Upcoming IP Investor Dialogues: Spirax Sarco, FAST, COST, CPRT, POOL, SNOW, DDOG - Register your interest at help@inpractise.com 

Anonymous

What key structural changes have you seen over the last decade or two in marketing?

The first thing to say is that the heads of everyone in marketing are exploding, and that's okay. Marketing is going through a massive transformation. All the old rules that were taught in business school and taught at places like Procter and Gamble are on the challenge. Some are still relevant; some are less relevant, but there's a whole load of new tricks to be learned. Every CMO in the world is being pitched multiple times a day or a week on new technology or data that they should be using; otherwise, they're going to become obsolete. Most CEOs are unhappy with marketing because they know it could be doing more, but they do not quite know what it should be. Marketing is in massive flux. And yes, it is in the process of being reinvented. I do not have all the answers, but I will have a go at some of them.

To me, the fundamental change in marketing is the move from mass marketing to people-based marketing. What does that mean? That means in the past, for anyone who watched Mad Men or anything to do with advertising in the 60s, 70s, 80s, 90s or people who worked in agencies a couple of decades ago, it was all about the big idea. Thus, the brands would need to be defined maybe every year or a couple of times a year. That's a big piece of work; that'd be a big campaign. The CMO would brief a few agencies on what they want to do, “It's about this product. This is our brand strategy; what should the campaign be?”. The agencies would go away, have lots of brainstorming, storyboard a few ideas, and then present back to the CMO. The CMO, based on mostly judgements and maybe a few bits of panel research, would pick one of the Agencies, or one of the ideas have to be turned in production. The production would be a very fancy affair, going off to Hawaii with some celebrities and champagne and lobster and basically spending millions on producing one TV ad, then a bunch of additional ads would be made for outdoor, and print and what have you, and then the campaign would be run. And it was one message to everyone. There were some subtle variants on that. But broadly speaking, that's what marketing was. It was about getting an idea as brilliant you could and as well-executed as you could. And that message went to everyone. ROI on most marketing was relatively unknown. And that was accepted. But you could see that if you run a big campaign, sales went up. If you run a bad campaign or no campaign, sales went down. So, it was something companies had to do.

If you fast forward to the future, you could argue that you could be serving a different ad to every single person. Now that's technically complex. Most companies are not quite there, but they're on a journey in that direction. So, what does that mean and why? The reality is that every single person is different. And every single person's relationship with a brand or a company is different. They have different past purchases, potential purchases, price sensitivity, the timing of purchases, and media consumption. Regarding where they might see ads: some people are on Facebook, some on Instagram, some watch TV, some read print newspapers, some see outdoor ads, etc. They have different experiences with competitive brands, some are loyal to another brand, and you need to win them over. Some are already loyal to your brand. You just need to keep them and a whole lot of switches in between. They will have different potential lifetime value.

For every customer, you need to be able to say, “If I got this right, what kind of relationship will I have this customer? What kind of products and services will they buy or consume? And what's that worth to me? And what's the right path to take them down in terms of interactions? And how much will that path of interactions cost? And when you add up all the revenue and profit coming from that customer and subtract the cost of the marketing, what's the overall customer lifetime value?”

Now to do that by person with potentially a different ad served to every single person is a massive undertaking. I can safely say there is not a single company that's there, but there are many on a journey in that direction. So if you talk about the transition: step number one for any company is, organize your data about customers. When I say customers, I don’t just mean existing customers, but lapsed customers and potential customers. You gather all the data from your own databases and gather extra data that maybe you did not cover in the past. Who's been to your website? Who's in your loyalty scheme? Who's used your app? Who's been in your store? What aisle did they walk down? If on the website, what did they put in the basket? Did they call customer service? And so on. You then enhance it with useful third-party data. Third-party data may include household income, marital status, education level, number of kids, what car they drive, what they buy in the supermarket, etc.

Using a combination of your own data and that third-party data, most companies are already starting to organize that data well, and create more segments with more tailored campaigns. So the most sophisticated advertiser that I am aware of, they're already targeting thousands or tens of thousands of people with different ads per group, based on those factors that make people different, instead of targeting ads at tens of millions of people with the same ad.

Are there any regulatory barriers to this shift to personalized marketing?

