Weekly Digest, January 1

Weekly Digest, January 1

Monday newsletter published by Data Science Central. Previous editions can be found here.  The contribution flagged with a + is our selection for the picture of the week.

Featured Resources and Technical Contributions

Featured Articles

Picture of the Week

Source for picture: contribution marked with a +

  Hire a Data Scientist | Search DSC | Classifieds | Find a Job | Post a Blog | Ask a Question


Weekly Digest, January 1

A Primer On Generative Adversarial Networks

A Primer On Generative Adversarial Networks

Guest blog by Keshav Dhandhania and Arash Delijani.

In this article, I’ll talk about Generative Adversarial Networks, or GANs for short. GANs are one of the very few machine learning techniques which has given good performance for generative tasks, or more broadly unsupervised learning. In particular, they have given splendid performance for a variety of image generation related tasks. Yann LeCun, one of the forefathers of deep learning, has called them “the best idea in machine learning in the last 10 years”. Most importantly, the core conceptual ideas associated with a GAN are quite simple to understand (and in-fact, you should have a good idea about them by the time you finish reading this article).

In this article, we’ll explain GANs by applying them to the task of generating images. The following is the outline of this article

  1. A brief review of Deep Learning
  2. The image generation problem
  3. Key issue in generative tasks
  4. Generative Adversarial Networks
  5. Challenges
  6. Further reading
  7. Conclusion

A brief review of Deep Learning

Sketch of a (feed-forward) neural network, with input layer in brown, hidden layers in yellow, and output layer in red.

Let’s begin with a brief overview of deep learning. Above, we have a sketch of a neural network. The neural network is made of up neurons, which are connected to each other using edges. The neurons are organized into layers – we have the hidden layers in the middle, and the input and output layers on the left and right respectively. Each of the edges is weighted, and each neuron performs a weighted sum of values from neurons connected to it by incoming edges, and thereafter applies a nonlinear activation such as sigmoid or ReLU. For example, neurons in the first hidden layer, calculate a weighted sum of neurons in the input layer, and then apply the ReLU function. The activation function introduces a nonlinearity which allows the neural network to model complex phenomena (multiple linear layers would be equivalent to a single linear layer).

Given a particular input, we sequentially compute the values outputted by each of the neurons (also called the neurons’ activity). We compute the values layer by layer, going from left to right, using already computed values from the previous layers. This gives us the values for the output layer. Then we define a cost, based on the values in the output layer and the desired output (target value). For example, a possible cost function is the mean-squared error cost function.

where, x is the input, h(x) is the output and y is the target. The sum is over the various data points in our dataset.

At each step, our goal is to nudge each of the edge weights by the right amount so as to reduce the cost function as much as possible. We calculate a gradient, which tells us how much to nudge each weight. Once we compute the cost, we compute the gradients using the backpropagation algorithm. The main result of the backpropagation algorithm is that we can exploit the chain rule of differentiation to calculate the gradients of a layer given the gradients of the weights in layer above it. Hence, we calculate these gradients backwards, i.e. from the output layer to the input layer. Then, we update each of the weights by an amount proportional to the respective gradients (i.e. gradient descent).

If you would like to read about neural networks and the back-propagation algorithm in more detail, I recommend reading this article by Nikhil Buduma on Deep Learning in a Nutshell.

The image generation problem

In the image generation problem, we want the machine learning model to generate images. For training, we are given a dataset of images (say 1,000,000 images downloaded from the web). During testing, the model should generate images that look like they belong to the training dataset, but are not actually in the training dataset. That is, we want to generate novel images (in contrast to simply memorizing), but we still want it to capture patterns in the training dataset so that new images feel like they look similar to those in the training dataset.

 Image generation problem: There is no input, and the desired output is an image.

One thing to note: there is no input in this problem during the testing or prediction phase. Everytime we ‘run the model’, we want it to generate (output) a new image. This can be achieved by saying that the input is going to be sampled randomly from a distribution that is easy to sample from (say the uniform distribution or Gaussian distribution).

Key issue in generative tasks

The crucial issue in a generative task is – what is a good cost function? Let’s say you have two images that are outputted by a machine learning model. How do we decide which one is better, and by how much?

The most common solution to this question in previous approaches has been, distance between the output and its closest neighbor in the training dataset, where the distance is calculated using some predefined distance metric. For example, in the language translation task, we usually have one source sentence, and a small set of (about 5) target sentences, i.e. translations provided by different human translators. When a model generates a translation, we compare the translation to each of the provided targets, and assign it the score based on the target it is closest to (in particular, we use the BLEU score, which is a distance metric based on how many n-grams match between the two sentences). That kind of works for single sentence translations, but the same approach leads to a significant deterioration in the quality of the cost function when the target is a larger piece of text. For example, our task could be to generate a paragraph length summary of a given article. This deterioration stems from the inability of the small number of samples to represent the wide range of variation observed in all possible correct answers.

