The transformation in retail that started in 2020 is not about new trends. Rather, it is defined by mass adoption of services that were previously offered by only selected retailers. The new purchase journey now takes four distinct paths: 1) drive and purchase – travel to a brick-and-mortar store; 2) click and pick – purchased online and pickup at the store; 3) click and deliver – purchase online and have the item delivered; and 4) subscribe and forget – subscribe online and have the item delivered a regular interval (CPG brands). Consumers have adapted to these services and clearly see trade-off between time/effort and money. The retailers that will come out on top will be those that can offer the most frictionless purchase journeys.
Every HR department and talent acquisition agency knows that in today’s tight labor market it is extremely difficult to find the talent that meets the organization’s business needs. In other words, the way we go about matching talent to job openings is still pretty much in the dark ages. For employers, the process is slow, expensive and labor intensive. For applicants, it is often difficult to find open positions that are a fit.
However, today we can leverage artificial intelligence (AI) and machine learning (ML) to transform the way we match talent to employers. An AI/ML solution could be built using multiple technologies but using the Google or Amazon Cloud offers cost savings, scalability and speed that few can match.
The above is a sketched architecture of what the platform could look like using the Google Cloud. I will call this solution Deep Signal.
Sample Use Cases
So, what could you do with Deep Signal? Here are a small sample of use cases:
1. Candidate Sourcing and Placement – Source candidates faster and more accurately (decrease the time from first contact to placement)
2. Source Specialty Candidates – Identify difficult to find candidates with specialized skills
3. Candidate Recommendation Engine – Recommend candidates based on look-a-like modeling (if you liked this resume, you will also like these other resumes)
4. Employer recommendations Engine – Recommend employers based on look-a-like modeling (if you are interested in this employer, you will also like these other employers)
5. Smart Search – Job search based on non-traditional features (commute time, vacation, etc.)
6. Next Best Action – Make recommendations based on actions taken by employer or candidate
7. Identify Prospective Clients – Identify prospective clients based on existing candidate database (this prospective employer will be interested in these candidates)
There are two things that will be needed in order to build a AI driven platform. The first one is data. Without extensive categorical and historical data, you cannot build accurate machine learning models. The second thing we will need is computing power. Without enough computing power, it could take weeks or months to run our models. Both Amazon and Google offer solutions with more than enough computing power to handle anything you throw at them.
Transforming into a data-driven business model based on machine learning and artificial intelligence requires more than deploying new technologies. It often requires cultural and organizational shifts as the business adapts to the realignment of data, people, process and technology.
As always, questions and thoughts are welcome. If you give me a use case and I will think of ways ML/AI could be used.
From a marketing perspective, enabling the healthcare or prescription journey is a complex landscape of intertwined stakeholders, regulations, technologies and touchpoints. The interests and needs of healthcare providers (HCPs) and patients will converge and diverge at different points. Having the right technology infrastructure and targeting strategies is key to the successful completion of this journey.
As usual, my sketch over simplifies things but it does capture the key technology platforms, data models and touchpoints at each stage of the journey. For the most part, platforms and strategies stack on top of previous ones as we move along the journey–rather than being solely use in one stage. Given that healthcare is a highly regulated industry, a strong privacy and data governance strategy is of the upmost importance.
Machine learning is an evolving and complex science. If one takes into account all possible scenarios, dependencies and models, it would be impossible to sketch an all-encompassing explanation. So I will focus on what the non-data scientist should know about machine learning. That is, they should focus on understanding available data, machine learning types, algorithm models and business applications.
1. First, determine if what is the nature of the available data. Do you have historical campaign data with historical results or an unstructured database of customer records? Or are you trying to make real-time decisions on streaming data?
2. The data available–and use case–will determine the appropriate machine learning approach. There are three major types: supervised, unsupervised and reinforcement learning. a) Supervised Learning allows to you to predict an outcome based on input and output data (e.g. churn). b) Unsupervised Learning allows to you categorize outcomes based on input data (e.g. segmentation). c) Reinforcement Learning allows you to react to an environment (e.g. driverless car).
3. Each machine learning type will use a number of algorithm. There are hundreds of variations. Supervised Learning typically uses regression or classification algorithms. Unsupervised Learning uses, but is not limited to, clustering algorithms. Reinforcement Learning will usually use to type of neural network algorithm.