There are all sorts of regulations. They're not barriers. They’re things to work within so that you do not break the law or spook consumers by bad behavior. The most famous of data regulation is GDPR, which is a European law on data. I think it's actually a very sensible law, which basically says that any company if they are gathering data from consumers, needs to (a) tell them they are gathering it, (b) let them have access to see what you stored about them, (c) tell them how you use it, (d) give them rights to tell you, yes, you can or cannot use it that way, and (e) give them the right to have it deleted if they do not want you to keep it. And most companies if not all companies are implementing those laws, because (a) they have to if they want to operate in Europe, but (b) it's kind of sensible.

That's how you maintain the trust of your customers. And so GDPR in many ways has become a law for the world because most big social media, or eCommerce companies are global or at least multinational. And if they operate anywhere in Europe, they will probably implement GDPR to their entire tech stack. That's the most influential law. My take on the laws elsewhere in the world is they either do not exist, or they're being created very similar to GDPR. For anyone who wants to understand the law and keep that knowledge management knowledge manageable, I would say learn about GDPR. But I would say that there are those five points I mentioned earlier and they do not prevent people-based marketing or data-driven marketing.

How do you explain the recent trends seen in TV ad spend versus digital? Digital has been growing over the last 5-7 years but TV spend has held up. I think people have been looking at views and spend, and that seems to be uncorrelated today where previously you would expect them to be following each other. How do you look at the correlation between TV spend and viewership and why has TV spend held up so much?

There are a few things I need to pull out of your question because (a) It is a complex question and (b) the answers are even more complex. It's not a simple black and white. The world is not quite as simple as digital versus TV, because you do have digital TV.

For example, Hulu or Roku are ads targeted in a digital way but affectively like traditional TV or virtually traditional ads. You have things like YouTube TV now that’s serving ads. But in essence, for YouTube customers who pay a subscription is $30 to $40 a month in the US. There are shades of grey. That's one of the ways TV companies have held off is by trying to expand audiences through digital TV and keeping up the ad revenue there.

The other thing I would say is that TV ad revenue is now under pressure in 2019. And in the US, the pressure started in 2018, as it is in decline, but, surprisingly, it did not happen three to five years earlier. In essence, for a long time, advertisers would realize they should be moving money from TV to digital, but digital looked very fragmented, and had a lot of brand risk associated with it. If you wanted to do one big, bold thing TV was still the biggest thing that you could do. On top of that, a lot of agencies who provided advice to CMOs and marketing teams would be encouraging TV because (a) they understood it better and (b) they made more money by doing mass-market campaigns on traditional TV. There was a mixture of inertia based on old school business models, and some pragmatic realities, it’s difficult to spend as much money with the same impact on digital.

Over the last few years, though, Google and Facebook in particular, but other digital properties have got bigger and bigger while the TV audience has been declining. Therefore, the argument that ‘the Internet is too fragmented, and TV is still really big’ is getting worse and worse. Further, YouTube, in particular, has done a much better job in the last year or two years of cleaning up its act when it comes to brand friendliness. So the last year they have deleted an enormous amount of unsavory content. Also, for content that might be questionable for brands but does not breach YouTube's quality guidelines, they have just stopped serving ads. By doing that, YouTube is now a much more acceptable platform to brands.

What that's done in the last few years, with Google, Facebook and others getting so big, advertisers are now moving that money over. That does not mean TV will die. But if you look at the ROI of TV advertising versus digital video advertising, in general, TV is a lot worse. Why is that? The CPMs, the price you pay, is typically higher. Then there's a whole load of waste in the system. For example, when you buy an audience on TV, let's say the Game of Thrones audience and the audience is 10 million people, that's 10 million people who watch Game of Thrones not 10 million people who watch the ads within Game of Thrones. This is because during the ads break, people are channel surfing, going to the toilet, grabbing a beer, going on Facebook, or whatever they're doing. Not many people are watching yet. So, my best guess is only about 30% of people are watching the ads, while the rest of are doing something else. So whatever CPM you're paying, you can triple it for the people who actually watch the ads, which makes inefficiencies even more glaring.

