Can automated media planning match instinctive expertise?

This week I saw a tweet from @leorayman asking if Pointlogic client’s can comment on Chorus and if automated media planning can match instinctive expertise: http://twitter.com/leorayman/status/10728837860036608

I never compare Chorus with an expert; I compare an expert using Chorus with him not using Chorus. But let me use this blog to do both and work towards my point of view on the question raised in Leo’s tweet.

Chorus vs. Expert

Communication planning today is incredibly complex and needs to balance data and insights in one hand with creativity and ideas in the other. The communication planner is literally juggling  reach, cost, consumer insights, brand objectives, creative plans, media properties, quality of reach, engagement, historical knowledge from ROI modeling, competitive activity and so on – all factual information available through various data sources. At the same time the communication planner tries to include more creative thinking in how to optimize the chances for viral campaigns, to be noticed; to be talked about; to be different. I believe it is clear that automated planning will beat the expert when it comes to assuring that all factual information is accounted for while the expert will beat automated planning for being effectively different.

So who will come with a better media plan? Many clients opt for planning to be data driven and fact based and Chorus would better satisfy these clients. However, I believe that more clients should show strength in doing things differently and rely on human expertise to guide them through ideas that could potentially stand out.

There is also a practical issue. Most campaigns I know (and for which I have seen the planning with no automated optimization) are actual not that different. It’s mind blowing how much of communication planning is based on “copy the competitors and tweak through ideas & insights”. Chorus provides a zero-based optimization where competitive activity is an input but not a starting point. As such, practice shows that plans delivered by Chorus often show more differentiation based on brand objectives than planner’s output. To set the score: Chorus 1 – 1 Expert .

Expert & Chorus vs. Expert

Now look at the situation where we compare two experts, one using Chorus and one not. The obvious reason why the expert that uses Chorus has a better chance of providing optimal campaigns is that an advanced user understands how to benefit from the fact-based planning results from Chorus and merge it with the more qualitative ideas from himself and other experts; the best of both worlds.

Also there are two key reasons why the optimization feature in Chorus helps creativity (instead of just being perceived as a black box delivering a solution). Firstly, creative ideas can be entered into Chorus (with assumptions on performance) and as such they can be better supported which leads to more clients relying on such creativity. Secondly, a more important and subtle reason, is that I strongly believe creativity and innovation start with insights. The optimization feature – the automated planning – and the possibility to compare plans deliver a whole new area of insights currently not available to planners. To give some examples:

• How does communication plans change if my objective changes from increasing awareness to increasing loyalty? These analyses show which channels perform across the board and which are more niche in nature.
• What happens to the brand results if I remove television altogether? These analyses show which brands and categories can make a major shift towards digital.
• How would media plans change assuming different roles for the media? These analyses show the ability to focus certain channels on one task (perhaps leading to the ability to plan using cheaper formats like half or quarter page ads).
• At what price point will newspapers get 10% of the share and which media will then see spend decrease? This is very useful for negotiations both to know what happens with the 10% price reduction as well as which media is specifically competing with newspapers.

I could give many more examples where Chorus delivers insights that can drive innovations in planning however I’ll close with the updated score: Expert & Chorus 2 – 1 Expert.

Thanks,

Peter

The meaning of synergy

As you know, Pointlogic provides a portfolio of systems and methods to help clients allocate budgets across a number of media. Yesterday I was demonstrating one of our tools and, as in many such demos; I was asked how we deal with synergy. As this is such a significant area, I decided to write my thoughts on this topic.

I like to think of advertising and marketing communication as creating value for the brand with individuals. If John sees an ad on television then something is wired in his brain that leads to value creation, let’s say the value created is $1. If Jack sees an ad outdoors then again there is some value created, let’s say this value is .75 cents. The question behind synergy is what is the value added when Joe that sees the ad on both television as well as outdoor. If the value is $2 then there is a positive synergy, if the value is $1.75 then there is no synergy and with less value there is a negative synergy (or cannibalization).

Understanding synergy is crucial if the value created is only measured in sales (as in most marketing mix models). These models could show sales uplift that we can contribute towards TV, an uplift that we can contribute towards outdoor and an additional uplift that we can contribute towards the combination – the synergetic effect.