Generative Adversarial Networks

GANs answer to the above question is, use another neural network! This scorer neural network (called the discriminator) will score how realistic the image outputted by the generator neural network is. These two neural networks have opposing objectives (hence, the word adversarial). The generator network’s objective is to generate fake images that look real, the discriminator network’s objective is to tell apart fake images from real ones.

This puts generative tasks in a setting similar to the 2-player games in reinforcement learning (such as those of chess, Atari games or Go) where we have a machine learning model improving continuously by playing against itself, starting from scratch. The difference here is that often in games like chess or Go, the roles of the two players are symmetric (although not always). For GAN setting, the objectives and roles of the two networks are different, one generates fake samples, the other distinguishes real ones from fake ones.

 Sketch of Generative Adversarial Network, with the generator network labelled as G and the discriminator network labelled as D

 Above, we have a diagram of a Generative Adversarial Network. The generator network G and discriminator network D are playing a 2-player minimax game. First, to better understand the setup, notice that D’s inputs can be sampled from the training data or the output generated by G: Half the time from one and half the time from the other. To generate samples from G, we sample the latent vector from the Gaussian distribution and then pass it through G. If we are generating a 200 x 200 grayscale image, then G’s output is a 200 x 200 matrix. The objective function is given by the following function, which is essentially the standard log-likelihood for the predictions made by D:

The generator network G is minimizing the objective, i.e. reducing the log-likelihood, or trying to confuse D. It wants D to identify the the inputs it receives from G as correct whenever samples are drawn from its output. The discriminator network D is maximizing the objective, i.e. increasing the log-likelihood, or trying to distinguish generated samples from real samples. In other words, if G does a good job of confusing D, then it will minimize the objective by increasing D(G(x)) in the second term. If D does its job well, then in cases when samples are chosen from the training data, they add to the objective function via the first term (because D(x) would be larger) and decrease it via the second term (because D(x) would be small)

Training proceeds as usual, using random initialization and backpropagation, with the addition that we alternately update the discriminator and the generator and keep the other one fixed. The following is a description of the end-to-end workflow for applying GANs to a particular problem

  1. Decide on the GAN architecture: What is architecture of G? What is the architecture of D?
  2. Train: Alternately update D and G for a fixed number of updates
    1. Update D (freeze G): Half the samples are real, and half are fake.
    2. Update G (freeze D): All samples are generated (note that even though D is frozen, the gradients flow through D)
  3. Manually inspect some fake samples. If quality is high enough (or if quality is not improving), then stop. Else repeat step 2.

When both G and D are feed-forward neural networks, the results we get are as follows (trained on MNIST dataset).

Results from Goodfellow et. al. Rightmost column (in yellow boxes) are the closest images from the training dataset to the image on its direct left. All other images are generated samples.

 Using a more sophisticated architecture for G and D with strided convolutional, adam optimizer instead of stochastic gradient descent, and a number of other improvements in architecture, hyperparameters and optimizers (see paper for details), we get the following results

 Results from Alec Radford et. al. Images are of ‘bedrooms’.

Challenges

The most critical challenge in training GANs is related to the possibility of non-convergence. Sometimes this problem is also called mode collapse. To explain this problem simply, lets consider an example. Suppose the task is to generate images of digits such as those in the MNIST dataset. One possible issue that can arise (and does arise in practice) is that G might start producing images of the digit 6 and no other digit. Once D adapts to G’s current behavior, in-order to maximize classification accuracy, it will start classifying all digit 6’s as fake, and all other digits as real (assuming it can’t tell apart fake 6’s from real 6’s). Then, G adapts to D’s current behavior and starts generating only digit 8 and no other digit. Then D adapts, and starts classifying all 8’s as fake and everything else as real. Then G moves onto 3’s, and so on. Basically, G only produces images that are similar to a (very) small subset of the training data and once D starts discriminating that subset from the rest, G switches to some other subset. They are simply oscillating. Although this problem is not completely resolved, there are some solutions to it. We won’t discuss them in detail here, but one of them involves minibatch features and / or backpropagating through many updates of D. To learn more about this, check out the suggested readings in the next section.

Further reading

If you would like to learn about GANs in much more depth, I suggest checking out the ICCV 2017 tutorials on GANs. There are multiple tutorials, each focusing on different aspect of GANs, and they are quite recent.

I’d also like to mention the concept of Conditional GANs. Conditional GANs are GANs where the output is conditioned on the input. For example, the task might be to output an image matching the input description. So if the input is “dog”, then the output should be an image of a dog.

Below are results from some recent research (along with links to those papers).

 Results for ‘Text to Image synthesis’ by Reed et. al

 Results for Image Super-resolution by Ledig et. al

 Results for Image to Image translation by Isola et. al

 Generating high resolution ‘celebrity like’ images by Karras et. al

Last but not the least, if you would like to do a lot more reading on GANs, check out this list of GAN papers categorized by application and this list of 100+ different GAN variations.

Conclusion

I hope that in this article, you have understood a new technique in deep learning called Generative Adversarial Networks. They are one of the few successful techniques in unsupervised machine learning, and are quickly revolutionizing our ability to perform generative tasks. Over the last few years, we’ve come across some very impressive results. There is a lot of active research in the field to apply GANs for language tasks, to improve their stability and ease of training, and so on. They are already being applied in industry for a variety of applications ranging from interactive image editing, 3D shape estimation, drug discovery, semi-supervised learning to robotics. I hope this is just the beginning of your journey into adversarial machine learning.

