Harbingers of failure

That is the title of a new paper by Anderson, Lin, Simester, and Tucker, the abstract is this:

We show that some customers, whom we call ‘Harbingers’ of failure, systematically purchase new products that flop. Their early adoption of a new product is a strong signal that a product will fail – the more they buy, the less likely the product will succeed. Firms can identify these customers either through past purchases of new products that failed, or through past purchases of existing products that few other customers purchase. We discuss how these insights can be readily incorporated into the new product development process. Our findings challenge the conventional wisdom that positive customer feedback is always a signal of future success.

Is there a comparable phenomenon with new economic ideas? For the pointer I thank Justin Nazari.


It isn't until page 25 that the authors address who are these people. It would be nice if they had more data about that, but based on their description I'm thinking the Harbingers are recent immigrants. They are people that consistently have different product selection criteria, and that could be people from other cultures.

I don't think this applies to economic ideas or very much else. It's not like these people are like that unlucky character in Li'l Abner who constantly had a cloud over his head.

Reading what little data they provide I think the male proportion are clearly either engineers or the sort of guy who goes to pickup artist seminars.

As to the women, I'm sort of clueless, but I suspect it is the sort of single/divorced middle aged or older professional woman who still dresses like she did at her first job out of college.

The prevalence of beauty products and the implied income required to make more purchases, coupled with my observations regarding people who bought zunes and complain that the store stopped carrying something, are the basis of my almost baseless assumptions

In my business, we see products that really on appeal to designers.

These often flop because they often have drawbacks that can be overlooked by designers who admire the overall concept.

+1 As someone who has worked in architecture and product design this is definitely true. Designers especially recent graduates create products that are just ideas and have very low practicability. Students show me portfolios of houses that fight against racism or more frequently household items stripped of usability, chairs that you couldn't sit down in etc. Consumers who think of themselves as hip early adopters are drawn to purchase this type of thing but is is a very small market.

The only examples I could see were Diet Crystal Pepsi and Frito Lay Lemonade. I don't know about DCP, but FLL is a kind of Frankenproduct . . . Frito Lay + Lemonade are not two things that should not go together. It's kind of like a Doritos Taco but so very different.

I wonder why they didn't market it through their parent Pepsico Brand.

Pepsi lemonade already exists, it's called Tropicana Lemonade. I think the idea was "people drink lemonade with Frito-Lay chips, so why not sell both?" But salty and thirst quenching don't go together in the consumers mind.

Like beer and pretzels?

I think what they missed is intent. People don't eat chips *in order to* get thirsty. It's like cross selling cholesterol reducing statin pills with steak.

It's a side effect people would like to forget not be reminded of.

This Frito Lemonade is making me thirsty!

I hope they included the Sega Dreamcast in their data....not that I'm still bitter about it or anything.

I'm definitely one of these folks.

My wife thought I should open my own market research firm where I'd be proprietor, sole employee, and entire sample size. Big companies would show me mockups of products they were considering launching, and if I'd say "I'd buy that!" then they'd cancel the program, fire the guy who came up with idea, and bury the mockup at Yucca Mountain under concrete and steel.

"Is there a comparable phenomenon with new economic ideas?"

Malcolm Gladwell and I are doppelgangers when it comes to coming up with ideas that are popular and true.

Reading the References section of this paper was a trip down memory lane for me because I've been sailing or golfing with quite a few of the authors of the still-classic papers in the field of New Product Forecasting.

Steve, just for validating your hypothesis, what car do you drive?

It would be funnier if I could say Pontiac Aztek ... But I mostly drive a Honda Odyssey minivan.

So, perhaps your preferences are closer to the average in the case of cars, but far from the average consumer preferences for other kinds of products. In my case, I'm the harbinger of failure for music.

Honda Odyssey is a fantastic car. Wife had to drag me kicking and screaming to the dealership to buy one, but now that I've driven it, it's the only car I want to drive.

An odd thing to read about a minivan

The Odyssey drives like a car, as long as you think a car should drive like a Honda Accord. I do miss having the flat load floor of an SUV - it might have the same cargo volume, but it's less usable.