Also, what makes it wasteful is that TV advertising is mostly still targeted based on people who watch a program, not people's demographics, spending power, intent, etc. So if you're a niche brand, let's say you're BMW, and you're serving an ad on Game of Thrones. Let's say 10 million people watch Game of Thrones in reality; that means only 3 million people see the ad. However, what percentage of 3 million people are actually going to buy a BMW or a luxury car in the next three years. The answer is not many. Some of them will be too young, too old, do not have a driving license, or do not make decisions about buying a car because they have a company car. Also, many cannot afford a BMW. When you actually narrow it down, it's probably 10 or 20% of the audience that actually can afford a BMW and might buy a BMW. So, you're down from 10 million people to 300,000 people. Yet you're paying to serve an ad to all the rest. So the inefficiencies on TV become glaring. Now, if I were running a TV company, I would radically fix that and let advertisers target people based on how Facebook and Google allow targeting rather than who watches what program. They still have a structural problem that the ad sets are too expensive even after all of that. Nonetheless, it would be a good way to mitigate the risk.

Is that even possible for broadcasters to gather that data and get that targeting capability?

It should be possible. Yet it depends on the country as to the structure, but there are a lot of things blocking it such as, who does the selling of the ad inventory, who gets the right of refusal, the ability and availability of ad tech stacks to make it happen, and how things get sold. A lot of stuff gets sold in like upfronts very early in the year, and it's sold against programs, not demographics. There is, therefore, a lot of barriers. But if you think about it, if you take YouTube TV as an example, YouTube knows who's watching which program and they can change the ads based on who's watching which program. It does not always happen. When I was living in the US, I would be watching YouTube TV, and then I would watch MSNBC, and I would see ads for old people drugs, which I do not need because the ad was sold by MSNBC. In the future, MSNBC should probably give the ad sales to YouTube and say, "You can serve more relevant ads than I can, why don’t you do that?" I think for the TV companies, they either need to outsource ad targeting to Google, Facebook or similar companies or get their act together and sell to people based on demographics, purchase intent, etc. Rather than who watches what program.

Does this mean that with this shift to personalized marketing, these media-buying platforms will be auction-based?

Yes, I believe so. Number one is if you have an auction-based platform, you can ultimately make more money. That might sound strange when I have just said TV is expensive and at risk of a price implosion. But in reality, if you think about the value in ad TV serving to someone who will never buy a product, you've just wasted your money. If you serve an ad to someone who can afford your products, is open to your brand, and is likely to buy in the near future, you've just hit the nail on the head. If you can, as a publisher, get 100% of ads to be well-targeted, you will generate the highest CPMs because, ultimately, the advertisers will be prepared to pay more. Therefore, all platforms will or at least should be auction-based, and the ones that are not will end up being discounted in terms of pricing because of all the waste.

How do you think advertisers will change their team composition or organizational structure? Will they bring media planning or buying in-house?

It’s tricky. The question on almost every CMO’s mind is, what should I do in-house versus what should I outsource? I think at a generic level, the stuff that is very strategic and needed day-in day-out and potentially sensitive should be in house. The one-off activities, the ones requiring specialist skills that you do not think you can build should be outsourced. Your brand strategy should be in-house, but maybe when you want to redefine your brand, you use an agency to help you do that. Your data strategy and data management should be in-house. You might license software. That’s fine, but in terms of working out where all the data sources are, how you gather them, how you organize them, that should be in-house.

Do you mean the CRM or marketing services?

Exactly. The customer database is the brain of marketing in the future. That cannot be outsourced. You can get consultants to help you with your strategy, but ultimately, employees of the company will and should own management of the data.

In terms of creative, it depends on what you want to do. There's still room for some big, bold breakthrough ads that cost a million dollars. You will probably still use agencies for those. There's also a lot of day to day activity, making thousands of different ads for all those different customer segments. That could be done from an agency if it's very well plugged in, probably embedded in your organization. Or it could be done in house. You could also license software and do it in house, but that needs to be so embedded in the marketing routines that it cannot be an arm's length relationship.

Regarding media buying, different companies have different strategies. Most companies these days run things like search media buying in-house, and generally outsource most other stuff. Obviously, various performance advertisers will do more in house. The CPG brands, government, and banking often outsource an awful lot more. There is a trend towards in-house, but it does not mean that's the end state. There's a trend towards it because most companies realize that they outsource too much and they were maybe not getting the best advice from agencies who are a bit slow to pick up on Google and Facebook, data, and other things.

Still, it's more a case of experimentation. Another thing that should be kept in-house is measuring ROI. In order to hold everything accountable, whether it's your in-house teams or your agencies, you need to make sure that you do all the measuring yourself. If you say to an agency was my campaign good, they're always going to say “yes”. You want someone in-house to be measuring that. So, I think it's around the data, planning, and judging in-house. Day to day activities that are strategically important will depend on whether you think it's more efficient in-house or outsourced and whether you can build the skills in-house.