About the Authors

Keshav Dhandhania: is a cofounder of Compose Labs (commonlounge.com) and has spoken on GANs at international conferences including DataSciCon.Tech, Atlanta and DataHack Summit, Bangaluru, India. He did his masters in Artificial Intelligence from MIT, and his research focused on natural language processing, and before that, computer vision and recommendation systems.
   
Arash Delijani previously worked on data science at MIT and is the cofounder of Orderly, an SF-based startup using machine learning to help businesses with customer segmentation and feedback analysis.


A Primer On Generative Adversarial Networks

Introduction to Market Mix Modeling

Introduction to Market Mix Modeling

Guest blog post by Jane Sandwood. Jane Sandwood is a professional freelance writer and editor with over 10 years’ experience. She decided to leave the corporate world and move into freelancing to take advantage of the flexibility and work-life balance it offers.

When it comes down to the absolute basics, marketing is all about one thing – getting the right offer in front of the right person at exactly the right moment. You’d have a hard time finding a marketer who doesn’t agree with that statement Oddly enough, many of these same marketers don’t actually live by this creed. According to data collected by Hubspot, nearly 25% of marketers don’t know whether their efforts truly affect the outcome of a conversion.

If I were a gambling man (which I most certainly am), I’d be willing to bet many of these people implement the “spray and pray” method of marketing – that is, they try a whole bunch of tactics, on a whole bunch of target customers, all the time. When one of the tactics works, they stick with it. When none of them do, they chalk up a loss and simply try something – anything – else.

But the problem here is they aren’t paying attention to why a certain campaign performed so well, and why another one fell flat. They don’t realize that what worked today might not work tomorrow, or what didn’twork today (that they end up scrapping entirely) might be hugely successful a few months down the road. In short: they’re not thinking strategically.

If you’re in this boat, the first thing you need to do is take a step back and analyze your overall marketing plan. Ask yourself:

  • What’s working in terms of generating leads and sales? What isn’t?
  • At which stage of the buyer’s journey is a certain tactic working? Where is it not?
  • Who are your customers? What do you know about them that could help improve your marketing initiatives?
  • How does all of this fit together?

When you’re able to determine the answer to these questions – especially the last one – you’ll be in a much better position to create an effective marketing mix. In turn, you’ll have a better understanding of whether or not a certain offer will resonate with a certain customer persona at a given time.

The Definition and Components of Marketing Mix

A marketing mix is the tools, strategies, and tactics a company uses to promote its products or services. When discussing marketing mix, we focus on all of the controllable elements of a marketing plan.

If you’re at all familiar with the term “marketing mix,” you’ve likely heard of the “Four P’s” of marketing.

While you may have seen the terms “marketing mix” and “Four P’s of marketing” be used interchangeably, it’s important to understand the two aren’t synonymous.

While “marketing mix” refers to the strategies and tactics implemented by a marketing team to promote its brand, the “Four P’s” refer to common factors that a team analyzes in order to develop these strategies.

The reason many people use the terms interchangeably is because, in his 1960 book Basic Marketing: A Managerial Approach, marketing wizard E. Jerome McCarthy proposed the “Four P’s” approach to marketing mix, which became the gold standard in terms of forming a marketing strategy for years to come.

However, in the early 1980s, Bernard H. Booms and Mary J. Bitner recognized that the “Four P’s” essentially ignored the fact that customer service is a huge part of any company’s marketing initiatives (even product-focused companies). In turn, they developed the “7P model” of the marketing mix.

An 8th “P” has also been added over time, which we’ll discuss momentarily.

Most recently, marketing experts have begun focusing heavily on the customer side of the marketing equation (rather than taking a brand-centric approach to marketing). This led to the development of the 4C model – again, which we’ll discuss later in this post.

The point, here, is that there are many elements to a marketing mix. Which ones a company chooses to focus on depends on the company’s overall marketing objectives, the products or services it offers, and the needs of its customers.

In the following sections, we’ll dig into the “P’s” and “C’s” of the marketing mix, and explain how analyzing each of them will contribute to a better understanding of how to reach your company’s marketing objectives.

The Four P’s of Marketing

Okay, by now you’re probably wondering:

“What, exactly, are the Four P’s of Marketing?”

Well, the reason I didn’t explain this earlier is because I wanted to be sure we were on the same page in terms of understanding that there’s more to the marketing mix than just the Four P’s.

We’ll get to that 5th P (and more) soon enough.

So, now that we’re straight on that end, let’s talk about the original Four P’s of Marketing. They are:

  • Product
  • Place
  • Price
  • Promotion

In this section, we’ll define each of these terms in greater detail, discuss some of the questions you should ask yourself and considerations you should make regarding each one, and explain how popular brands have used each of these factors when developing a new offer.

Let’s get started.