Who bought the new debt Greece sold last year?

Well you can start with Marc Mezvinsky, in his case it is clearly hereditary.

So in the future, when Apple introduces a new product, they should be concerned if the first people who show up to buy them are wearing an iWatch.

Using a Newton.

The "Apple Watch sales plunge 90%" headline was splashed everywhere yesterday, but bear in mind it's based on data from just the last 3 weeks of June. In the first week, sales showed no decline from the previous 2 months. Is this a real, permanent trend? Too soon to say.

Also, even at the current rate, that's over $1B in annual sales for the Watch. At any place but Apple, that would be considered success.

Yeah, but, the problem is that Apple really needs a hit new product category because it's running out of headroom in existing product lines. Already, fewer and fewer people are upgrading Macs or iPads any more just because there's a new model:


And the same likely to be true of iPhones soon.

Apple has been running on "cool" for a long time. People bought their products because of the perceived status and novelty associated with them. They have to keep generating novelty to maintain the "cool". But now that everyone has a tablet and a smart phone there's little novelty, so people are moving to the cheaper Android-based alternatives. Why pay extra for the 6th generation iPhone when there's a million smartphones out there and nobody is going to look at your phone and think "wow, an iPhone!" anymore?

Apple products have a better build quality. The problem is often their software and their constant updates sucks so bad that I prefer Android.

"Apple has been running on “cool” for a long time."

That's certainly true of a lot of Apple products. But the iPhone is still an easy to use, high quality product that works well and is reasonably priced for it's market niche.

The Apple Watch is a new category -- it's a first of a kind. Just as the original 128K Mac needed time to build both a user base and an ecosystem of applications, the watch needs that time too. This is like 1984 all over again. And it's the only game in town, so like eBay and Amazon they are going to own their niche.

Unless, of course, Google makes an effort to get ahead of the curve. I wouldn't bet on that. An effort, maybe, but not a successful effort.

I think the problem is whether or not very many people are going to want one at all. The screen HAS to remain tiny. Processing power and battery life are going to remain problematic given how thin and light the device has to be. But even supposing the battery/power problems could be solved in a couple generations, and the watch could do everything that a smart phone could -- I still wouldn't want one. I'd much rather have the pocket version with a usable screen than the wrist version with a tiny screen.

The "smartphone on my wrist" concept is why this product is flailing at the moment. Who wants a smartphone on their wrist when there is a much better one in their hand?

Apple (or someone) needs to find out what the application is for this hardware on the wrist, that is neither a timepiece or a smartphone.

I actually didn't see any headline. My comment was based on the fact that I've literally never seen one "in the wild" and I work in the middle of the city. It's also pretty much never talked about any more. I like my iPhone as much as the next guy but there are some real Apple-heads out there who probably feel condemned to wear their iWatch for the rest of their lives lest they ever admit they don't find much value in it.

The only people I've ever heard of who get quality use out of their Apple Watches are people who bike to get around NYC a bunch. They can get directions as they go from the watch and also still be able to send and receive texts.

Needless to say, this is a small demographic.

Likely a shrinking demographic if that is how it is used.

Works for Republican "base" who glomed for the idea that tax cuts pay for them selves, then for alienating minorities (anti-immigration reform), then jumped onto using the debt ceiling to shut down the government. They were also early adopters of "Obamacare is a failure."

Ain't is strange for such a crew of failures that the Republican position in state and federal legislative bodies is as advantageous as it has been since the 1920s.

It wasn't alienating minorities that lost Republicans the White House; it was alienating White people. One year after the gov't shutdown, the Republicans won a major election victory. Most Americans still don't like Obamacare (and it remains a failure). Most Americans still like tax cuts.

Except there's no evidence that the GOP won over any new voters in 2014 (or 2010). Both mid term elections featured low turn-outs, in fact the 2014 turn out was the worst we've seen in a long time. The real truth is not that the GOP was one white more popular than it had been only that the Democrats were less popular in so far as they weren't bale to get their 2012 (or 2008) to turn out in the same numbers.