So far so good. However, in our research methods like Compose or M3 we look at synergy slightly different. This is because sales are not perceived as a result of advertising but instead as a result of perception changes. Think of it this way, perhaps someone is much more likely to buy a certain mobile provider if they (a) like the brand, (b) understand the pricing, and (c) believes the carrier provides excellent client services. If the TV ad delivers on likability of the brand and shows the pricing and promotions while the outdoor ad delivers a positive perception towards the client services than we’ll see a positive synergy of TV and outdoor towards sales (even if there is no synergy towards the underlying three attributes). Simply put, there is no need for synergy between the media TV and outdoor, instead there is a synergy between the underlying drivers of sales. If someone likes the brand and understands the price the value creation could be $1 and if someone believes the carrier provides excellent client services the value creation is .75 cents. However, if you’re successful on all three drivers the value creation could be $2 (independent on the media that delivered the perception).

I do believe there is synergy between media, for example, seeing a brand across multiple sources could help the perception of “big” (they’re everywhere so they must be big) and other attributes. But this kind of synergy is dwarfed by the much larger effect of different media having a different role to play.

For Pointlogic this means that we strongly endorse looking at synergy based on actual creative executions – synergy is so much more about the actual creative than it is about some mysterious effect of multiple media. For sales models we have to translate media executions to sales and then we include synergy. If Pointlogic is a part of the research design – as in Compose and M3 – we normally recommend including underlying attributes that lead to sales so that synergy can be looked at from the perspective of the roles of the different media.

Yesterday I compared synergy with the situation of a football (i.e. soccer) team. To optimize the chances of winning games you need defenders and strikers. Is this a synergetic effect between defenders and strikers? Probably not, it’s simply that you need both roles to win games!

Thanks,

Peter

How tablets will change businesses

About six weeks ago I purchased an iPad with no idea what I would do with it. Now, six weeks later, I use this device more than I have ever used my laptop. I use it to take notes (I’m writing this on my iPad right now), to send and read emails, to read a Dutch newspaper for which I have a subscription, to play chess, to read business papers, to browse the web, and to share pictures of my two daughters with friends and family. It has truly become a device that merges my personal needs for entertainment with my business needs for communicating and staying informed.

So is the iPad just like a computer but more convenient because of its size? Or is it more like a mobile device catered for more serious work than smart phones? I would say no to both. The core reason I use the iPad more than I could imagine is because it functions as an interactive piece of paper. For me, that insight is the way forward for companies successfully implementing the iPad as a business tool.

Senior executives at companies spend most of their time in meetings or reading. Different departments provide them with financial statements, marketing plans, HR summaries, market information, sales figures, competitive activity, and so on. These end up on someone’s desk as papers that he needs to read (or quickly scan) and those papers lead to the strategic decisions he or she needs to make for the business. The senior executive doesn’t use a laptop or web-based dashboards for his information.

And this is where the iPad, and its competitors, come in as I believe the senior executives will use these devices once the information is available. It can simply start by delivering the documents as a PDF loaded onto these devices but soon we’ll see use of the interactivity to give executives better information, faster and easier.

This provides a huge opportunity for providers of, outsourced or in-house, business analytics and intelligence. These providers will get their chance to interact directly with senior executives in ways that was impossible before.

Thanks,

Peter

What’s in a word? Everything!

Yesterday I had a debate with someone from Google about search which was somewhat impractical, but perhaps many debates are. The debate was about: Is search a communication channel?

His argument was yes because search delivers an audience. True, and I agree it’s a very valuable and targeted audience. Marketing is at its best if it gets consumers into the inquiry stage and search is the logical medium to allow consumers to inquire.

My argument was no because it’s not about search but about find. If search marketing was called find marketing, everyone’s mindset would be different. Search is a pipe, just like a cable is the pipe for television. Through the pipe you’re bringing consumers to a place where they might find what they’re looking for and that is the channel. Search is more like a free shuttle bus bringing consumers to a mall.

