Share this project

Done

Share this project

Done
I'm writing a detective story set halfway between San Francisco and the internet. And the more people who reserve a copy, the better each one will be!
570 backers pledged $13,942 to help bring this project to life.

Naming characters with Google AdWords

Posted by Robin Sloan (Creator)

Here's a new trick.

In this book, I'm trying to craft a central character with some of that same iconic strangeness that makes Sherlock Holmes so appealing. There's a lot that goes into that, but for now, focus on the name. Sherlock Holmes. It leaves an indelible mark on the brain.

So, I have a name in mind for this character, and I was looking for a meaningful way to test it out—without giving it away.

That's where AdWords comes in.

Here's what I did:
Created a campaign attached to a bundle of search terms: mystery, detective story, sherlock holmes, noir, and more like those.
Came up with a whole set of names, basically wide variations on a theme. One was my original pick, but I liked all of them. Then, I created an ad for each one, all with the same body text but each with a different name swapped in for the headline.
Allocated a small budget ($40, to be exact) and kicked off the campaign. And wow there are a lot of people searching for stuff on Google. Over the span of 24 hours, my ads made about 100,000 impressions.

So the question—and I do think it's a serious question, insofar as it's a simulation of a decision that will confront many potential buyers of this book—the question is, which name worked?

The results, pixelated for secrecy's sake:
Here's the way I read this: The four names at the top all did about the same. I wouldn't choose a name with an 0.23% click-through rate over a name with an 0.20% just because of that measly 0.03 margin.

But the 0.07% at the bottom? I think there's real signal there. As it happens, the name at 0.07% was one I really liked—but it didn't make the cut. Alas.

My original idea—the name I came into the exercise with—is the one at 0.21%. So basically, I see this as validation: The name works. People don't see it and go "ew" or "meh."

But okay, I'll be honest. This was mostly just an excuse to try a new tool. Any nerd will tell you that tools can provide their own intrinsic rewards. There's an aspect of exploration to it, too: you're pressing out into new tool-territory, learning about what you can and can't do.

This little AdWords test is a first step. Mechanical Turk might be next. I mean, imagine—this is the sci-fi extrapolation—imagine highlighting a block of text, choosing a menu item called Test the way you'd choose Spellcheck today, and when you do, a little timer appears next to it. Five minutes later, ding—the timer goes off and you have the results right there, floating over the text. Aggregated feedback from an anonymous swarm of readers: "I stumbled here," "this variation works better," "this line rings false."

That might sound naive—it's definitely oversimplified—but I think there might be something useful lurking in this particular tool-territory.
So, finally, here's the irony: I'm making a big deal out of keeping this name secret. Functionally, it is secret—none of you know what it is yet! And yet... 100,000 people out there have laid eyes on it. Thousands of Google searchers have seen her name. What kind of secret is that?

Ah, liquidity. Ah, scale.

Sometimes, the vast sea of clicks can be a comfort.

Comments

Only backers can post comments. Log In
    1. Saheli on

      I just wanted to say that I don't think 0.07% should be villainous, but should be a misunderstood last-minute heroine, like Sydney Carlton.

    2. Missing avatar

      tim inman on

      This is a great idea, but it would be much more expensive to do it well. As an Adwords customer, I know you aren't getting a very big sample for forty bucks. Don't rule out your favorite name over maybe a dozen or two clicks difference.

    3. Robin Sloan Creator on

      @Stephanie: Good question. It was a placeholder page that said essentially "Hey, thanks -- you just helped me out with my writing, even though you didn't realize it." Then a link to a short story.

      @Jeff: Love that! Yes, Mechanical Turk is next.

    4. Jeff Barr on

      I'd suggest Mechanical Turk (even if I didn't work at Amazon).

      A few years ago one of my colleagues was trying to come up with a creative yet descriptive name for an internal accounting system. I asked Turk to suggest some first names that would be appropriate for an accountant. I got 100 good answers in a very short time and we ended up choosing "Penny."

    5. Stephanie Rosenblatt on

      My question is, when people clicked through the ad, what did they see on the page they landed on?

    6. Jimmy Stamp on

      I'm once again blown away by your use of technology. This is an amazing project!

    7. Missing avatar

      Kevin Meyer on

      We can only hope Carmen Sherlock Holmes-Diego use such innovative detective techniques in the book!

    8. Robin Sloan Creator on

      @Lionfire: Then again, maybe the whole thing is a ruse... ;-)

    9. Robin Sloan Creator on

      @Lionfire: Someone else pointed out the same thing. Shhh. (Lots of digital detectives out there, it turns out.) And yeah, I agree w/ you re: not over-interpreting the results. I think it's a good data point to have, though.

    10. Lionfire on

      No one knows any of those names. No one except, of course, those detectives who recognised the that you left a clue or two, and that Google caches its own ads.

      Perhaps a little more secrecy is required, but the idea has some merit. I'd be a little hesitant to assume that click-through implies a like or dislike of the name. Perhaps it just means that the name is familiar, and therefore possibly too similar to another character?

    11. Lily Sloan on

      I agree with Gavin. .07% = villainous

    12. Gavin Craig on

      So are you going to make the 0.07% name a villain? I think you really need to use that name somehow, especially if it was your favorite.

    13. Robin Sloan Creator on

      @Burt Yes! That's what I'm talkin about. It's like design thinking for fiction. Maybe this whole thing is just a prototype.

    14. Burt Herman on

      Awesome use of adwords, go user testing!!!

    15. Andy Baio on

      Clever. Let me know if you need MTurk help.

    16. Jack Cheng on

      I see what's going on here. This is the old bait and switch, right? Make us think it's a sci-fi book when It's really going to be a book about how write and market a book in the 21st century ;)

      But seriously, I'm loving the updates and that you're keeping the what's actually going to be in the book secret.

    17. Yancey Strickler
      Superbacker
      on

      great post, robin

    18. Missing avatar

      Dan Bouk on

      This is geek-tastic. Love it!

    19. Tim Carmody on

      It's the lesson of "The Purloined Letter"! The only way to truly hide something is to hide it in plain sight.