When you say "Most Americans do not like Obamacare" what you are saying is that most Americans react negatively to a word. Most (no, not all) the specifics of the ACA remain fairly well accepted and most Americans want the problem areas fixed, not the whole thing repealed.

I have no idea what you think the GOP did that alienated "white people". Unless you are referring back to the dismal denouement of the Bush years. Pointless and endless foreign wars and a collapsing economy are a bit off-putting, though not just to folk of non-swarthy skin.

"The real truth is not that the GOP was one white more popular than it had been only that the Democrats were less popular in so far as they weren’t bale to get their 2012 (or 2008) to turn out in the same numbers."

Yes, the Republican Party had low numbers, almost as bad as the Democratic Party.

And where did the voters go? Certainly not the Communist Party, Libertarian, etc. Maybe we need a Crowd Party -- no fixed agenda, except that it reflects what its members say they want.

Lower taxes and free health care -- that's what we want! No nukes or fossil fuels and plentiful cheap electricity -- that's what we want! Vote Crowd Party!

Both mid term elections featured low turn-outs, in fact the 2014 turn out was the worst we’ve seen in a long time.

From 1960 to the present, the median turnout for mid-term federal elections has been 37.15%, The turnouts in 2010 and 2014 were 37.8% and 36.3% respectively. The only recent mid-term election which was any higher than that in 2010 was that in 1994, the result of which does not fit your thesis. The last federal mid-term which had a turnout any higher than that in 2010 and wherein the Democratic caucus increased in size was that in 1982 (and, by the way, the Democrats gained no Senate seats that year). Anyone old enough to have voted that year would now be 50 years old at least.

they weren’t bale to get their 2012 (or 2008) to turn out in the same numbers.
The median turnout in Presidential elections since 1960 has been 54.35%. That in 2012 was 53.6%. There have been no novelties in turnout patterns of late.

"When you say “Most Americans do not like Obamacare” what you are saying is that most Americans react negatively to a word."

-And that's what matters, doesn't it?

"I have no idea what you think the GOP did that alienated “white people”"

-Romney was nowhere near conservative enough to become president.

Your thesis is that demobilization of voters due to Romney's pallor summed with the number of swing voters attracted to self-assuredness exceeds the number of swing voters who'd have defected to the Democrats had someone more robust run. The problem with that thesis is that you did not have a depressed turnout generally. Turnout was about normal. Also, the population of swing voters tends to be modest and much smaller than it was a generation ago (Karl Rove claimed about 7% of the electorate ca. 2004). All that may be true, but it seems a stretch.

If you pop over to The Guardian you will find people who believe that every electoral failure by the left has as it's root cause the abandonment of true "socialism". If you go to some right wing equivalent every failure on the right is because the party is no longer following "true" conservatism. In reality most people want neither of these things and parties win elections either by bringing the centre or by convincing people that their views represent the centre, or by lying about their views.

Mr.Ed. Spot on. In politics, the extreme Left or Right electorate (usually all morons on both sides) is the harbinger of failure. When a Party adopts policies that appeal to its moronic base first, it will surely lose.

Especially since the extremes on both ends are moving further away from the center, thus making their policy preferences even more unappealing to the center.

That's why the GOP is going to lose in 2016 (bar some miracle where a moderate Republican gets nominated). The Dems will move further to the center with Hillary, while the GOP will move to crazy land with Rubio or some other cartoonish character.

I'm not sure. Romney was pretty right down the middle. He's not in the White House because a lot of conservatives just didn't show up to vote.

If Romney had been more to the right, he'd have lost even worst.

That’s why the GOP is going to lose in 2016 (bar some miracle where a moderate Republican gets nominated). The Dems will move further to the center with Hillary, while the GOP will move to crazy land with Rubio or some other cartoonish character.

Rubio's a perfectly banal Republican pol (former speaker of the lower house of the Florida legislature); don't know what you've been smoking. The salient thing about him is not abnormal programmatic preferences but a deficit of preparation for the office.

