After working in two jobs where my job was using quantitative analysis to try figure out what marketing channels to invest in I have to say this is incredibly hard to get any answers - to the point where you might ask why bother? The current solutions in the marketplace AFAIK are either extremely expensive vendors (think $50k pa starting price) or startups which are most focused on direct response (an example I was shown was web traffic from a TV commercial which let you "win a jeep".)
The best steps you can do for your business are have a very firm commercial idea of what success looks like (i.e. measure LTV, churn etc by channel) and have a robust framework for market research. I personally like the idea of splitting spending marketing spend between branding (where your aim is to promote unprompted brand awareness) and acquisition (where you want CPA < LTV.) Also remember there are A LOT of businesses which spend money in things that an expert can identify as not commercially sound - so audits by external agencys can be useful (i.e. not using negative keywords in AdWords and spending many thousands on direct response campaigns that don't work.)
I'll add the cavaet that rhe worst thing about "attribution" I've found is it puts an unfair amount of pressure in the bottom of the funnel. Brand focused marketers aren't really held to scrutiny at all (lots of marketing teams in businesses I know don't even do total spend / total sales) - I think this can cause some resentment between teams.
Thanks for the links and comment, I'm on the adtech side and largely agree - true measurement of effectiveness is unsolved, and we conflate measuring lift with causing lift.
IMO anyone paying attention in the industry knows the answer is probably less ad spend, aka most inventory is overpriced...but then the incentives take over and the firehose stays on full blast. Publishers, agencies adtech/martech, are not incentivized to right-size ad spend at all, and even for advertisers orgs, the slightest bit of silo/politics incentivizes CMOs to spend & protect their budgets.
It is a select handful of orgs with serious CEO level buy in that are willing to honestly tackle this and spend less, more efficiently. That tends to be a competitive advantage in a sea of waste.
Yep also I think there are incentives to look like you are investing in this space ... and for the vendors who are selling these tools, not many marketing managers are going to be happy if you make their placements or creative choices look stupid
All good points. I think brand marketers aren’t held to standards often because it’s incredibly difficult and expensive to do so. It’s not like you can’t just calculate a quick roas from your tv spot. I’d love for an easier way to measure branding. All I know of are brand lift studies (expensive), PSAs (waste of money at low spend levels), and attempting to segment other ways which is difficult long term with view through conversions.
My first job out of uni was quantitative market research (read: surveys) and at the time I thought it was a waste of time - studying UX you learn that for a lot of questions customers will give you incorrect or unhelpful answers (depending on the wording.) Compared to the more advance marketing technology it is probably better ROI though - I read something recently which I haven't been able to find again that most marketing leaders DO NOT trust their DMP (data management platform).
One of the biggest problems I think of with making brand lift studies more cheaper is you encourage the supplier to recruit people who answer surveys as side-income to answer questions. Places like usertesting.com are now full of "professional website testers" rather than your average customer who may be adverse to tech.
The thing about view through conversions and cross device tracking is maybe only a handful of marketing analytics professionals I know can actually measure the uplift they get from installing the required tracking from these. The first question I think people NEED to ask when investing in analytics is "will I get ROI on this or am I just making more work for myself"?
> "will I get ROI on this or am I just making more work for myself"?
Or more often I see a brand spend money on a brand lift study when that $50k would've been better spends on more advertising.
I work in a trading desk and use a dmp daily. I definitely don't trust it 100% but there isn't really a need to. As long as I set up ~7 strategies (some of which use data from the dmp) I only care which strategy is the best, regardless of how accurate the data is. Of course more accurate is better, but that isn't something I have control over.
Sounds like we've had some similar jobs over the years!
Can't find the more attribution focused talk (see he's given it a few times but can't find the slides - think I found them once searching through slideshare.) Here is one Willem did which is probably a bit too brief to get much out of (was at a CDO conference) https://www.slideshare.net/Corinium/foxtel-at-the-chief-data...
(in the very small digital segment of the much larger advertising market). Astoundingly, the vast majority of ad spend is still done with the traditional level (close to none) of quantitative rigor.