Regarding the CRM point, agencies over the last decade evolved and bought businesses more on the marketing services side to diversify and offer new services. What is the value-add of those businesses for the agencies, for the end customer in terms of CRM, PR, communications, marketing services segmentation of their business?

I know that's a very broad canvas and most of the agency's holding companies have acquired thousands of companies in the last few years and incorporated them to different levels. What I would say, I have a lot of sea level friends, agencies and technology companies, CMO, CEOs, etc. Most of the agencies would say “we are still nowhere near getting it right when it comes to really helping clients build an awesome CRM database segment and do segmented tailored personal marketing.” The way that agencies’ incentives are organized, and how their skill sets are organized is still much more around the big idea. Most technology companies, such as Accenture, Deloitte, and PwC, focus much more on selling big tech solutions without really understanding how marketing works or should work. So they're not awesome either.

I have not yet seen a company, like a third party company whether tech companies or agencies, that's really nailed an approach to helping guide customers and companies through strategy, data strategy, gathering the data, organizing the data, managing the data, and then turning that into activities of communication with ever smaller customer segments. That would be awesome. It's also a massive undertaking. I have thought over the last years that maybe one of the big tech companies will buy one of the agencies and somehow create some magic with a combined skill set. There's a risk; it could cause a two-headed beast and can’t work as well. But I think that would be the dream agency of the future, or just the alternative is to be good at one niche.

I think that CRM side of the business seems to be obviously growing much faster, or at least faster positive growth versus the other side of the business?

Yes, clearly, it's a growth area. If you think about the value of an amazing CRM database versus no database, you can at least double but more likely triple or quadruple the ROI of your marketing by having an amazing customer database. Simplistically, if you were to apply a pure economic model to that, companies should be or could spend three times more on gathering and organizing data than they do on media. Now they will not do that because they probably do not need to, but that's how valuable it is. Companies are realizing that more and more.

It's not just about what's good practice versus bad practice. Companies like Google and Facebook and increasingly all the others are auction-based, which means CPMs are going up year after year, and advertisers that do not do really good, data-driven, people-based marketing will end up being priced out of the market. Because if you're doing mass marketing, while prices are going up dramatically, then the ROI on your spend is getting worse every year at a pretty fast rate. There are no alternatives. You have to do better data-driven marketing. When you just see the ROI, you can see that you should be spending more and more on data and CRM and even if that money gets pulled out of media or creative.

You're saying that because of the auction framework of Google and Facebook, that's going to drive up CPMs naturally, so if the agencies or advertisers do not segment their customer base adequately, they will just be priced out of the market completely?

Completely, yes. I think talking to customers, as in serving ads to customers, is a key part of your survival because you need it to acquire new customers to offset the lost customers or whatever it might be. To build your brand, you have to be professional at people-based marketing; otherwise, you will not go for the CPMs that are going to come. My best guess is that CPMs on the big platforms in the next decade will triple. People think that pricing is expensive now have not seen anything compared to the future.

Because you get better segmentation and better targeting?

It's better for everything. If you look at what the platforms are trying to do, Facebook is setting the standard, but the others are trying to keep up. They're doing four things better and better every year. The first is around targeting: giving advertisers more and more ways to target. Best practice at this point in time would include (a) allowing companies to pull a segment out of their CRM database. It could be lapsed customers or potential customers. Whatever it might be, you pull a segment and take their email addresses or mobile phone numbers, upload them into Facebook, and target them with a certain ad or communication. That's called custom audience targeting. The second way of targeting is pretty cool. It is called look-alike targeting where you upload a segment from your database to the platform; in this case, Facebook works out what these people have in common and serve ads to people like them.

That allows you to expand your audience. The third way is obviously demographic targeting. So, using platform data, third-party, or whatever, and then targeting people based on age, sex, marital status, number of kids, what car they drive, or where they go on vacation. Every platform is going to get better with that. That's the one important lever. The second important lever is ad formats: having ever more ad formats that meet marketer's goals. Again, I will use Facebook as an example just because they're doing the best job, but you can break ad formats into two buckets: brand building and performance. Brand building would be things like display ads, video ads, and other forms of engagement ads where the goal is not really to get people to do anything tangible other than engaging with your brand and learn about your brand, etc.