4. The uses of machine learning are nearly limitless. Although I listed here as the last step, determining the use case or business application should probably be the first step. For example, we would use logistic regression to determine whether a house would sell at a certain price or not and we would use linear regression to predict the future price of a house. You should keep in mind that machine learning business applications typically require the use of more than one machine learning type as well as multiple algorithms.
I hope this sketch serves as a useful guide. Feel free to share and use as needed.
Thoughts and comments as well as suggestions for other sketched explanations are welcomed.
There are countless algorithms we can used to mathematically predict an outcome to a business challenge. However, the most widely used algorithms will fall into four categories: classification, continuous, clustering and recommendation.
Let’s use a real life example to illustrate how we choose the right algorithm to solve the right problem. For illustration purposes we are making a number of assumptions to keep things simple for the non-analyst.
Let’s say that a realtor is trying to answer the following questions:
- Will a couple buy a house? Here we are looking for a categorical answer of Yes or No. For this we would use some kind of Classification algorithm, which could include: Logistic Regression, Decision Trees or Convolutional Neural Network
- How much will they pay of the house? For this question we would use Continuous estimation as we trying to determine the value in a sequence. Is this case, one would likely use a Linear Regression algorithm.
- Where will the buy the house? Clustering would be the best approach to determine where they are likely to buy a house. K-means and Affinity
- If they buy a house, what else will they buy? Recommender System Algorithms are commonly used to determine next best offer or next best action. The most commonly used Recommender algorithm is Collaborative Filtering: either user-to-user or item-to-item.
A new business paradigm is forming, one that aligns business intelligence, creativity, big data and technology to deliver relevant connected experiences to a new generation of hyper-connected consumers. To succeed in the coming years, companies will need to fuse together a myriad of technology platforms, segregated business data, and a somewhat ineffective media landscape to reach Millennial consumers and the generations that follow.
From crawling to flying, the journey to becoming a data driven organization that can deliver connected experiences has four distinct phases. Each of these transformational phases are enabled by five core drivers: data, technology, intelligence, content and experiences.
This is just a simplified reference architecture of today’s marketing technology stack. Excuse all the acronyms but I will be doing a post soon on the most used/overused acronyms in marketing.
Privacy and Data:
- The Cookie Crumbles – With Google following on Apple and Mozilla’s steps and putting in motion plans to phase third-party cookies, marketers will have to reassess how they track and target customers. Apple’s coming changes to its Mobile OS tracking policies will only compound the problem for brands. While we don’t know exactly what will replace third-party cookies (i.e. Sandboxes, FLOCs, etc.), content will be king and contextual targeting will make a comeback.
- GDPR and Privacy Regulations – The effects of GDPR have been felt far beyond the EU as brands embrace it as the gold standard to safeguard consumers’ privacy. In the US multiple states have introduced laws similar to California’s CCPA. Furthermore, Japan, India, Canada, Brazil and Korea have passed privacy laws similar to GDPR. In 2021, governments across the world will continue to enact stringent privacy laws.
- First Party Data and Universal IDs – These changes will force brands to either develop robust first-party data assets or be at the mercy of Walled Gardens—likely both. 2021 will see significant investments in the technology infrastructure and partnerships needed to build these assets. Another way to counter the Walled Gardens will be to use Universal IDs from vendors such as The Trade Desk, the Advertising ID Consortium or ID5. However, until an industry standard is accepted, identity management will be something marketers will have to focus on.
- Headless Commerce – Headless commerce, which is the decoupling of backend and frontend operations, is a must of any brand that wants to compete in digital commerce. One area where marketers need to quickly pivot is social as it has it shifts from being a mostly brand channel to becoming a performance medium. For more on headless commerce, see my other post.
- Direct to Consumer – One of the doors that headless commerce opens, is that it makes easier to CPG brands to bypass retailers and sell direct to consumers. From Procter & Gamble to Colgate, CPG brands are rushing to build the infrastructure needed to reach consumers directly. This will also require that CPG marketers expand their skillset.
- Retailers as Media Networks – Another trend that will continue to proliferate is that of retailers becoming their own media networks. This is a strategy that Amazon mastered early on and now other retailers (i.e. Home Depot, Target, Walmart, Instacart, etc.) are trying to mimic. These retailers are now becoming mini walled gardens. In 2021, the retail media landscape will continue to fragment and marketers will have to deal with a more complex media landscape.