That chart is interesting. Realistically how likely is it that digital spend is going to double in the next five years? Is the consumer economy big enough to justify that without taking share from other media (presumably there are other advertising media)? If (when?) it does stop that's going to have a pretty big impact on google/facebook's valuations and growth rates.
As a consumer, not an ad tech guy, I don’t get digital advertising at all. It rarely if ever drives behavior on my part, and I just don’t see effective use of it. The most useful ads I see are the circulars in the newspaper that I seek out.
I still remember brands and jingles from my childhood on radio and TV, but today, I can’t think of anything digital advertising that stands out as other than clumsy and dumb or useless retargeting. I’ve been a Gmail user for 12-13 years, and the best they can do is some phoney “Win an Aldi Gift Card (very enticing) for taking surveys” and “Milfs are looking for you”, seemingly triggered by some spam I receive.
I’m also getting bombarded with air conditioner ads from Amazon because I bought an air conditioner, from Amazon, a few weeks ago. For all of this big data and machine learning investments, they are trying to sell a man sitting in a 60 degree room another AC — which given the price and free shipping they probably lost money on.
People always say that on HN because they think they are unique or immune to advertising somehow. Its not like advertisers toss out some ads and then you automagically go buy a new roof for your house. All advertisers do is see you own a house and send you some new roof ads. Your reaction is not "Oh I better go get a new roof from Steve's Roofs". Your reaction is 18 months from now you're searching for a new roof after a hailstorm, you see "Steve's Roofs" in a google search and click on it because you've seen the logo/tagline before. (Or maybe their is some latent brand awareness, you might associate Steve with trustworthy due to an ad but not realize it, etc).
With the AC.. that is due to a lack of data. They can't always match all of your devices to you, maybe you cleared your cookies etc. They missed that you bought it, and so as they see it you're still a very valuable prospect. IF there was more/better data that annoyance would be solved.
The AC advertisements are in case you're someone who regularly buys AC units and installs them, I think. DIY landlords, general contractors, etc.
Advertising platforms sell that ad with the maximum bid at auction, given what targeting categories they can place you in. Value of serving an ad is roughly customer lifetime value (LTV) times conversion rate (maybe divide by about three, but that doesn't change the winning ad). You don't have to be in a key target demographic to get oddly specific ads, you just need to get lumped into a similar cohort that has enough targets that are valuable enough.
Remember those random television ads from law offices about mesothelioma? Same concept. Pursuing a civil suit was lucrative enough that the trickle of leads you'd get from blasting everyone with that message was good enough. Doesn't matter if it's relevant to <1% of the target audience - high LTV can make up for it.
Literally in marketing 101 (okay, 301) one of the first things you learn is that marketing is not necessarily about directly selling things, but about brand recognition and subconsciously instilling it, and implied quality, in the consumers mind. For instance Coca Cola spends an obscene amount on advertising each and every year, even though they have no significant competition. A big part of that is because it creates an image in your mind the next time you go to buy something, that Coca Cola is higher quality more desirable than alternatives.
So for instance for your AC unit, perhaps you see a Samsung and GE. Your decision between which to buy is the purpose of marketing and advertising. For instance this is also why companies don't really like running discounts. Discounts can produce immediate increases in sales, but they simultaneously can also subconsciously diminish the perceived quality of a product in the consumer mind. Places where discounts, across all brands, are ubiquitous and independent of the manufacturer are presumably a game changer here since it's not the product being discounted per se but the site itself running a discount.
Apple's 1984 ad is still probably one of the most famous ads. Check it out [1]. It doesn't really make you think, "Yes I must go immediately buy a Macintosh." Literally no details are given about the product at all, and 90% of the commercial is a cinematic that has nothing to do with the product. Nonetheless it instills a image, and one of perceived higher quality, in your head of the company and the product. So the goal of advertising is not necessarily to immediately make you buy a product, but to change your perception of a product.
---
This is about marketing in general and not digital market exclusively, but it's relevant since for instance the vast majority of people also don't think commercials are effective on them - yet companies spend billions of dollars running them. Somebody's wrong, and it's not the side that can directly quantify the cost:returns of the marketing.