 Then we had another beer and concluded the discussion was very academic. But this morning I was thinking about why nobody is positioning themselves as the find agency. Even the search engine “ask.com” decided not to be about “answer” but about “ask”. Ah well, another industry is even worse. They (and we) don’t find, we don’t even search, we re-search. Although it works as a business model: “We have found an answer! Never mind, we believe you should look again.”

 Peter

Can you beat the law of gravity?

Well of course you cannot beat the law of gravity (but it does make a catchy title)! That’s what makes physics different from economics; as many laws of economics sometimes don’t hold. One of my favorite examples I recall from high school is a so called “Giffen Good”( http://en.wikipedia.org/wiki/Giffen_good). A “Giffen Good” sees an increase in demand when the price rises (violating the law of demand).

 I’ve just been house-hunting in Singapore and I believe I have found another economic law that is being violated. The number of houses and apartments built and entering the market is enormous (sometimes this city feels like one big construction area with new high-rise condo’s popping up throughout the city). So supply is quickly increasing and according to the law of supply and demand, prices should go down. But that is not happening, on the contrary, prices are still going up. So why is that?

My theory is that home-owners have a very high confidence in the real estate market (which by the way is already very high, comparable to New York City). Instead of fearing the increase in supply, it is actually further boosting their confidence. The city is building so many new condos; apparently those constructors and investors know about and plan for an ever increasing influx of people moving to Singapore. Combining that with the fast growing economy and the limited amount of land leads to home-owners asking a premium price and not minding if some nay’s lead to the property staying empty a little longer.

Any economist that wants to react? Hope to be able to claim such a product as a “Kloprogge good”!

Thanks,

Peter

Why a sample works

If you’re in a casino watching the roulette table which of these sequences is most likely: RED-RED-RED-RED-RED-RED-RED-RED-RED-RED or BLACK-RED-RED-BLACK-RED-BLACK-BLACK-BLACK-RED-RED?

As a mathematician/statistician I am sometimes asked why a sample works? If you have a population of 100 million, how come that taking at-random just 200 people will give me a pretty accurate indication of the 100 million?

To understand this best to think of the RED’s and BLACK’s in a casino. The answer to above question is that both sequences have the same probability (they are both just as likely or unlikely to occur). However, most people would say that the sequence with the random red’s and black’s is more likely and the reason is that such a sequence has many, many more “look-alikes”. Another sequence (and a look-alike) with again the same probability would be R-B-R-R-B-B-B-R-R-B. Notice that there is only one unique sequence (with no look-alikes) with only reds.

Now let’s say that we have a big bowl of balls, about 100 million balls of which half are red and half are black (but obviously we don’t know that the reds and blacks are split 50/50). If we take out 200 balls each individual sequence of red’s and black’s is just as likely to occur and as such there is a chance that we only grab 200 red balls, but the probability of this happening is extremely small (so small that if all of humanity was able to take 200 balls every second with no sleep it would still be extremely unlikely to ever see an occurrence of 200 in all of our lifetimes, even if we all lived for a million years).

The reason this is so unlikely is that there is only one way to grab 200 red balls and that is to grab red, and then red, and then red, etc 200 times. It is much more likely that we grab close to 100 red and 100 black balls as there are simply so incredibly more ways of doing this. We could start with one red, then a black, etc or we could start with three blacks, then the first red, etc.

So with even a sample that looks small compared to the total population the number of ways you could sample that will lead to the right conclusion (in this case that half of the balls are red and half are black) simply outnumbers the number of ways you could come to a wrong solution. And by the way, this story as you now might understand is independent of the number of balls in the big bowl; this could be 100 million or just 100 thousand it does not influence the size of the sample we need.

So finally, what do you do if you are in a casino and you see the sequence RED-RED-RED-RED-RED-RED, what should your next bet be? Of course red, the bloody thing is probably broken!

 Thanks,

 Peter

Sell value, not eyeballs

New media: Still all about technology

Words I’ve been hearing a lot the last couple of weeks: Java, HTML 5, Flash, Android Applets, etc. So did I talk with our developers? Did I go to a technology conference? No, I visited a couple of marketing and media conferences.