Product

This probably goes without saying, but the product (or service) you offer your customers largely determines how you should market your brand as a whole.

But it’s essential to remember that your product/service exists in the real world. In other words, you can’t simply assess your product/service in a vacuum.

When developing your product or service, you need to consider:

  • Whether or not there’s a need for it in the current marketplace
  • Whether or not there will be a need for it in the future (or what you’ll need to do to adapt to future needs)
  • The life cycle (growth, maturity, and decline) of similar products or services in the industry

Think back to when Apple released the original iPod. In no way did Steve Jobs invent the MP3 player – but he did notice that the ones that were available at the time certainly were not perfect. Even so, consumers were flocking to the new technology that did away with the need to carry around dozens of CDs at a time; clearly, there was a demand for the new product.

So Apple quickly got to work developing an MP3 player that was as close to perfect as possible at the time.

However, by today’s standards, the original iPod seems almost primitive. Of course, Apple knew this would happen – which motivated the team to create the iPod Touch, the iPad, and, of course, the iPhone. By predicting the needs of consumers in the future, Apple has managed to stay one step ahead of almost every other electronics company in the entire world.

When assessing the product or service you offer (or will offer) your customers, ask yourself the following questions:

  • What do customers want?
  • How, where, and when will they use it?
  • What are the essential features of the product/service?
  • What features are superfluous (either of your own product or of ca competitor’s)?
  • What sets your product/service apart from your competitors’?

Understand the answers to these questions, and you’ll have a much better idea of where your product fits in with the current market. In turn, it will be much easier to showcase the value it can bring to your target customers’ lives.

Price

Again, pretty straightforward – to be successful in any industry, you need to offer your products or services at a price that is affordable to your customers while also allowing you to maximize returns.

Think about:

  • How the pricing of your offer affects your profit margin
  • The perceived value of your product or service
  • Along with perceived value, your brand’s reputation throughout your industry
  • The price of similar products being offered by your competition

It’s also worth noting that the price you set for your products or services can inherently affect how consumers view said product or service in the first place.

On the one hand, pricing your products on the low end of the spectrum can border the line between “affordable” and “cheap”; set the price too low, and your target consumer might assume your product is rather low in quality.

On the other hand, setting your price too high will certainly turn off a decent amount of your target consumers who either refuse to spend more than a certain amount of money on the product you offer, or who literally can’t afford to pay the price you’ve set.

It’s important to understand that the price you offer your product at should never be set in stone, for a variety of reasons. For one thing, demand for your product will fluctuate (as alluded to in the previous section); for another, the cost of producing your product will likely change over time. Again, these are only two of the many factors that affect product pricing; you’ll need to revisit these factors from time to time to ensure you’ve set an optimal price for your offering.

For a detailed analysis of how coffee giant Starbucks increased its revenue from $331 million to $417 million in one quarter by raising the price of “tall” sized coffees by 1%, check out Tucker Dawson’s article featured on Price Intelligently. A 1% increase in price gave Starbucks an extra $86M revenue… in one QUARTER

In general, when determining the pricing of your products or services, you should ask yourself:

  • How much does it cost to produce one unit or provide one instance of service?
  • How much do you aim to make during a single exchange?
  • How much do customers expect to pay for your product/service?
  • Is your price hurting sales? Helping?
  • Is your price hurting or helping your profit margin?

While everything about your marketing initiatives is, at one point, a matter of trial and error, this is most often the case when it comes to pricing. As mentioned earlier, you should always be prepared to assess the way you’re pricing your offerings to see if there’s room for improvement.

Place (Placement and Distribution)

Place refers to the logistics of where (and how) you actually deliver your product or service to your customers.

When considering placement of your product/service, you need to know where, why, and how your target customers tend to do their shopping. Additionally, you’ll want to determine where – among all of these possible places – they’re most likely to purchase your product. Then, you’d put all of this information together to determine which channels you should use in order to maximize output while bringing in the most profit possible.

Mall vacancy is at record highs because of the current e-commerce boom. For example, say your company develops canes and walkers for elderly individuals. After doing research on your customer base, you determine that only 5% of them purchase your products via your online store, while the rest prefer to visit your brick-and-mortar locations.

Using this information, you’d know to focus your efforts on opening more physical stores, or partnering with already-existing distribution centers (such as drug stores and department stores) rather than trying to force your customers to use a channel they’re not comfortable with.

In terms of distribution, there are three main strategies to choose from, depending on your customers’ preferences and needs:

  • Intensive Distribution: Companies using this method of distribution attempt to get their products in front of as many eyes as possible, in a variety of places. For example, if you were in the mood for Coca Cola, you could head to any one of your local convenience stores, drug stores, or grocery stores to get a bottle or two.
  • Selective Distribution: A company that offers more than one product (each of which varies in quality and price) might employ a selective distribution strategy. For example, a hardware company might partner with Walmart to sell its more basic, lower-end tools, and will partner with Home Depot to sell its more advanced, specialized wares.
  • Exclusive Distribution: Exclusive distribution refers to instances in which a company’s product is onlyoffered at a specific store or location. For example, grocery stores such as Aldi, Whole Foods, and Trader Joe’s offer certain brands that can only be found at these respective stores. Companies that utilize exclusive distribution place high importance on building brand loyalty.

