By Stuart Renshaw

For years CPG brand marketers, agencies and publishers have talked about food content personalization across multiple channels in the race to grow revenue, engagement and ROI through more relevant user and shopper experiences. Despite a lot of smart thinking and discussion, most vegetarians still have to plow through meat dishes in their social feed and marketers still struggle to understand what content is working hardest for their consumers.

So why, when many brands are using sophisticated Data Management Platforms (DMP) and Content Management Systems (CMS) aren’t we hearing more marketers and technology product managers shouting about the benefits of personalized experiences?  Well, the truth is, not many companies have fully integrated the disparate elements of their content ecosystem to deliver the consistent and meaningful personalization that 63% of today’s users expect from their brands

Until recently, the challenge of coordinating globally distributed teams across clients and agencies and securing the right level of resource has made it impossible, impractical or unaffordable to undertake the heavy lifting required to make real headway. Today, emerging technologies and machine learning (like Samsung Whisk’s deep learning based natural language processing) make it possible to not only achieve what companies have dreamed of for years, but to do it more consistently, more accurately and more affordably.

At Whisk we have identified 5 key steps to building a personalized content ecosystem:

  1. Understand User Needs
  2. Analyze and Tag Recipe Content
  3. Integrate with your CMS
  4. Personalize Content Delivery
  5. Optimize with Purchase Behaviour Data

For traditional teams and processes, this could take years and cost millions to achieve, but Whisk’s plug and play technology accelerates the entire process, taking only weeks to implement. 

Personalize food experiences with Whisk

1. Understand User Needs

Today there are vast quantities of published recipes available for users to search, with more created every day. The value of creating a new recipe is small, especially if the user can’t find it in all the noise in their social feed. However, finding one perfect recipe at the very moment a user wants to see it is valuable and is the foundation of consumer loyalty for any brand or platform. 

To do this, you have to start by understanding all the possible factors that influence a consumer’s choice and then identify which factors matter most to that consumer in the decision-making moment. Only once all the rich and relevant data feeds have been connected can all possible factors driving preference at a given moment be understood. At this point, Whisk’s artificial intelligence technology can determine from all the data it sees which has the greatest influence on how relevant an individual will find a piece of food content at a particular moment in time. 

These are some of the key data signals that Whisk has identified as driving brand choice and that feed into our recommendation engine. 

food content personalization

Samsung’s Whisk team identified three factors that influence what an individual is going to cook from available data sources.

Context

  • Is it grilling weather or more about cozying up with a soothing bowl of soup?
  • What’s in your kitchen right now and what is about to go out of date?
  • What deals are available at your local store?

Preferences 

  • What diet are you following, e.g., keto, paleo, vegan, diabetes-friendly, etc.?
  • What are your favorite dishes, flavors and cuisines, e.g., Indian, Italian, spicy, etc.? 
  • Do you have food allergies and avoidances, e.g., gluten-free, seafood, etc?
  • Are there particular ingredients you love or hate, e.g., sprouts, blue cheese, mushrooms, etc?
  • How many people do you normally cook for? Are you single or do you have a big family?
  • What is your weekly budget? 

Behavior

  • What recipes have you viewed, liked, shared or shopped?
  • What products or brands do you tend to buy?

Over time, by using and connecting data sources, an organization may build both a detailed picture of its customers as well as a nuanced understanding of what influences a consumer to cook a particular dish. 

2. Analyze and Tag Recipe Content

Once an organization identifies what matters to their consumers, it’s important to understand which components should be highlighted for each consumer, in each moment. 

This is where well-structured content with accurate, relevant and consistent labels added as metadata becomes really important for 3 reasons:

  • Google rewards rich content in ranking for SEO
  • It’s how we understand the kind of content people are interacting and engaging with. For example, it’s hard to know if a pasta recipe is vegan without an accurate label to show it. Metadata allows us to build a picture of user preferences.
  • It’s how your CMS or Whisk’s AI recommendation engine knows the right content to provide to a particular user. 

For someone starting from scratch without any of their own proprietary recipe or food content, licensing content from a provider like Whisk is a quick way to access thousands of high quality, well-structured recipes from the start. However, many of our enterprise clients have masses of old legacy content that is poorly structured or out of date. This can be one of the biggest obstacles to providing personalized experiences for their users. 

When you have thousands of recipes and images amassed over many years and websites, it requires significant work to upgrade every recipe. Yet, it’s increasingly important for recipes to provide robust, complex culinary, ingredient and nutritional information as search engines, such as Google, reward this rich content with higher search rankings and it is key to recommending relevant content to users.

Getting the content up to date is only half the task as you then need to maintain it. Keeping complex content fresh and current, particularly around changing nutritional calculations and diet trends, is a full-time commitment. There was a time in 2016 when no one had really heard of keto but for legacy content, it’s a big job to go back and understand which recipes are or aren’t keto. Applying consistent rules, tags and definitions across many markets and teams is a logistical challenge but one that Whisk’s technology is made for.

Today, a nutritionist may go into a recipe, calculate micro and macronutrients and apply all the relevant tags in a CMS, however, one recipe may take an hour or more to update. Additionally, if values or rules or data changes, such as the normalized ingredient nutritional values from the USDA (which happens every couple of years), the nutritionist then has to recalculate the value of each recipe.