And one other final point here. Even in 'naive' marketing, it's a numbers game. For instance did you ever buy viagra back in the day when viagra spam was everywhere? Probably not. In fact, almost nobody did. But if each of those impressions cost a company $0.001 and their profit margin on a single sell (let alone return purchases) was $15 then they only need to get 1 out of 15,000 people to buy the product for it to be a profitable endeavor.
> As a consumer, not an ad tech guy, I don’t get digital advertising at all. It rarely if ever drives behavior on my part, and I just don’t see effective use of it. The most useful ads I see are the circulars in the newspaper that I seek out.
That's very different from my experience as a consumer. Nearly all of the ads I see are digital - I mean I look at the stuff in my physical mailbox long enough to sort (I'd happily pay the post office $100/yr to just block the 'bulk rate' and 'resident' email. Maybe more.) but I don't open up and look in them, even though they are local and usually contain discounts, just 'cause they are rarely what I'm looking for. (where I am, it's mostly fast food and off-brand tools, as far as I can tell. I mean, I do spend a fair amount on tools, but I usually either want a fancy tool, or just the exact right tool. I'm not really willing to wait for the pozi-drive #3 screwdriver I need to go on sale.)
What sorts of things to you seek out in the circular? and why do you check your physical mailbox rather than using a search engine?
I'm asking out of interest; I used to run a business and I used to buy advertising. Once, I even bought space on the conveyor belt dividers at the grocery store, that realtors use. (I looked for pictures, and all the pictures I have of it are terrible. ) To sell Virtual Private Servers. It was good for a laugh, and there was a little bit of attention, I mean, it was a grocery store in silicon valley. There is something really gratifying to the ego, from the perspective of a small business owner, about buying local physical advertising. But I'm not really convinced of it's effectiveness.
It didn't really occur to me to use mail... just 'cause I personally hate physical spam so much.
>I’m also getting bombarded with air conditioner ads from Amazon because I bought an air conditioner, from Amazon, a few weeks ago. For all of this big data and machine learning investments, they are trying to sell a man sitting in a 60 degree room another AC — which given the price and free shipping they probably lost money on.
I get this a lot, too. I mean, amazon knows I just bought that thing, right? I bought it from them. why do they keep serving me the ad? I'm wondering if a lot of it might be third parties trying to arbitrage google ads vs amazon affiliate rates, and those people might be able to see I'm looking for a thing, but not that I bought the thing.
Why, from the mass market perspective (mp3, not vinyl), does traditional media need to go on at all? In the absolute limit of global ultra-draconian GDPR-style laws the capabilities of internet advertisements will at best be reduced to those of TV spots. If TV advertisements haven't already merged with TV over IP and the data market (and thereby become indistinguishable from online video advertisements), they'll just be replaced as ATT, Comcast, Amazon, Netflix, and the rest shift more and more to online platforms.
As far as I can see, unless advertisement gets regulated back in to the stone ages (not that I'd be personally opposed to that...), it's going to be down to Google, Google competitors, and highway billboards before long.
I realize I'm kind of missing your point, but in my hometown of Austin, TX lots of billboards are advertising billboards, a charity (probably for a tax writeoff), or else have had the same ad for so long (often something like a concert that has already happened) that it's clear the billboard company is not getting paid anymore. I think billboards are stuck in the same downward ad revenue spiral as a lot of other traditional media.
Charity billboards are likely PSAs that brands use to test the effectiveness of a billboard. You find two very similar towns, each with a walmart. Buy a billboard in each town, but in town A put up a walmart ad, in town B put up a charity/PSA billboard. Over simplifying, but if sales in town A increase more than town B then the billboard is effective.
I honestly don’t know. I’ve only read about this type of testing and a friend does it for direct mail at work but I’ve never used it. I think in digital or direct mail that you have the ability to track those users, which I guess doesn’t apply with billboards.
Weird, in my town it’s the opposite. Heavy rotations of ads on billboards, and a variety of topics from retail, politics, political influence and brand awareness.
Paying for a tv spot with an audited and reliable audience estimate seems to be less fraudulent than any digital ad spend I've heard of with all those phantom bot clicks and impressions. Also you have the advantage of knowing the context of your ad rather than being surprised to be displayed on "John Doe's Racist Blog (tm)"
But you could buy 10x the impressions on an Apple TV for the price you’d pay on traditional tv. So if fraud is less than 90% it’s probably a better deal to do connected tv.