As with placement, distribution also takes into consideration how to provide a product in a way that’s most convenient for both the company and the consumer.

Promotion

Promotion refers to any action, taken by a company or its customers, that gets the company’s name out in the public eye in a positive way. Such promotion can either be paid for in some way or another, or it can be organic.

Some of the most common forms of promotion are:

  • Paid advertisements
  • Public appearances and events
  • Discounts, sales, and freebies
  • Contests
  • Social media marketing and content marketing
  • Referral marketing

When deciding on a promotion strategy, consider the following questions:

  • What mediums do your target consumers use (especially when in “buying mode”)?
  • When are they most likely to notice and come into contact with your promotion?
  • How are your competitors promoting their products?
  • What is your budget for creating and implementing these promotions?

Above all else, a successful marketing campaign is one that provides value to the consumer. This value might come in the form of a discount for frequent shoppers, or it might be a series of blog posts explaining how to get the best use out of your product. It all depends on what your customer is looking to get out of their engagement with your brand.

What’s Missing?

As mentioned earlier, the original Four P’s of marketing focused mainly on product-related marketing, and didn’t offer much in terms of best practices for marketing services.

Not only that, but even product-focused companies now tend to define their unique selling proposition based on the added services they provide (rather than just the features of their product).

Because of this, three new “P’s” (and a fourth) have been added to the marketing mix equation, which we’ll discuss in the following section.

The 7 P’s Of Marketing (And the 8th P, Too)

The four “new” P’s of Marketing, which focus more on the service side of things, are:

  • People
  • Process
  • Physical Evidence
  • Performance

There’s still one more to add, too…Let’s take a look at each of these categories in greater detail.

People

Simply put, without qualified, competent personnel working hard behind the scenes and on the floor, no company can succeed.

But it’s really not enough for a company’s team members to be qualified; they need to also be truly dedicated to the company’s mission in order for the company to truly reach its full potential.

Not only will such alignment aid the company in terms of productivity, but – as alluded to earlier – it can also become a part of the company’s USP. Ensuring your company is full of competent, diligent, and dedicated employees will inherently improve the services you provide. In turn, this will attract consumers who are looking for that little “extra something” from their experience with your brand.

I’ve talked about Chick-fil-A in a bunch of different posts on Fieldboom’s blog, but there really is no better example of how a dedicated team of employees can make all the difference to a customer’s experience. Everyone who works at Chick-fil-A – from the fry cooks to the franchise managers (and beyond) – all embody everything CEO Dan Cathy believes the company should be.

Without such a heavy focus on customer service, Chick-fil-A may very well be just another fast food joint. For companies like Southwest, it’s all about letting employees be themselves, which attracts and retains customers.

Processes

Processes refers to all of the actions that a company implements when delivering a service to its customers. Of course, this is a rather broad definition, as essentially everything a company does ultimately goes toward providing quality services to its customers. So, for our purposes, let’s discuss the processes in which customers are directly involved.

In thinking of processes in this way, we need to think of all of the touch points a customer experiences from the moment they walk into a storefront (or log on to a website, or contact a customer service helpline…) to the moment this engagement ends. A few questions to consider:

  • How do your employees greet and assist your customers? Do they “assign” themselves to walk-ins, a la Apple Stores? Do they offer assistance if necessary, like most clothing stores? Or do they take a hands-off approach and make themselves available only if a customer seeks them out?
  • How does the checkout process work? What payment options do you offer your customers? How about options for making returns?
  • How does your company handle complaints? Is a manager always available during operating hours? Do you have a fully-staffed, on-site customer service department? Do you offer 24/7 support either by telephone or via email/social media?

It’s worth noticing that “processes” deals with both the planning behind a company’s operations and the people who carry out these operations. As noted in the section above, it’s often the dedicated employees of a company who bring value to said company’s customers. However, without proper guidelines in place for these employees to follow, even the most talented and dedicated individuals would have a tough time carrying out a company’s mission.

Physical Evidence

In service industries, physical evidence refers to all the tangible parts of the otherwise intangible experience customers have when they engage with a company.

Physical evidence, in this regard, refers to three different aspects of a brand:

  • Ambience
  • Spatial layout
  • Corporate branding

Ambience refers to the overall “feel” of the location in which a service is rendered. This includes aspects such as color scheme, volume and genre of background music, and level of background noise. For example, a coffee shop might be decorated with relaxing colors, play instrumental music softly in the background, and have sound dampeners placed on walls to reduce echoes – allowing students and remote employees to focus and get some work done.

On the other hand, a gym would likely be decorated with “loud” colors while blasting rock and hip hop music at full blast in an effort to pump *clap* you up.

Spatial layout, quite simply, refers to how furniture and other “equipment” is placed throughout the service area. Think of the last time you went to a restaurant that was way too crowded: even if the food was amazing, you probably will at least think twice about returning any time soon.