How attractive is that today? I think previously over the last few years, the big brands have been hesitant about putting dollars on Facebook and Google, because of some of the content they have. Have they changed their minds about this?

I think they need to get over themselves. There were some valid concerns two years ago. Obviously, Facebook had some bad press around, fake news, etc. But as an app from an advertiser's perspective, your ad is not associated with any content. It sits between content. So by advertising on Facebook, Twitter, Snapchat or whatever, it’s not like you're sponsoring the content that people see on that site. Therefore, there should be no concerns over them. For YouTube, it's slightly different because YouTube's primary form of advertising is pre-roll ads that appear before content. It appears as if the brand is sponsoring the content, you're about to watch next. So, there is much more association. I understand the sensitivities of the past, but I think advertisers need to get over themselves. First, YouTube is just too big to be ignored.

It's bigger than pretty much any other TV station and hits a certain demographic massively like people in their twenties, teens, et cetera. Second, they have done a great job of cleaning out the vast majority of unsavory content. Third, for the content that's questionable, they will not serve ads. Thus, your brand will not appear next to it. Finally, there is a suite of tools that you can use to set your own brand sensitivity. A combination of those things means there should be no concerns, it just requires a sensible approach so that your brand appears where you want it to appear.

So these platforms clearly offer great ad formats. For example, you’ve mentioned that Facebook offers brand display and video for brand advertisement and performance advertising. Clearly, they're the best at targeting.

Absolutely. For performance advertising, there are so many ways to be creative. To pick out a few examples, a mobile app install is obviously a big one, or click to a website or click to conversion is a way to get people to visit your website and maybe buy something or take an action of some sort. There are dynamic product ads that are like dynamic retargeting. So, based on whatever you are doing on a company’s site, they can target you with an upsell, cross-sell or whatever it might be. But there's also book-an-appointment ads; there are store visit ads. What Facebook's done well is they look at all the marketing goals that exist and they try and have an ad format for each one of them to help marketers get the results they want. So, it can go in a lot of ways. That's bucket number two. So, going back to the buckets of what drives performance for a platform and what drives CPM growth; number one is targeting, and number two is ad formats.

Number three obviously is the audience. You keep finding ways to grow audiences and clearly to date, Google, Facebook, Twitter, and Snapchat have done a great job of growing their audience and at the expense of older media like print or TV.

Finally, number four is challenging and training the ecosystem around them to do a better job every year. It involves partnering with companies who have useful plugin technology, challenging or training agencies do a better job, and telling your advertisers what's possible. So, they either do it themselves or challenge their agency. It involves showing what's possible with data, which then encourages companies to do a better job of organizing their data. However, that alone also drives masses of improvements in marketing, and thus ROI, which drive up CPMs.

Every time an advertiser can organize their data better, or segment it better, they improve ROI, which increases CPMs. Every time they do a better job of tailoring creative it improves ROI, which increases CPMs. There are so many levers. Facebook is now a 12+ year-old company. While Google is 20 years old, Facebook’s CPM growth is still about 20%. And it's probably going to be in the twenties for the next decade. So, from that perspective, you just extrapolate. CPM growth in the twenties for ten years in a row and you're going to have pricing something like tripling in a decade.

Do you think CMOs, more generally, are aware of this, in terms of getting priced out of the market?

Not so precisely. There's a wide range of CMOs, but increasingly, the mass marketing dinosaurs are being fired and replaced with more data-driven digital natives. I would say most CMOs, newly appointed or digitally savvy, which is increasingly the majority, are well aware that data is really useful. They are well aware that you cannot avoid Google or Facebook.

Also, all the big digital platforms are well aware that investments in technology are as important as picking which TV ad to serve or whatever it might be. They're well aware of the big important buckets. I think the subtlety is around CPM growth and the levers: what is driving it? And the fact that it's an existential issue. The way I speak to CMOs is, "If you cannot increase the ROI per ad served by more than the price of CPM inflation every year forever, then one day you will not be able to afford to talk to consumers." Now, that might sound complex, but it's a relatively straightforward statement. I think that particular point is probably still missed by most CMOs.

Let's say that you're the CMO of P&G. I was reading the JP Morgan report in 2017 when P&G cut their agencies from 6,000 to 2,500, which even then is a huge amount of partners. How would you look to structure such a huge business like that? Is there any need for that many agencies?

