Henry Ford famously said "any customer can have a car painted any colour that he wants so long as it is black." Apparently toothpaste operates on the same principle. You can buy toothpaste in any flavour that you want, so long as it's mint.
I combed through the Crest and Colgate websites and tried to record every flavour of toothpaste. The dataset is available here.
Common Flavours
These are the flavours that are used by both Crest and Colgate:
- mint
- peppermint
- spearmint
- clean mint
- cool mint
- fresh mint
- effervescent mint
However, the majority of flavours are unique to one brand or another1.
Flavour Visualization
I found around 50 flavours in total. To analyze them, I created vector representations2 and then projected them to 2D using Uniform Manifold Approximation and Projection (UMAP).
Automatically Generating New Toothpaste Flavours
Since all the good3 mint flavour names are taken, we'll need to use NLP magic to generate some new ones. We can do this by taking the vector representation of each of the above flavour names4, averaging these to produce an "average mint flavour" vector, and then searching a list of phrases of the form "adjective mint" to find out which phrases are most similar to the average mint flavour5.
The resulting top 30 new toothpaste flavours are:
- minty mint
- sweet mint
- warm mint
- orange mint
- fine mint
- nice mint
- perfect mint
- kind mint
- some mint
- delightful mint
- bright mint
- good mint
- terrific mint
- even mint
- delectable mint
- delicious mint
- marvelous mint
- wonderful mint
- silky mint
- fantastic mint
- agreeable mint
- pleasant mint
- palatable mint
- little mint
- hearty mint
- tasty mint
- subtle mint
- mellow mint
- neat mint
- great mint
I'm not too sure about "palatable mint" and "some mint"6, but "silky mint" and "subtle mint" seem very plausible. And based on some brief googling, the flavours in bold all actually exist in the wild.
In case you were wondering, here are the 30 least likely toothpaste flavours:
- quick-witted mint
- extroverted mint
- gregarious mint
- studious mint
- talkative mint
- criminal mint
- disloyal mint
- chief mint
- warlike mint
- corrupt mint
- far-flung mint
- married mint
- wide-eyed mint
- obese mint
- quarrelsome mint
- decimal mint
- illiterate mint
- defensive mint
- self-assured mint
- unlawful mint
- black-and-white mint
- ill-informed mint
- self-reliant mint
- long-term mint
- ill-fated mint
- high-level mint
- short-term mint
- well-to-do mint
- happy-go-lucky mint
- hard-to-find mint
- And several of them are trademarked, as though some competitor is going to take advantage of the significant goodwill towards the Frosty Mint™ name that Colgate has painstakingly built up. ↩
- Derived by averaging the word embeddings from spaCy's medium English model. ↩
- [citation needed] ↩
- I restricted this experiment to flavours ending in "mint." ↩
- I used cosine similarity, but other metrics are available. ↩
- Insert Charlotte's Web reference. ↩