One thing to note, here, is that proper ambience and spacing alone isn’t enough to bring in customers if the service you offer isn’t on-par. On the other hand, a sub-par ambience or spatial layout can certainly detract from the overall customer experience.

The best way to explain corporate branding is to ask the question: “If I were to be blindfolded and brought to a random service area, would I be able to tell where I was?” Or, similarly, “If I were to be brought to a website, would I immediately be able to tell which site I was on?”

Think about the differences between Target and Walmart, of McDonald’s and Wendy’s – not in terms of product or service provided, but in terms of the recognizable (sometimes trademarked) aspects of the environment that tell you right away where you are. No company understands this better than Apple – and that’s reflected in their value as a company.

Performance (Or Productivity and Quality)

Whether you call it Performance or Productivity, the 8th P of Marketing is all about how well your company operates – in the eyes of your customers – in comparison to all of the other companies in your industry.

Often, your company’s ability – and drive – to go the extra mile to ensure your customers’ satisfaction is the key differentiator that will keep them coming back whenever they’re in need of the services you offer. Performance is incredibly important in service related industries.

Note: This isn’t to say it’s not important within product-focused companies. But, for example, most consumers wouldn’t choose Pepsi over Coca Cola because of Pepsi’s superb customer service.

Businesses such as restaurants and hotels can definitely leverage performance as a selling point that sets them apart from the crowd. For example, this story of how staff members at a Marriott went above and beyond to make life a little easier for a mother who had recently given birth certainly led to some positive publicity for the hotel chain.

Performance is all about a company’s drive to really “wow” their customers, providing them with top-quality value by any means necessary. It means never being satisfied with the status quo, and always probing deeper to see what else you can do to make your customers happy.

The 4 C’s of Marketing

As mentioned earlier in this post, the marketing world has, in the past few decades, undergone a shift in focus and philosophy.

The “old way” of marketing has all but disappeared. Casting a wide net and reeling in any ol’ consumer in the hopes that they become a paying customer is no longer effective. The modern customer yearns to be treated as an individual by even the largest corporation.

Understanding this, in the early 1990s Robert F. Lauterborn developed the 4 C’s of Marketing as an updated version of the 4 P’s.

They are:

  • Customer wants and needs
  • Cost of providing value
  • Convenience
  • Communication

A more modern take on the 4 P’s of marketing. Let’s dive into each, and discuss how each relates to a corresponding “P” – but does so in a way that focuses on the consumer receiving a service rather than the company providing it.

Customer Wants and Needs (Instead of “Products”)

When you think about it, focusing on products before focusing on consumer needs is sort of putting the cart before the horse, isn’t it? In the modern business world, developing a product or service before thinking about whether or not there’s a demand for said product or service in the first place will almost certainly lead to failure.

Really, it’s always been this way. The major difference today is that uncovering the needs of your vast base of target customers has never been easier. That’s not to say that doing so is easy – it’s just much easier than it was in the pre-digital age.

Rather than gambling on whether a product you’ve already developed will succeed, you can now easily validate your idea at a fraction of the cost before you sink money into developing it.

The internet gives you an opportunity to:

  • Reach out to your target customers to understand their wants and needs
  • Research customer reviews of similar products to understand where there’s room for improvement
  • Create content relating to your idea to see if it creates enough of a buzz to warrant development of a product

In doing so, you can enter into the world of business with a much better understanding of who your customers are, what they want, and how you can give it to them. Equinox doesn’t sell “working out” – they sell a better lifestyle.

Cost of Receiving Service (Instead of “Price”)

When you set a price for your product or service, you’re unwittingly placing your offer in a vacuum. Of course, your customers don’t live in a vacuum. They live in the real world.

If a customer purchases your product, they’ve traded a lot more than just the money they give the cashier at the register. They’ve also traded gas (and gas money) to drive to your store. They’ve traded time that could have been spent elsewhere for time spent browsing your wares. They’ve expended energy and stress wondering if your product will be worth the amount they’ve paid for it.

At first glance, it might seem like there’s not much you can do to alleviate these extra costs to your customer. But think about it – maybe you could offer free shipping for purchases made online – alleviating your customer’s need to drive to your brick-and-mortar location.

You could improve your customer service, or change your store’s physical layout, to make the process of finding and purchasing products more streamlined. Or you could offer a timed return policy or warrantee on all items, easing your customer’s worries that they might lose money on the deal. Simply put, rather than focusing on price optimization, focus on alleviating cost to the consumer.

Convenience to Buy (Instead of Place)

Amazon is the king of a high-converting buying experience when it comes to e-commerce. This goes along with the last section, and also deals with optimizing processes as well. Let’s assume that your storefront is located in a rather convenient area (or, if more applicable, your online shop is easy to navigate).

When it comes time for a customer to make a purchase, is this process just as convenient? By today’s standards, there are a ton of different ways to make payments: cash, cheque, credit card, debit card, PayPal, Apple Pay and mobile payments…the list goes on. If your company doesn’t support one (or even more) of these methods, you’re bound to lose business at some point in the near future.