The first thing I would say is I have to assume that there were 2,500 agencies on a global level. There's no way they would be 2,500 in the US or the UK typically marketing is still done at the national level. Maybe brand identity is done on the global level, but most marketing is still done at the country level. For example, 2500 on a global level translates into 50 at a local level. That's probably still too many because the reality is most agencies do compete with each other because they have overlapping services. They do not know each other very well. They are not very slick at working with each other. There are a ton of activities that need to all work with each other in a seamless way and be iterated incredibly quickly. That second part is critical. If you've got an internal data science team, combined with three or four or five media buying agencies, two or three creative agencies, a separate PR agency, a different benchmarking agency, and maybe a customer insight panel agency. There are technology companies who are pushing their services as well as licensing their technology. Trying to get that lot to play nicely and efficiently and iterate campaigns in a real-time way, is just impossible.

I think it will evolve. In the past, and for some companies now, there is what's known as an agency of record who would be the big agency who coordinate the other agencies. I think that does not work anymore. For two reasons. First, in the past, the agency of record would generally want to work with agencies from their own holding company, hide profits in places with transfer pricing, and give bad advice because they gave advice in which they made the most money rather than advice that was best for the advertiser. It's a business model that risks corruption and incompetence. Second, having a client brief, an agency who briefs other agencies and coordinates other agencies, means the client is too remote and the ability to iterate fast and to keep improving is too slow.

Therefore, I think companies need to become their own agency of record. Thus, obviously have some in-house teams and some agencies, but ultimately, the client is the one doing the coordination. It's relatively straightforward to do because there are really talented people at agencies of record who will be redundant soon. So you can hire some of them and say “what you used to do as an agency of record, you are now in the marketing team, do the coordination from the inside rather than the outside.”

Can you highlight the bit on the transfer pricing and run an example of a 10 million campaign on an agency of record versus a small digital agency, for example.

An agency might either own a media buying platform or a demand-side platform or a data exchange or maybe have a partnership deal. If they take client’s money through that platform or through that company, they will get a kickback. For example, in Mexico, their companies that will say, “last year, Mr Agency, you spent 10 million. This year you spent 12 million. I'll give you a million kickback money back under the table.” It's typically not illegal, but obviously it's not moral. It's like a volume discount. Then the agency will want to make sure they meet that goal. So they will pump their client's money through the TV channel, even though it's not in their interest. Now that money, the volume discount, will not appear in the client's P&L and it might not even appear in the margins of the media buying company. It might appear in obscure transfer pricing because maybe the kickback goes to a different company within the holding company portfolio.

Or, alternatively, these days you might get the agencies who own a stake in an ad tech company or a data exchange or demand-side platform. They use that platform to manage their clients’ money, and then effectively make money because the platform they're using is also making money, or going up in value or whatever it might be. So there's a whole load of ways in which obscure agency holding companies with lots of different companies and investments and some really bad practices will hide money from their clients, which again, is a reason for clients to become their own agency of record. And that way, if they're managing agencies on a direct basis rather than an indirect basis, there's a lot less room for funny business.

Can you elaborate on the split between work and non-working dollars? So I am an advertiser. And I have got a $1 million campaign. How much actually goes to the buying of the media versus the non-working?

It varies. There is the normal rule of thumb, and some CMOs have different rules. If you've got a campaign budget, let's say a million dollars, you should probably limit 20% of it to creative and 80% for media. That's kind of a tried and tested number. I hate to say tried and tested because I like to say everything should be changed, but ultimately, it's a number that's hard to disagree with. You will see examples of campaigns that had such amazing creatives. They went viral, great, but they are very few and far between. And you will also see examples of very cheap creative facilities. There are examples of all extremes.

The problem is relying on something to go viral is a very high-risk strategy. If you said, “you know what, for Christmas, our Christmas campaign, we're just going to do an amazing job with creativity and hope it goes viral.” What if it doesn't go viral? likewise, saying, "we're going to make cheap creative and just get your ideas and hope for the best," might be great if you are a 24/7 eCommerce company selling thousands of products all year round. You can keep it with lots of creativity. But if you're a company that wants to have a big, bold, successful product launch or whatever it might be, it's pretty risky too. I think the combination of appropriate risk ratios and risk management typically suggests about 20% of the campaign budgets we spend on creativity. I would say most advertisers would be wise to use that as a starting point, and then feel free to iterate from there.