We can feel it coming. In a couple of years from now we’ll be talking about the old media like search, display, widgets and the new media where location is key and virtual/social lives are influenced linking those behaviors to what people do offline. We know it’s coming, we know it will shake the industry (perhaps more than web 1.0 and even 2.0) but nobody knows exactly what is coming. For now it’s in the hands of those that understand and influence technology.

It’s just like the introduction of the internet. Innovative startups with strong believes in certain technologies will lead innovations and one or two will get it right to become market leaders.

My personal prediction of the big success story: “Suggest marketing”. People are going to be so distracted that serving up relevant suggestion to consumers based on where they are and what they do. No need to search, find it before you’re looking. I actually like the idea of my mobile phone telling me: “If you want to purchase something for your wife on this business trip, you might want to try this book as she has shown interest in this subject lately.”

What we can learn from the traveling salesman?

For those that have worked in the area of operations research, mathematics or computer science the traveling salesman is an intriguing and well known problem. The problem sounds very easy, what is the shortest route traveling from one destination to the next and returning to the place you started. What makes this problem intriguing is that no mathematical algorithm exists that guarantees the shortest route in an acceptable amount of time (to be precise: it’s an NP-Complete problem and as such the most efficient algorithm to date – and probably forever – solves the problem in a time that increases exponentially with the number of cities).  In practice this means that normal CPU’s would need days  or even years to solve problems with a relatively small number of cities (e.g. 100).

Applying marketing data includes many optimizations for which we can prove that no efficient algorithm exists. Examples are television reach optimizers or allocating inventory towards brands. For these problems, as a real optimal solution cannot be guaranteed, mathematicians need to develop heuristic algorithms; an algorithm that tries to get to the best possible solution in a limited amount of time.

And that brings me to the key subject of this blog. The deflation I’ve seen on what it means to optimize. Ten years ago, my clients asked me about how our optimization techniques work; they even got independent mathematicians to audit our methods. Nowadays this doesn’t happen anymore. To have a button in software that says “optimization” seems to be enough to convince prospects. And this is a slippery slope as some optimization routines I’ve seen has moved towards simply users clicking (with some basic information) what they think is best. And let me guarantee, users are not able to optimize as good as good old-fashioned mathematical optimization techniques.

Don’t believe me, please download a little piece of software I wrote many, many years ago. This software allows you to try and find the best route for a traveling salesman before the computer gives it a go. With about 40 cities, on average – in my case – the computer does 8 to 10% better. That’s a lot if it is a 10 mln dollar decision!

And yes, as the computer algorithm cannot guarantee the best solution, you might be able to beat the algorithm. Do let me know (with screenshot!) if you’ve been able to do this.

Is ROI killing creativity?

There are many discussions out there about the future of the agencies and the consensus seems to be that analytics will get a more prominent part in marketing departments and agencies. The discussion than seems to revolve around the tension between ROI and creativity and I remember having done some master classes in Europe with the title “Is ROI killing creativity?”.

I’m obvious happy with more analytics, it’s Pointlogic’s core business. However, as I mentioned in my first blog post, I’ve always specifically enjoyed the tension between numbers and creativity. But my true believe is that these two could and should go hand in hand. If ROI is killing creativity, the only good reason would be if creativity indeed doesn’t deliver a better ROI. I’m guessing most would disagree.

The key problem is, I believe, that most marketers and agencies are comfortable with ROI in a historical sense. How well did this piece of creative perform? How well did this media plan perform? Then the strategic application of the ROI insights are to redo what worked and to stop what didn’t. And yes, this is like glue for creativity as creativity is about new and fresh and not about redoing.

But the solution is relatively straightforward although I know in my daily practice that it is out of the comfort zone of many. But analytical thinking and ROI should be applied in forward looking programs. Demand that analytics provide answers to expected ROI of campaigns before they happen. You might find that it could really transform the way you look at and apply research (with more assessments based on the unknown and more focus on how to monitor and adjust in real time). It’s only then that you’ll find that ROI helps creativity and that creativity helps ROI. And finally, it makes our job as analysts more fun!