Not only that, but it’s also important to consider what else is involved in the payment process. For example, do you require customers to provide ID when not using cash? Is the signup process for your mobile payment system tedious and time-consuming? Does your credit card reader accept the newly-implemented chip card?

When implementing modern methods of payment, make sure the rest of your payment process is simple and straightforward – otherwise, you may end up unintentionally making things less convenient for your customer.

Communication (Instead of Promotion)

Remember how we talked about how relatively easy it is to get in touch with your target customers nowadays? Well, because of this, the modern consumer essentially expects it from the companies they do business with.

While one-way advertisements and promotions (such as commercials and magazine ads) aren’t exactly relics of the past just yet, they certainly have taken a backseat to the more interactive forms of marketing and advertising we see today.

Content marketing, social media marketing, even email marketing…these are all two-way streets – if you allow them to be. In other words, it’s not enough to “set it and forget it” by publishing a blog post, scheduling a post on Facebook, or sending out an email blast; you need to actively engage with the members of your audience that respond to these campaigns.

In this way, you’re not simply promoting your product to consumers; you’re promoting your value to individuals who are in need of your services. In doing so, you tear down the wall that acts as a barrier between supplier and customer, and forge a more authentic relationship with the people who need you most.

A Few Notes on International Marketing Mix

Although we often discuss the idea that every consumer is an individual, we tend to do so while at the same time assuming they still have certain foundational characteristics in common.

But this only applies when discussing your company’s “domestic” customer base (even this isn’t exactly accurate, either).

When promoting your brand to an international audience, you essentially have to start from scratch. Although there are bound to be some similarities between your domestic and international audiences, failure to recognize even the most minute difference could lead to disaster for your company.

Let’s go back to the original 4 P’s of Marketing to get a feel for some of the important things to think about when marketing to an international audience:

  • Product: The products you offer – for a variety of reasons – may or may not resonate with a certain international group. You’ll need to consider the group’s cultural and religious background, their average level of income, their buying habits, and more. For example, McDonald’s restaurants in India don’t serve beef products in adherence to the Hindu religion.
  • Promotion: Again, when promoting your product or service to an international audience, you need to consider cultural and religious norms, and you need to have a firm grasp of the language of the people. For example, while a character giving the “thumbs up” sign on a billboard in America would conjure up positive thoughts, the same photo would be considered incredibly offensive in many Middle Eastern countries.
  • Pricing: We mentioned earlier that you’ll need to consider your international audience’s level of income when promoting your brand abroad. But, once your audience begins paying for your products, you’ll need to consider other factors, such as taxes, the exchange rate, and the cost of transporting your items throughout the world.
  • Place: Domestically, the chain of product transfer generally goes: manufacturer, wholesaler, retailer, consumer. In many other countries, this chain can be more like a tree, in that it branches off in certain areas (such as involving multiple wholesalers). Without a clear understanding of how your product will be delivered, there’s little chance of being able to optimize your reach in a certain area.

Whether marketing your product or service domestically or internationally, it’s essential that you clarify as much information as possible. In doing so, you’ll be able to avoid everything from minor setbacks to major pitfalls while ensuring your customers get the most value possible from your brand.

Conclusion

Attempting to market your product or service without knowing and understanding what will or won’t work is a fool’s errand. Sure, you might get lucky and gain some major publicity through a single campaign. But even if this does happen, you’ll have little to no idea why it worked – and won’t be able to replicate the results in the future.

On the other hand, if you work to understand your market – and the individuals who make up the consumer side of this market – you’ll have a much better chance of creating something that your target audience will find valuable.

And, even if you miss the mark, you’ll at least be able to pinpoint why you fell short of your goals…

Originally posted here


Introduction to Market Mix Modeling

2018: The Year of the Self-Learning Data Organization

2018: The Year of the Self-Learning Data Organization

As 2017 ends, Ramon Chen, Chief Product Officer at Reltio, the creator of data-driven applications, has peered into his crystal ball to decipher what 2018 will bring in data management. Find his predictions below. 

2018 will be the year of AI and Machine Learning … again: There have been repeated predictions over the last couple of years touting a potential breakthrough in enterprise use of Artificial Intelligence and Machine Learning (ML). While there are no shortage of startups – CBInsights published an AI 100 selected from over 2000+ startups – the reality is that most enterprises have yet to see quantifiable benefits from their investments, and the hype has been rightly labelled as overblown. In fact, many are still reluctant to even start, with a combination of skepticism, lack of expertise, and most of all lack of confidence in the reliability of their data sets.

In fact, while the headlines will be mostly about AI, most enterprises will need to first focus on IA (Information Augmentation): getting their data organized in a manner that ensures it can be reconciled, refined and related, to uncover relevant insights that support efficient business execution across all departments, while addressing the burden of regulatory compliance.