But don't the agencies still have scaling and media buying that can provide some kind of competitive advantage going forward?

They would like to talk about it, but it's pretty non-existent. First of all, whenever there is an auction-based platform, agencies do not get any kind of bulk discount by buying a lot. For Google, Facebook, Twitter and others, there's virtually no rebate. Agencies might get some rebates if they buy the preferred inventory, but in essence, that's a scam because that's been pumped up in price by the platforms, in order to give the kickback to agencies, so it means the ROI has got worse. Over time that preferred inventory will mostly be done away with by the platforms and in a world of everything being auction-based. Then agencies say, "Because we buy in bulk, we get a discount," is irrelevant.

The main way for any agency to offer a better ROI to their client is through a better approach to media buying, targeting, creative, optimizing, all of those, doing better tailoring, basically having a better suite of technology and data and activities that improve ROI day in, day out throughout.

Advertisers can go directly to Google and Facebook because they have got auction platforms, and you can be direct, and you can target great. But those businesses are not really set up for marketing businesses. Do you think there's always going to be an intermediary between the advertiser and the media outlet, whether it's Google or Facebook?

I am not a Doom monger for agencies, saying that there'll be no agencies or the big media buying agencies will implode. I think there's always going to be a case, a good economic case for agencies. Obviously, they need to change a lot of their practices. I will give you examples of when outsourcing makes sense. Let's say, for example, a really good media buying person, who's really world-class at running Facebook campaigns or Google search campaigns, whatever, probably cost you fully loaded, at least $10,000 a month.

If you're a large, diverse advertiser, you need to be doing well across campaigns on Google search, Amazon, Facebook, Instagram, Twitter, Snapchat, Pinterest, maybe LinkedIn, plus some display advertising, et cetera. You probably need at least five or six people, that's $50-60,000 a month for your headcount.

Now, first of all, that's a lot of costs. Now, if you're going to justify their costs, let's say your media spend is only 60,000 a month. Do you really want half of your cost to be wages and a half to be media? Probably not. So you probably have to outsource to an agency. If your media spend is a million a month, you have 60,000 of costs might make more sense because it's six per cent of your media spend. There's an element of what's your media spend versus what's your headcount costs and the ratio of those matters.

The second thing is, let's say you've got a world-class Facebook person and they quit. Do you stop advertising on Facebook for three months? No, you do not. But in essence, there's a bit of a risk there. Whereas if you outsource to an agency and one of their people who's running your campaign quits, they can backfill probably same day or next day, or whatever it might be.

There's an element of risk management too. For example, when I was running AdParlor, which was mostly in North America, we had clients in places like Columbus, Ohio. And they said, "there's no way we want to build an additional team because (a) the talent probably is not here (b) if it is, we'll get it and a few months later off the train, they will quit and go to go somewhere else."

Just having the continuity of talent was just too hard. I think there's both an economic argument for agencies, what's the cost of the people versus the cost of media, and then there's a talent and continuity perspective that makes sense. And then what the agencies would say, which I would also agree with this, when lots of Facebook buyers sit together all running campaigns for lots of different clients, they learn from each other around best practices and new ad formats, et cetera as well.

If a Facebook media buyer sits in a company on their own, then they do not have anyone to learn from. That does not mean I am all for agencies. I am all for appropriate balance. For a company that spends 10 million a month, they probably would go almost everything in-house. And maybe now I am going to get an agency in to audit what they do and challenge them and give them new ideas. When your spend starts getting smaller, then you probably outsource more.

There still will be an intermediary. The value chain will remain, but you're going to see a huge fragmentation in those agencies, which we’re starting to see already.

What would be a concluding remark on your overall view of the shift of personalized marketing?

I probably would come back to the whole issue of this being an existential issue. For companies that do not do it and do it faster than their peers or their competitors, there is a serious risk that at some stage CPM inflation will mean that they cannot afford to talk to consumers. This is not just about, "Let's implement best practices because it's good." It's, "Let's implement best practices because otherwise at some stage you're out of business."

Former COO of Google, Europe

Facebook: Personalized Marketing

What are the potential reasons TV spend has held up as viewership has declined?

What are the challenges for TV advertising to compete with the ROI from digital advertising?

How would you organize a marketing department of a large CPG company?

As Head of Marketing at a large CPG brand, how would you split your advertising budget between the various channels?