There is almost no precedent in the history of popular balloting for the presidency for a political party to experience anything but an erosion of support in the course of 3d, 4th, and 5th attempts to retain the White House. This happens under optimal conditions (see election of 2000). The remarkable self-confidence people assume in assuring us that will happen this year is...fascinating (and amusing when done in conjunction with calling other people 'morons').

As I said, all you got to do is look at who supports Rubio.

Wait till they start yapping about "foreign policy" to see real voter alienation.

The "conservatives" within the GOP are definitely the harbingers of failure. Whatever that moronic base latches on to, is for sure an electoral defeat for the GOP.

I think you need to stop pretending you know manure from apple butter.

Of course, the flip side is that some customers stick with old products and some economists stick with old economic ideas long past their sale by date.

John Kenneth Galbraith is dead.

Long time ago I collaborated on a college radio emission. So, it was about rock and just a few of the new bands we presented enjoyed moderate success, zero had wide commercial success. I understand perfectly the early adoption issue, i.e "man, you gotta listen to this band". The curious issue is that you got your space on the radio and you call yourself an expert....... don't look down at me, I was in my second year of bachelor.

Perhaps the lesson is look at your consumer profile. For example, if you sell music to young people ask popular DJs and average teenagers not smug college radio idiots. If you sell luxury cars, you ask the 50+ year old consumer. I always think about this when someone preaches about Tesla success saying that young people is a good % of consumers. How buys $100K+ cars? People with grey hair. Stop listening to the young and rich, they're not a significant % of the consumer population.

Is it statistically more efficient to use inverted results from a focus group whose preferences you know are antipodal to your target market? Rather than the conventional strategy of flogging a cohort with median preferences aligned to your target?


Exquisitely bad taste is more consistent than exquisitely good taste ?

I think we're arguing that they're more reliable?

The coolest kid on the block may vary from week to week, but the biggest dork is always the biggest dork?

So, this approach is more of a risk mitigation strategy? Than trying to identify a super duper product.

Could be. Interesting hypothesis.

It would seem in the hypothesis is true, you would use both approaches, assuming the cost of tow different market studies is reasonable. Also, perhaps you can use the information to tweak the product in some cases.

Preventing failures is easier than predicting success?

Isn't that like telling you don't add vinegar to your new ice cream product? How does that help much.

Like science reporting (think the ENCODE "revolution"), I would offer that any economic theoretical novelty that quickly gets into the mainstream press with either an "!" or a "?" after the title is a signal of non-importance (or overselling, at the very least).

There are a very small number of people who are so sightless as to cause me to reflect considerably on my position if they and I agree. And if these same people were wearing a new brand of clothing, I probably would not wear it.

Now you need to ask yourself, "Are they the Harbingers? Or am I?"

If you look around the table and don't see who the Harbinger is, it's you!

"Is there a comparable phenomenon with new economic ideas?" Yup, it's called macroeconomics.

My wife asks my advice on what to wear so as to know what not to wear.

Love my Zune!

You are consistent. Look at your 6:18 am comment.

The study looks a little like it is finding a statistical artifact:

Most new products fail (61.1% within three years) so people who buy new products that go on to fail 67% or more of the time, just look like people with random judgement--or people who really lack of any judgement at all. The real harbingers are people who buy a lot of new products, but not the new products that suck.

The sample size is huge, so the effect is statistically significant even if it is fairly small (an increase of about 10% in failure risk). But, there are some important methodology issues:

1. The use of only four bins, when you have that big of a data set, makes no sense. If the effect is as hypothesized and you broke it into say ten bins, you would have a much more convincing result. The failure to do so suggests that this may have not be presented because it didn't pan out.

2. The study does a poor job of distinguishing vultures from the perennially unstylish, which timing and discounting would show as noted below. If the Harbringers only buy after the product is in a death spiral (which seems to be the case for many of them) then there is no predictive ability.

