Amanda Hesser has a piece on Google recipe searches (approximately a billion searches a month) and the resulting impact of those Google results on how people are cooking today. As discussed previously, users overwhelmingly stick to the first page of search results, and thus, for many users, what is returned on the first page of results is what matters. According to Hesser, the higher results in Google are those sites with a lot of metadata on ratings, calories, cooking times, and pictures. Hesser says this hurts the smaller cooking blogs and cooking sites, which may have the better recipes, in favor of larger sites or content factories that play the SEO game better. Hesser suggests that the best approach would be seeing which recipes generate the most comments related to page views, have the most FB likes, and are shared the most. This sounds right, although I am not quite sure how Hesser knows this is not incorporated into the search algorithm.
Hesser’s piece got my attention. Recipes found on the Internet have been at the heart of my learning to cook and then expanding my recipe base over the last 12 or so years. My process is as follows. Having a dish in mind that I would like to learn to cook, I will look at a number of recipes online, and by comparing and contrasting, along with my own knowledge of cooking and others’ comments, I will make a judgment on what looks like the best recipe.
Sometimes this is on AllRecipes, Epicurus, or the Food Network; sometimes it is on a random blog. My favorite go-to recipe is found on a decade old personal web page of apparently a former CS student at UVA that has nothing to do with food other than a couple of recipes. Even if I search directly for a recipe for “pav bhaji,” it will not appear until the 3rd page of search results. Given this, I doubt many people are finding this recipe, and there is probably virtually no chance that someone will stumble onto this recipe in a more serendipitous fashion (i.e. when just looking for something great to cook as you might when browsing one of the major recipe sites).
The other thing I do a lot is tweaking recipes. If I look at ten recipes, I may hone in on one recipe, but I may find something interesting in a different recipe (particularly if I find the same ingredient in several recipes) that I may use to adjust the recipe that I have chosen.
My takeaway is that recipes are an example of where some other method than search may yield better results. As Hesser says “the most relevant recipe is the best recipe” rather than having to do anything with associated metadata that may or may not appear on a site. Given the number of searches for recipes, this seems like an area that is bound to be a target of the content mills compromising the efficiency of search.
In a better approach, I would love something that helps make the culling and sorting process more efficient, and perhaps does something similar in its algorithm to return better recipes, whether for a particular dish the better recipe is on a major web site or on a random personal page. I also would love something that can suggest possible tweaks to a recipe that I could consider. I am sure there are a number of other such recipe-specific features that would be helpful. Clearly a tasty opportunity here for someone!