We’re always talking about how NomNomNow is different from your traditional dog food company. Even calling ourselves a dog food company feels a little confusing, when we look at how far options for feeding our dogs have come.
In order to make meal time as healthy and convenient as possible, you could say we’ve gotten it down to a science. But our engineer Wenzhe will be quick to clarify: what we’ve gotten it down to is an algorithm. A perfectly precise, sophisticated algorithm, that makes it possible for us to serve your dogs in a way that we couldn’t have dreamed of a few years ago. For those who want to nerd out a bit, here’s how we use technology to build an experience that can hardly be described as a “dog food company”.
Our baseline algorithm
When we began NomNomNow, we worked with our veterinary nutritionist Dr. Justin Shmalberg to create our baseline algorithm. This is the formula that turns the information in your dog’s profile into a precisely measured amount of food to be cooked and delivered each week.
Justin and Wenzhe worked together to bring the latest science of dog nutrition to the web for the benefit of all dogs. When you input your dog’s weight, age, and activity level on the NomNomNow online profile, this algorithm tells us that your dog should be eating x number of calories per meal on a two meals per day feeding schedule as recommended by Justin. It even calculates the precise supplements of our NomNomNutrient Mix, to make sure each recipe is appropriately balanced. Of course, two dogs that weigh the same, exercise the same, and are the same age, are rarely the same. So, we began adding additional layers to the algorithm.
By putting your dog’s current weight and goal weight into your profile, our algorithm can then rely on Justin’s recommendations for weight loss (or gain) programs. For example, if you input that your 35-pound dog should ideally weigh 30 pounds, the algorithm knows how to gradually (and healthily) decrease calories over time, to transition from what a 35-pound dog eats to what a 30-pound dog eats. Because Justin oversaw the creation of this, we can be sure that the calculation, fancy and high-tech as it may be, ultimately relies on Justin’s experience and initial input as a veterinary nutritionist with years of experience placing dogs on weight loss, weight gain, and weight maintenance plans.
Similarly, your dog’s birthday in the profile helps us determine your dog’s life stage, and serve them accordingly. If your dog is a puppy, you’ll notice that we feed them larger portions during the growth period, and graduate to the adult formula over time. Based on their breed and estimated adult weight, we can alter this from one puppy to another.
Allergies, conditions, picky eating, and other personal dog facts inputted in the ‘things we should know’ section can likewise be considered in what our algorithm determines is right for your pup. If you select our Tasty Turkey Fare recipe, but then input that your dog is unable to eat egg, you can anticipate an email from us with a more appropriate recommendation (our Turkey recipe includes eggs).
The result? By the time your dog’s order gets submitted to our kitchen, we know the exact number of calories that should be fed, and the exact number of grams that this translates to within each recipe. Portions, perfected.
On the kitchen’s end of things, the algorithm breaks down your dog’s order into the number of ingredients we will need. The number of additional grams of meats and vegetables to purchase are then added to a long list, and our team shops for them and cooks everything on the weekend. That way, we only buy and cook exactly what we need each week, allowing us to dynamically portion every meal for every dog at scale. There’s no waste, and no leftovers.
After cooking for all of our dogs, we use scales to portion each meal into an individual bag, to keep the line of precision going away from the computer as well.
Great as it is, we’re always changing
Like the dogs we feed, our model is always growing and changing. “We train our model to create the best personalization for each dog, by constantly refining the way we create each dog’s meal”, says Wenzhe. We leverage tools such as Google BigQuery to analyze large cross-sections of dogs, and glean insights based on breed, weight, and longer-term nutrition history. As we collect more and more pet data, learn about more of your dogs, and continually bounce this data between Wenzhe and Justin, we adapt our algorithm to improve with every new bit of information we gather.
And while it’s important to us to use this information to provide perfectly balanced and portioned meals for your dogs, another long-term goal of ours is to continue bringing these health and nutrition insights to the veterinary community. Because NomNomNow is able to track your dog’s caloric intake and health changes over time, we will provide a new way to deliver this information to your veterinarian, leading to insights into your dog’s health, and the health and nutrition of dogs as a whole. As our customer base of #nommers grows, we can only imagine the possibilities.
A world away from the pet food companies of the past
For decades, pet food companies have had largely one-way communication with the dogs they feed, and it’s up to the owner to do the research and determine which food is right, and just how much to serve. We’re getting rid of the one-size-fits-all model, because each dog deserves something special. By reversing the order and asking about your dog first, then directing how to appropriately feed through our algorithm and Dr. Shmalberg’s guidance, we create a completely new experience, while gathering valuable data greater than any pet company has been able to gather thus far. What this could mean for the veterinary community (and most immediately the general wellness of our beloved dogs) down the road is especially exciting, as we are able to track valuable details about a large number of dogs.
About Wenzhe Gao: Before becoming the NomNomNow engineering wizard that he is, he spent 9 years managing teams and building dev tools at Google.