Saturday, May 14, 2011

Birding App Study - Advanced Search Discussion

Something interesting to come out of our focus group relates to the usefulness of the advanced search tools in the apps.

I presented the participants ten photos of birds, alternating between easily recognizable and challenging species.  I asked them to pretend that they didn't know what the bird was and to write down three to five characteristics of the bird.  Then by entering some of these characteristics into the search tool, it was our hope to be able to find these "unknown" species within the app, just like a beginning birder relying on the app for help.

Of the ten species presented, only a couple species were were found on any of the apps using these filter search tools.  It was pretty frustration for the participants.  It is my belief that there is a fundamental flaw in the programing for these so called advanced or smart search features.  Birding apps to help people identify birds should use "probability search" programming rather than "filter searches".

Let me explain.  Filters narrow down the possibilities.  Each time you enter a new filter item, it further narrows the possibilities.  What we ran into in our focus group is that participants filtered themselves down to zero possibilities very quickly.  The problem is that we all have a slightly different opinions about certain bird features, but the apps are designed with one designation for each bird species, so if you enter the "wrong" characteristic, your bird will no longer be among the filtered search results.

Probability search programming will show users something like this:  "Of the 5 characteristics you entered, the following birds match 5 of 5 or 100% [or 4 of 5 (80%)...or 3 of 5 (60%), etc.]"  This way, if a birding app user enters one attribute incorrectly, the bird they are trying to identify will at least still be on the list somewhere.  As they add more "correct" characteristics of the bird, the correct bird will rise on the probability search list, eventually revealing the correct identity of the bird to the app user.

That being said, the existing filter search features can be used more effectively with some training and experience.  Most apps allow you to select more than one characteristic under each category.  For example, if you aren't quite certain of the size, you can select a couple size options and it should yield more possible birds than if you selected only one size.  The same goes for color and other attributes.

App developers, I invite you to weigh in here.  This is all about helping the app developers help the birders who use their products.


  1. Search, as implemented in most birding apps, is absolutely the wrong approach for beginning birders. As you found, the temptation is to give the app far too much information. It only leads to frustration when the app doesn't return the expected results.

    There is a well recognized — but unfortunately named — approach to identification that is much better suited to beginning birders. Jizz — bastardized acronym for generalized size and shape - advocates getting a quick general impression of a bird that provides a starting point for further investigation. For example, it's chicken-like or it's sort of like a thrush.

    Houghton Mifflin's Peterson Field Guide to Birds of North America, and it's East and West sub-species, implement this jizz approach in the way the books are organized. With a little familiarity, a beginning birder can quickly orient themselves at roughly the right place in the book, and then flick through pages of illustrations to find a group of birds that looks similar to the bird they observed.

    When we designed the Peterson Birds of North America app, we kept the organization of the Peterson book intact, knowing that years of expertise had gone into making this the best way to get a positive identification. The first thing you see when you open the Peterson Birds app is a visual representation of groups of similar birds. Same as the book. You can then quickly flick through virtual pages to get to a screen with likely looking birds. The Peterson ID system arrows on each illustration show important field marks that a beginning birder can focus on, and you have quick access to range maps and songs that can help further narrow down candidate birds.

    Our approach to aiding identification is radically different to that of other apps. We use the computer's search engine to deliver screens of information fast — because that's what it's good at — and we use the user's eyes and brain to finish the identification — because that's what they're good at.

    I'm not saying search has no place in a birding app. It's great for filtering birds by location or for experienced birders looking up specific characteristics when they know how the search works. But for beginning birders, search is definitely the wrong way to identify a bird.

    Nigel Hall
    Appweavers, Inc.

  2. @Nigel - thanks for sharing this helpful information!