenever it is precise to individuals.
- Fully disclose what they keep, and allow anyone to opt out. The small numbers of opt-outs will pale in comparison to the value of avoiding bad press.
If it doesn’t happen with self-regulation, it won’t be too long before an external force intervenes- either through consumer revolt, or worse, government intervention.
Analytics should rarely live alone- they should support (but not replace) a broader ecosystem of decision inputs. But we can empathize with a common challenge- “what do I do when I’m being told two contradictory things?”
Hence the adage, known sometimes as Segal’s Law: ”The man with a watch knows what time it is. The man with two watches is never sure.” (more…)
I love Derek Thompson at the Atlantic. But this article compares “attention” and “users” across marketing vehicles, which ends up having very little to do with marketing ROI for advertisers.
The largest advertisers are not dumb- they utilize advanced statistical modeling and toolsets to optimize their marketing spends pretty precisely. So why do they still spend so much on print? Because it works.
The true story is more complex and nuanced than the charts suggest. If I were to show these to the large advertisers, they would say “Yep. So what?” They know that ”attention” and “users” are not equivalent across marketing tactics. TV controls the whole screen with video and sound, while a little box next to your facebook page is easily ignored, and radio catches you in your car- when you’re less likely to change channels.
But at the end of the day, they simply don’t care much about the mechanism of how or why different ads work. There is enough evidence and proof that they do work, and they work differently, but they drive sales and margin, often with positive ROIs.
It’s this kind of aggravating nonsense and irresponsible writing that sets the field of marketing measurement back years. And infuriatingly, it is read by a lot of people.
The first paragraph: “Measuring marketing spending used to be pretty easy, at least in theory: You’d run a big ad campaign and then see if your sales rose. If they did, great. If not, then you wasted your money.” (more…)
There’s no shortage of coverage of late for “big data analytics” and the hype around it: Forbes’ “How Target Figured Out A Teen Girl Was Pregnant Before Her Father Did” and The NYT’s “How Companies Learn Your Secrets”. For the moment we’ll set aside whether this is new (we don’t think so), whether bigger data equals better data (nope), and whether the future will look very different than the present (not likely). Instead we’ll focus on how new grads should best prepare for this burgeoning marketplace.
Download the below in a 1-page whitepaper on circular ROI.
There’s been a lot of noise in the advertising industry about the impending death of print as an advertising tactic, and it’s certainly hard to argue with ongoing circulation declines from data at ABC and NAA. However, the ROI on circulars is actually increasing. Why?
Many of our clients have trouble striking the balance between the Art (their intuition and gut feels) and Science (analytical results and ROI) of their marketing plans. As a rough starting point, here are the three most important times in which to ignore the analytics: (more…)
Even more evidence of retailers struggling under the continuing assault of in-store price scans is here at the Chicago Tribune. A while back we predicted this would happen. Next up: doubling down on in-store service (or, back to the basics of retailing).
The NYT is running an op-ed piece about Amazon’s new salvo with it’s Price Check app which I discussed a few days ago. It explores Amazon’s mindset in creating it, though I doubt the conversation started from the question “What can we do to crush small independent booksellers?” In my experience, things like this often start with a few developers saying “hey, wouldn’t it be cool if…”
A great read is here at NYTimes.com on their “You’re the boss” blog about running a small business. Our reactions are:
1) Pretty amazing that the subtle nuances of a single marketing vehicle can be important enough to make the paper- and no small paper at that!
2) If this experiment is representative, it’s a testament to the usage of search engines these days.
3) … and that there’s only one player!
4) But more distressing is the implication that paid search can influence natural search. I’m a fan of Google, but much less if they’re breaking their “Don’t be evil” pledge.
We recently delivered a marketing assessment to a marketing team led by someone who had recently come from a large chewing gum brand. He relayed a story that gave us a real-world validation of our analytical results.
At his prior brand, they had used Marketing Mix Modeling to measure the impacts of his marketing activities. The contribution was measured at about 20% of his total sales volume, which meant that without marketing, his sales would have only been 80% of what they actually were.
When people look to measure the response of Hispanic advertising tactics, the knee-jerk response is often to try to isolate Hispanic populations sets so as to see the impacts more clearly. Unfortunately, not only is that very hard to do, but it’s also misleading.
An article in today’s NYTimes cites a McKinsey study that shows the value of analytics to healthcare in the US could be $300 Billion. They cite a challenge of finding people with “deep analytical” skills, which is certainly true, but what I’ve seen is that the cultural and operational difficulty of changing processes in healthcare (in hospitals, at least) is a far bigger hurdle.
I’ve been lucky enough to experience a lot of different firms in a lot of different industries through my career. Some places that loathe change, some fear challenging (or sometimes even discussing) an executive’s point of view on a topic, but these are usually limited to certain pockets. However, the attitude universal in healthcare seems to be “yes, it’s broken, and is causing inefficiency if not outright bad patient care, but it’s too hard to change it so I’m going to keep my head down.”
In fairness, it is far harder to change things in a hospital than in business: the incentive structure dictated by regulation and insurance are certainly daunting. But you could argue that the gains in human and economic terms are larger than in the private sector, so the opportunity should be that much more worthwhile.