On a side point, there really are duds out there and stopping them quickly can be economically important and also important from brand image. For example, someone I know was selling a new product at a $400 price point in a well established brand with a huge nationwide roll out and expected nationwide sales of about 200 units in the initial promotion rush with a significant specially trained sales staff. Instead, total sales, nationwide in the initial promotion rush were roughly 4 (i.e 98% below expectations), some of which may have been from sympathetic relatives of the sales staff. Getting something like that off the shelves ASAP is an economic imperative.

My guess:
Retail outlets quickly realize that certain products are not selling and sharply discount them. And the customers in question are the sort of people who purchase steeply discounted products because they like to think they are getting a good deal. i.e. you might find them buying up the steeply discounted frito-lay lemonade in droves, because it's priced very cheaply and the retailers want to unload it.

So I would bet less that these people are not really "early adopters" but rather this is some secondary phenomenon to retailers being very sensitive to what is selling and what isn't.

From the paper:
Harbingers tend to choose less expensive items and are
more likely to purchase items on sales and items with deeper discounts.

So what comes first? The product failure? The deep discounts? Or the Harbingers?
Paper suggest that the Harbingers come first, but if the harbingers are the sort of people who buy steeply discounted items it may be the product has *already* flopped by the time the harbingers start buying it.

So maybe they are more like vultures.

Getting the consumer carrion. I sometime read columns by Susan Llloyd (who has 8 or 9 children, I forget). Having a large family, she has some strategies for household budgetry. Her advice: buying cheap is not always buying economically.

Link? I sometimes hunt for large family writing that's not geared towards the very poor, but I've been mostly unsuccessful.

She writes for The Latin Mass, which does not have an electronic edition.

Oh well, thanks anyway.

True, but it might be the difference of "I'll try it' or 'Pass'.

If carrion birds are circling your product that's a harbinger of failure.

"If carrion birds are circling your product that’s a harbinger of failure."

Thread Winner!

Try adding "LDS" to your search terms.

Krugman is right about 25% of the time

That's only when the odds are even money.

More likely is that a segment of the market will buy anything and being able to sell to them is significant of nothing about the product. The ability to sell beyond these buyers of novelty is a sign of market success.

Look at movies, some segment of the population will go to even the worst movies ( Fans of Adam Sandler ) but mega hits break into broader markets.

Or Music.

Personally, I like quiet, uncrowded espresso shops. But the ones I find keep closing!

Possible data sources on harbingers and products.

Retail shops routinely throw away products that flop or slow moving. Some dumpster drivers resell them on FB groups.

"Inside the World of Dumpster-Diving Beauty Scavengers"

while it wasn’t so clear at the markets, in these Facebook groups the origin of these products is totally transparent. Many of them even including the word "dumpster" in the name.

If a customer tends to buy failures, then the next new product they purchase is more likely to be a failure. For example, customers who tend to purchase a successful product like a Swiffer mop are more likely to buy other ultimately successful products, like Arizona Iced Tea. In contrast, if a customer tends to buy flops, then their next purchase is also more likely to fail. For example, customers who purchased Diet Crystal Pepsi are more likely to have purchased Frito Lay Lemonade (both of which failed).

I.e. essentially customers who buy conventional and established products are signals that a product will succeed.

Not really unexpected, but I expect purchase datasets have been hard enough that this work doesn't happen repeatedly (so this paper is "novel", in the sense of being unfamiliar, though not actually new, especially not as an idea).

I wonder, here is how much their definition of product failure actually works though - "We will initially define a product as a “failure” if its last transaction date (in the store-level transaction data) is less than 3 years after its introduction. If the last transaction date is after this date then the product is a “success”. How many of either short run, novelty products or products in fast moving industries are a success by this criteria? It seems that almost all books and movies "fail" under it, albeit with a much longer tail reflecting their ease of storage and production. A more accurate definition of failed products would be ones which have a causal, negative connection to company profit or size, or which directly lose money for the company (even if they are overall beneficial for consumers). But that would actually mean you need more information than they have.

Comments for this post are closed