 The elements of data strategy (Source: HBR May-June 2017)

Enterprise data organization, not management, will be the new rallying cry: For over 20 years, the term data management has been viewed as a descriptor, category and function within IT. The term management represented a wide variety of technologies ranging from physical storage of the data, to handling specific types of data such as Master Data Management (MDM), as well as concepts such as data lakes, and other environments. Business teams have lost patience with the speed, and efficiency in which they are able to get their hands on reliable, relevant and actionable data. Many have invested in their own self-service data preparation, visualization and analytics tools, while others have even employed their own data scientists. The common refrain is that data first has to be made reliable, and connected with the rest of the enterprise, so that it can be trusted for use in critical business initiatives, and isolated initiatives such as MDM and Hadoop-powered data lakes have not been successful.

Organizing data across any data type or source, with ongoing contribution and collaboration on limitless attributes, will be the new rallying cry for frustrated business teams as it describes a state of continuous IA (Information Augmentation) that enterprises want to achieve before they can even consider AI as a potential next step.

 

Data-driven organizations will expect to measure outcomes: While being data-driven continuous to be vogue, companies have had surprisingly little in the way of measurable, quantifiable outcomes for their investments in technologies and tools. Certain Total Cost of Ownership (TCO) metrics such as savings realized from switching to cloud vs.on-premises are obvious, but there hasn’t been an obvious and clear direct correlation between data management, BI, analytics and the upcoming wave of AI investments. What’s missing is a way of capturing a historical baseline, and comparing it to improvements in data quality, generated insights, and resulting outcomes stemming from actions taken.

Much of this can be attributed to the continued disconnect between analytical environments such as data warehouses, data lakes and alike where insights are generated, and operational applications, where business execution actually takes place. Today’s Modern Data Management Platforms as a Service (PaaS) seamlessly power data-driven applications which are both analytical and operational, delivering contextual, goal-based insights and actions, which are specific and measurable, allowing outcomes to be correlated, leading to that Return on Investment (ROI) Holy Grail, and forming a foundation for machine learning to drive continuous improvement. As an added bonus, multitenant Modern Data Management PaaS in the Cloud, will also begin to provide industry comparables, so companies can finally understand how they rank relative to their peers.

 

Multi-cloud will be the new normal: With the Cloud Infrastructure as a Service (IaaS) wars heating up, players such as Amazon Web Services (AWS), Google Cloud Platform (GCP), and Microsoft Azure continue to attempt to outdo each other on all vectors including capabilities, price, and service. With fears of being “Amazoned,” some retailers have even adopted a non-AWS Cloud policy. For most, however, it’s about efficiency and cost. Multi-cloud means choice and the opportunity to leverage the best technology for the business challenges they face. Unfortunately, multi-cloud is not realistic for all, only the largest corporations who have the IT teams and expertise to research and test out the latest and greatest from multiple providers. Even those mega-corporations are finding that they have to stick to a single IaaS Cloud partner to focus their efforts.

Today’s Modern Data Management PaaS are naturally multi-cloud, seamlessly keeping up with the best components and services that solve business problems. Acting as technology portfolio managers for large and small companies who want to focus on nimble and agile business execution, these platforms are democratizing the notion of multi-cloud for everyone’s benefit.

Companies will execute offensive data-driven strategies, and should expect to get defense for free: Effective May 25, 2018, the European General Data Protection Regulation (GDPR) will force organizations to meet a standard of managing data that many won’t be able to fulfill. They must evaluate how they’re collecting, storing, updating, and purging customer data across all functional areas and operational applications, to support “the right to be forgotten.” And they must make sure they continue to have valid consent to engage with the customer and capture their data.

Meeting regulations such as GDPR often comes at a high price of doing business not just for European companies, but multinational corporations in an increasingly global landscape. Companies seeking quick fixes often end up licensing specialized technology to meet such regulations, while others resign themselves to paying fines that may be levied, as they determine that the cost to fix their data outweighs the penalties that might be incurred. With security and data breaches also making high-profile headlines in 2017, it’s become an increasingly tough environment in which to do business, as the very data that companies have collected in the hopes of executing offensive data-driven strategies, weighs on them heavily, crushing their ability to be agile.

As previously outlined, organizing data for the benefit of machine learning, or other initiatives results in clean, reliable data that is connected and forms a trusted foundation. A natural byproduct is a defensive data strategy, with the ability to meet regulations such as GDPR, and to ensure compliant, secure access by all parties to sensitive data. This is an amazing two-fer from which regulatory teams and CDOs can both benefit.

Whatever the industry or business need, organizing data in 2018 should be a top priority for companies big and small.

DSC Resources

Popular Articles

2018: The Year of the Self-Learning Data Organization

Selected Recent Articles from Top DSC Contributors – Part 1

Selected Recent Articles from Top DSC Contributors – Part 1

This is a new series, featuring great content from our top contributors. Some of these articles are rather technical in nature, but many are business-oriented and written in simple English. The entire series consists of about 120 articles. We intend to publish a new set every two weeks or so. This is the first edition. To read more articles from a same author, read one of his/her articles and click on his/her profile picture to access the full list. Some of these articles are curated or posted as guest blogs.

Selected Recent Articles from Top DSC Contributors

Source for picture: articled flagged with a +

DSC Resources

Selected Recent Articles from Top DSC Contributors – Part 1