I’m generally pessimistic on the ability of analytics to identify, measure, and capitalize upon improvements in care or efficiency in a hospital setting, until the organizations fundamentally change their cultures enough to allow it.
How you measure marketing is important, but some of the most critical conversations happen before any analytics. I shared some time with the CMO of a large retailer some time ago whose performance (and a rather sizable bonus) was contingent upon his ability to drive bottom line profits at the company.
This particular company was experiencing some significant operational challenges- there were systems issues and significant customer service problems that were hurting sales. He half-jokingly said that he should drop all this marketing stuff he was doing and start learning how to debug IT systems to get his bonus. (more…)
One of my favorite outputs from our marketing analytics models never gets its due. It shows which factors are moving the business results more than others. For instance, if I were running an online retailer I’d certainly be interested in hearing whether my sales are more sensitive to consumer confidence or gas prices. Enter the Tornado Chart: (more…)
Tom Davenport’s Competing on Analytics was extraordinarily helpful when I was setting up supply chain, marketing, pricing, and competitive intelligence analytics teams inside Home Depot. His framework for how companies build analytics capabilities is spot on (we started at a low stage 2, and made it to a mid-stage 4) He’s still at it, and continues to be way ahead of the curve. Scott Brinker has a great summary of his talk at eMetrics here.
There was a great article on “black hat” search engine optimization in today’s NYTimes. Under most circumstances I’m the last person to recommend reading comments on the internet, but in this case there are some fascinating points. Many focus on Google’s transition from a spunky “Don’t Be Evil” startup filled with hypernerds into the behemoth that it currently is (one poster even calls Google “the singularity”).
I’m lucky enough to be a guest lecturer at both Columbia Business School and NYU’s Stern School of Business. The session is in two parts: first we do a general overview on the state of analytics, and then we cover how they should become an “informed consumer” of analytics in their marketing roles.
I’m happy to report that the future marketing leaders are sharp, full of energy, and are the exact opposite of the old marketing stereotypes. There is still the excitement and strong creative bent, but these students are using marketing as a strategic toolset for the larger business, and will go toe-to-toe with any finance or line operating role who considers it an optional nice-to-have. It’s thrilling to discuss their ideas, and their energy is contagious.
When we’re doing marketing analytics work, we’re often asked by clients how much they can spend in a certain media. The implied second half of the question is “…before my ROI tanks.” It’s a reasonable question, and one that often opens the door to some analytics as well as starting good strategic discussion around the goals of marketing (both financial and non-financial). Most of the time we’re measuring both offline (TV, Print, Radio, etc.) and online (search, banner, social) tactics. One key input to the measurement is Adstock.
What is adstock? In traditional media like TV, Radio, and Magazine are subject to two effects: saturation and decay. Saturation is the idea that twice the media doesn’t always give you twice the impact, because of diminishing marginal returns. The theory is that the as you continue to advertise more and more, you either hit the same people more and more, or you hit people less attractive or likely to respond. Decay is that your marketing has a lag effect that decays after the actual execution.
Once these factors are measured, they can help advertisers understand how much to spend by showing where diminishing returns happen. But few advertisers think this way when it comes to digital advertising tactics, because they rely on the instant response metrics that emerge from simple conversion tracking. But with a fairly simple analysis using no more than excel, digital adstock can be measured and then incorporated into budgeting and allocation decisions.
When clients engage us repeatedly to track their marketing effectiveness and ROI over time, we are allowed a much deeper and more strategic role with them. But sometimes, we report large swings in ROI vs the prior year- and it’s a shock.
It shouldn’t be. Here’s why: (more…)
Passing Google’s Advanced Search Exam was harder than I expected.
Though I’ve posted on this topic recently, I’ve had requests for a more specific look at how to start and maintain data today for future analytics endeavors. So, some tactical and specific advice: (more…)
We often find ourselves in the situation in which our recommendations require changes that are large in magnitude or challenge the sacred cows of an organization. In many of these cases there is tolerance for a test- either as a way to double check results before committing to a large change, or to “put your money where your math is” to dispel the organizational mythology.
But then a critical question is “how much should we spend?” (more…)
We work with many companies that are just getting started on the pathway to analytics. Their data is usually better than they think it is, though it frequently is far from usable in the state we find it. Just this week a client said that they had the data we wanted, but that it would take 150 man-hours to get it together for us. A little skeptical, we probed deeper to find that this client had between 12 and 19 individual data files per week, each representing a category of products. We needed 156 weeks worth of data, which meant we were sent over 2,100 individual files. We built a quick macro to extract the relevant data out of the files (which made us heroes in the eyes of the client), but it was still an unpleasant task.
So, if you’re looking ahead to building an analytical capability and wondering what you can do today to ensure success tomorrow, here are a few rules of thumb: (more…)
As digital tactics continue to explode in breadth as well as depth, we’re being asked about how to think about remarketing in the context of other advertising tactics.
The answer is simple: you shouldn’t.
“Cotton prices have jumped 95% in the last 12 months, so we’re going to cut advertising.”
That doesn’t seem like it makes much sense, yet it’s happening today. Why?
Here’s a quick primer on when you can use a test vs. control (or experimental design) and when you should go for a fully specified regression model.
Test when you:
- Need just one answer: You get only a measurement of the impact for which you test. (more…)