When looking at these costs, an organization with 10,000+ recipes could be looking at costs of over $250k just to update their content and that doesn’t even include maintenance costs.  

Technologies, such as Samsung’s Whisk, are now able to do this work for global clients in a fraction of the time and cost. With a team dedicated to continually updating and improving the core database and calculations for all, marketers are able to focus on the creative, innovative work that they specialize in while adding more value to their organization’s core business.

3. Integrate with your CMS and content team

There are a lot of very smart CMS platforms out there that enterprises have taken a long time to select and develop for exactly their needs and the last thing anyone wants is to have to completely re-architect their entire technology platform and retrain their team.

Modern solutions allow various options to integrate with leading CMS systems, from WordPress to Sitecore or Magento. 

Some may use a batch approach to update new and delete old content updating .XML files every night, week or month. Others prefer a more robust integration using a plug-in API solution to deliver personalized content in real-time as a user moves through different pages on the website but either way you get the best content delivered to your CMS when you need it.

At the same time, our clients have content teams and nutritionists inputting new recipe content or editing recipes into basic CMS screens that take time and require a lot of manual effort to get content into shape. 

Whisk clients prefer to use the clean front-end experience in our Content Platform to pull up automated tag recommendations and nutritional values in real-time, allowing users to edit and tweak them if they like before uploading.

This simple step reduces the time it takes to manage content, gets richer and more consistent results and frees up the content team to create new content rather than administrative tasks. Let’s be honest, the reason why the content is in such a mess in the first place is because no one likes doing this kind of grunt work. 

Very quickly, this allows you to create and share high quality, always up to date, structured content across your enterprise and therefore benefit from increased organic traffic, higher engagement and the ability to personalize experiences for your users in any channel. 

As well as enhancing content engagement platforms, grocers and occasionally brands managing e-commerce solutions can benefit from integrating recipe content into their e-commerce sites. Simple plug and play SDKs are easy to integrate into the front-end of e-commerce experiences without interfering with the highly complex and often creaking infrastructure of e-grocery platforms. Once in place, they can power recipe recommendations based on a user’s tastes or the content of their cart, recommend items to add to cart to complete new recipes, or recommend perfect accompaniments for favorite dishes.

Together these are powerful ways to grow cart value and customer loyalty by adding inspiration and utility to the often boring shopper journey so focused on price and promotions.

4. Personalize Content Delivery

With your user needs and motivations well understood, your content in great shape and your owned platforms firing on all cylinders to attract and engage users, the next step is to work out how you will deliver that content to your users for greatest impact.

You may want to take the first step by amassing user data from first-party sources and internal research documents and then add third party data collected from interactions across multiple sites to segment your audience and define actionable personas.

Once armed with these personas you can start to set rules for how you want to deliver content to them and then manually optimize and tweak the way content is delivered as you better understand user engagement with different types of content in your CMS.

Many CMS solutions offer AI technologies that can over time start to automate these recommendations for your user. Most of these are generic technologies designed to work for a wide variety of customer verticals from car manufacturers to cosmetics brands and beyond, so having all that rich recipe metadata gives these engines something to get their teeth into.

When Samsung was building intelligent recommendation engines to power their connected kitchen devices, such as the Family Hub Fridge Freezer, they realized that food is like almost no other category in how the ‘perfect recipe’ for one person changes depending on the weather, on store offers, or on what’s in their fridge. That’s why Samsung acquired Whisk and now both the Samsung Research AI team and the Whisk team of data scientists work together to continuously strengthen Whisk’s AI recommendation algorithms and that help clients create smarter personalized experiences. 

All of this intelligence and rich data capabilities are available in the Whisk Content Platform which can power any CMS and help distribute content and personalize experiences across all channels 

Combine this understanding of content with Whisk’s reach across hundreds of millions of unique users and 500 million monthly recipe interactions from many of the world’s most popular recipe publishers and Whisk can offer the most accurate and robust food personalization platform in the world.

In future, we want to make all this data available through identity resolution services like Liveramp to understand millions of user preferences and use that knowledge to deliver valuable food recommendations programmatically.

Our partners can use this resource to power paid advertising campaigns through their DMP, personalize website experiences and serve more relevant content through social media channels. This results in better ROI on recipe content and more engaged, loyal users growing your LTV and lowering CAC.

5. Optimize with Purchase Behaviour Data

At any time in the process, our clients can add shoppability to their content and platforms, but until you have the four earlier steps in place, the real value of shoppable content won’t be fully realized.

Once the content is understood and being effectively delivered to users, the value of knowing which content drives specific users to a shopping action is priceless.

Once a DMP is able to connect specific content to specific user purchase actions, it allows advertisers and content creators to really understand what drives results, with which shoppers. This is the kind of hard data point that really ratchets up ROI as you apply it across lookalike audiences and produce less content with more impact.

Using one provider to make your website’s recipe and product pages shoppable online and in-store is the best way to build a consistent understanding of your user engagement and optimize your media and content budgets. 

While Shoppable Recipes make web pages shoppable, Shoppable Media links can make almost any piece of content shoppable in any channel – which increases the dataset dramatically. To really reach hundreds of millions of users, leveraging Whisk’s Sponsored Products across our recipe publisher partners can help build a formidable set of data to inform incredibly effective campaign and content strategies.

If you’d like to find out more about how Samsung’s Whisk can accelerate your marketing effectiveness, contact me at stuart@samsungnext.com.