- Connected TV – Connected TV is expected to grow 30% to 205 million users by the end of 2021. With a myriad of vendors now offering solutions that allow for the delivery of targeted content both dynamically and programmatically, marketer must rethink content and strategies.
- Supply Path Optimization – Although it is not something new, the proliferation of SSPs and DSPs has made programmatic supply path optimization (SPO) a priority for 2021. SPO has two major components. On one hand, it consolidates and prunes inefficient SSPs. On the other hand, is uses machine learning to determine the most optimal path to inventory.
- Business KPIs – As the current recession puts pressure on marketing budgets, marketers are demanding that programmatic strategies be optimized to business performance indicators and not simply impressions, clicks or engagements. Machine learning now gives marketers to optimize their media buys based on content, product inventory or profitability.
- Dynamic Content Optimization – Investments in technology and data will not only allow marketers to identify the right consumers but also to deliver content and messaging dynamically and at scale through Dynamic Content Optimization (DCO). Today, this is easier than ever for both small and large brands. For example, marketer can use solutions from Adobe, Salesforce, and Oracle to dynamically manage content in their digital properties and at the same time use solutions from Flashtalking, Jivox and Sizmek to optimize their media content.
- Organic – Creating, managing and optimizing organic content should be at the top of any marketing strategy for 2021. One catalyst for this will be Google emphasis on its EAT (Experience, Authority and Trust) algorithm. Under EAT, Google will give higher ranking to pages that demonstrate relevant quality and context. Another catalyst is the reduction in paid media and marketing budgets due to the current recession.
- Streaming and Video – The pandemic has brought streaming services, video content and video chat technologies to the forefront. Two trends are of note. First, streaming has become nearly universal so it must be part of every marketer’s strategy. Second, the subscription model will not be enough for all players to survive. Thus, advertising opportunities will be abound and the largest of these streaming services will become walled gardens. Finally, the introduction of 5G technology will mean that streaming content will be increasingly consumed through mobile devices.
- Retail – Google might still be king of the hill when it comes to search but it is no longer the only name in town. In particular, when it comes to products 54% of searches now take place through Amazon. As more retailers become media networks, search strategies can no longer focus solely on Google and must include a multi-pronged approach.
- Voice – 50% of consumers a currently use voice search and by the end of 2021 sales of smart speakers will surpass that of tablets. Marketers will have to rethink their keyword query strategies and take into account speech patterns as they integrate voice into their search marketing strategies.
- Visual – Given that 90% of all information processed by the brain is visual, Visual Search is the natural evolution of where consumers are going. This is already the case in 2021. For example, there are over 600 million visual searches on Pinterest every month and Google Lens has reached 500 million downloads in the Play Store. Marketers must now take into account how consumers visually search and how the consume/use visual search returns.
Abundance of data and advances in technology–including the proliferation of low code and no code platforms–means that Artificial Intelligence and Machine Learning (AI/ML) Marketing is at the finger tips of most organizations. If we think in terms of Data, Technology, People, and Process, the first two are not the impediment to leveraging AI/ML to drive marketing effectiveness and sales. Rather, it is the shortage of skilled people and the lack of understanding of which models and processes solve different marketing challenges.
I put this basic framework together to give an idea of different AI/ML techniques and their applications in marketing. Do keep in mind that the combination of models and applications can number in the hundreds. This is why skilled data scientists and data engineers are so important to marketing organizations. To keep things simple I selected 12 tactics with their respective algorithms across four categories.
Product Segmentation <> Collaborative Filtering
Journey Stage Segmentation <> Gaussian Mixture
Behavioral Segmentation <> K-Means
Churn <> Logistic Regression
Conversion <> Decision Tree
Customer Lifetime Value <> Random Forest, XG Boost
A/B Testing <> Multi-Arm Bandit
Timing <> Sequence Analysis
Next Best Action <> Nearest Neighbor
Sentiment <> Convolutional Neural Network
Semantic Search <> OWL-Search
Content Extraction <> Support-Vector Machines
Again, any of these marketing tactics could use completely different AI/ML models or non at all depending on what we are trying to solve.
For more on this topic, see my other blog posts on understanding and applying machine learning.
Contextual targeting is done through the use of keywords and categories. For example, we might serve an ad for Nike basketball shoes on the NBA website. Behavioral targeting relies on known user behaviors that have been tracked using cookies and is sometimes enhanced with demographic data. With third-party cookies phasing out, contextual targeting is making a comeback.