Maybe but let's think it through. Why do you think it is 10x cheaper? (Is it, I don't have numbers, please link them?) Is the market really so inefficient?
Apple is looking to maximise its revenue and will put up prices as far as they can in accordance with that. ie discounts are to get more revenue now and/or in the future. Why is traditional tv able to charg 10x more?
The idea is ridiculous. The math men didn't overthrow the Mad Men. The Mad Men found that math is awesome, because it's a topic that your customers don't know. So they also have no way to figure out that the Mad Men don't know math either. They call the mathematicians "quants". Does that sound like respect to you? Is that how one speaks in public about one's overlord?
There's still way more money being invested in the traditional world of advertising than what's being invested in the analytics and data side. Creative agencies haven't gone away. Agencies of Record are still the traditional shops (DDB, Ogilvy, etc.)
The data side is still very new (<20 years old). I’m expecting that once the old agency guys and old client side guys die off you’ll see bigger changes.
They overthrew the medium not the ad agencies. Its just those agencies that didn't adopt the new medium struggled.
Most large companies I have worked with still use an agency to manage ad spend for a cut. And though the pitching has an analytics bent I cant imagine its that far off how they used to pitch TV & Billboards. And they are still run by sales people not machine learning grads.
Yeah, the sales pitch nowadays its “with our new technology we can do this” where I’d imagine previously it was “with our partnerships we can do this” or something similar.
The article doesn't really align with the title. I was hoping for some history on how math men came up or how their value compares to traditional mad men. Unfortunate just another hot topic hit piece on google/fb.
The sentence, "math people and mad men" isn't a play on words, and "math people and mad people" doesn't make sense at all. Nobody uses "math men" in speech, this was done for a unique reason.
I don't think MLK would be opposed to a more inclusive statement, and I don't think MLK would have downvoted someone for wanting women to be included in a population that is often presented to the world as largely male, even though it's really not.
We really don't know what his position on a play on words in a feature article in The New Yorker written in 2018 would be. He's been dead for 50 years, and I'd venture to say he'd be more concerned about poverty and criminal justice reform.
Maybe black and underclass social problems are not the same thing as making sure writing is bland yet as inclusive as possible in a publication that mostly rich left-leaning white people are going to read on their $1000 itoy?
I'm female, I work at a university that is considered in 'the ghetto' as a software developer. There are a lot of issues in my city that range from everything from homelessness, to gentrification, mental illness, addiction, etc.
I've seen the consequences of discrimination from everything from skin color (I'm white, many of my siblings are black or mixed), to gender, to socio-economic status, etc.
I've also spoken on the internet for a long time, and I know what it feels like to feel silenced with the push of a button. A downvote. It's annoying. It's annoying to know people studying machine learning are only doing it to control how people think, to control behavior, to control wealth, or whatever the fuck else they want.
So, how many times does this have to wrap around itself for people to understand, downvoting people on the internet for complaining about how the world of math and logic seems largely presented as a male world, is part of the problem?
Even more ironically, MLK gets quoted.
All men are created equal. Are you focusing on the word men in that rhetoric, or are you understanding the message?
The best steps you can do for your business are have a very firm commercial idea of what success looks like (i.e. measure LTV, churn etc by channel) and have a robust framework for market research. I personally like the idea of splitting spending marketing spend between branding (where your aim is to promote unprompted brand awareness) and acquisition (where you want CPA < LTV.) Also remember there are A LOT of businesses which spend money in things that an expert can identify as not commercially sound - so audits by external agencys can be useful (i.e. not using negative keywords in AdWords and spending many thousands on direct response campaigns that don't work.)
I'll add the cavaet that rhe worst thing about "attribution" I've found is it puts an unfair amount of pressure in the bottom of the funnel. Brand focused marketers aren't really held to scrutiny at all (lots of marketing teams in businesses I know don't even do total spend / total sales) - I think this can cause some resentment between teams.
Good articles on this are few and far between but this one I don't mind https://www.cmo.com.au/article/641243/iag-marketers-must-end... as well as Willem's talk on what they did to bring programmatic in-house at Foxtel.
Happy to connect with anyone interested in this space.