8 Reasons Recipe Creators Are Embracing Artificial Intelligence
If your first thought, when you see artificial intelligence and recipes in the same sentence, is a funny recipe written by AI, don’t dismiss the value that AI can create for existing recipe content.
Though it might still not be ready to write recipes better than humans do, AI has been doing a great job in understanding and reading recipes as well as enriching existing content with relevant tags and labels.
Is Your Recipe Content Getting Noticed?
With millions of recipes to choose from and more being published every day, just writing great recipe content won’t help users to answer the question “what shall I cook tonight?”
Any user typing “what to cook” in the search bar might want to give up and go back to the safety of their favourite six or seven recipes they cook on repeat day in and day out as that particular search term on Google returns a whopping 1 730 000 000 search results.
Even typing “chicken recipes” will give you 944 000 000 results to choose from.
Yet it seems people, and especially Millennials, are hungry for inspiration when you add 15B food & drink pins on Pinterest and 310M Instagram #food posts on Instagram to the Google search numbers – let alone Facebook.
This proliferation of great content becomes challenging for anyone who publishes recipes as it is extremely difficult to make your recipes stand out and engage every unique visitor.
Recipe creators are saving time to get more creative while making the most out of existing recipes by using machine learning technology to automate the slow tedious tasks.
How AI is Making Recipe Content Smarter
1. Automatically calculating nutritional values for any recipe
Eating is not only about what tastes good anymore, but more and more people are interested in ‘well-being’ and thinking about the impact of food on their health. A report from Label Insight and the Food Marketing Institute showed that 75% of shoppers will switch to a brand that provides more in-depth product information.
All the data that Whisk uses to feed our ontology – the Food GenomeTM – is based on government databases in the US, Europe and beyond and the automated calculations free up publishers and nutritionists from the time-consuming need to calculate this data manually.
Showing macro-nutrient information such as energy, fat or sugars as well as micro-nutrients such as minerals or vitamins is helping users choose the best recipe – the one that not only tastes good, but that will help them reach their health goals.
It’s also great content that Google’s spiders like to see and can help grow session duration and engagement further helping your SEO performance.
2. Showing additional nutrition labels
Showing additional nutrition labels is a great way to add value to your recipe. And the best part is, all of the data is provided automatically thanks to Whisk’s deep learning based machine learning and smart algorithms, so you don’t have to calculate any of it manually.
3. Making it easy to understand nutritional information with a health score
Whisk uses deep-learning based NLP to map extensive nutritional datasets from the USDA and EU sources onto our Food GenomeTM food ontology. This dataset allows us to understand the nutritional composition of millions of ingredients and recipes in detail and to use it to create smarter food experiences.
Whisk has taken USDA and NHS definitions of “Healthy Eating” to calculate a simple health score for every recipe. This looks at 6 “good” macro-nutrients and 5 “bad” macronutrients and uses them to understand their density per 100 calories of food consumed. The higher the score the more “good” nutrients per 100 calories of food and the fewer “bad” nutrients so it’s easy to find healthier recipes that taste delicious.
Whisk also calculates the Glycemic Index and Glycemic Load values of recipes which we use in combination with the Health Score to tag healthy recipes as “Diabetes-Friendly” to help people find recipes that are tailored to their needs.
4. Normalising ingredients to grams
In a blog post where we share our advice on how to write tech-friendly recipe content, we addressed the importance of keeping your recipes consistent in terms of measures – for example, if your ingredients are measured in grams, don’t talk about oz all of a sudden. But, people are not used to seeing 100g of apple in a recipe as they would usually better understand if you referred to that part of the recipe as 1 small apple.
The way normalisation works is it provides standard measures and translates those to the measures AI can understand. For example, if a recipe contains the measure of ‘1 cup of flour’, we calculate grams based on the volume of a cup and the density of that particular ingredient. In this case, it would translate to 125 g but a cup of water would have a completely different weight.
Normalising ingredient measures allows us to calculate nutritional values accurately and match those ingredients properly to grocery store items – which we’ll talk about below.
5. Scalable recipe portions
One of the issues users find with recipe content is that servings are not always tailored to their needs. While most recipes serve 4 most occasions might call for different portion sizes, for example, if you are a couple, a large family or have friends coming to dinner.
Once recipe ingredients are normalised and have been ingested into a well-structured ontology, it’s then possible to allow users to scale the recipe servings and see the measures scale with it.
With one third of the world’s food going to waste every year and ever-growing food prices, it’s a simple way to be more efficient – and to eat more healthily.
6. Matching ingredients to store items
Creating a frictionless journey from inspiration to purchase is one of the best ways for recipe publishers to build an engaged and loyal audience.
The way this is made possible with artificial intelligence is that Whisk uses AI to extract the product, brand, attributes and quantities of items in the text and matches them to a list of store products at grocery retailers.
7. Using labels to help users find a recipe they need
Labels are an important part of any recipe as they help the user quickly scan, search and find a relevant recipe for their particular need. Labels are useful for:
- enriching your recipe content and helping visitors quickly scan recipes for their prefered dietary needs
- improving your search by adding more filters
- creating consistency in labelling recipes with predictive machine learning
- serving personalised recipes to visitors
- boosting a particular recipe you’d like to promote
It’s also something that Google likes to see and will help your SEO.
Manually tagging recipes with relevant tags and labels can be a time-consuming process.
Help your users understand whether your recipe fits into their preferred diet style, for example, – whether it’s vegan or Mediterranean, by showing an easy-to-spot label on your recipe, you’re making your recipe stand out.
By teaching AI to understand recipes and predict which labels each recipe should contain, recipe publishers are saving time and offering more value to their users.
8. Making sure allergies and avoidances are quickly seen
Avoidances are specific items that someone restricts from their diet. Use this parameter to return recipes that exclude one or more of the following avoidances.
People are more and more reacting individually to certain ingredients and are looking for recipes that are tailored to their own preferences, they want an easy way to go from finding a recipe to cooking without having to look for alternative ingredients to remove those they need to avoid
Easily show allergens on your recipes automatically and serve personalised recipes to your visitors.
As the number of recipes on a particular website increases, it becomes impractical to manually analyze each one to add tags, calculate nutrition, measure multiple serving options, etc. but without that kind of data, it’s very hard and expensive to get your content to surface on people’s newsfeeds or search results.
Smart recipe creators are harnessing the power of AI to accelerate their content development and free them up to do the creative work they love.
Once all the data is in good shape, the potential for recipe content to be used to power shopping lists, create personalised recommendations and even power the connected kitchen is vast.
Thanks to AI and machine learning, you can make your content future-ready, attract eyeballs and advertising revenue and help users take the friction out of cooking new, tastier and more healthy dishes.