Selling Software, Sexist Google Data, and What the Actual Hell, People?!

So, we've been messing around a lot with Google's A/B testing tools on our website lately. Like, a lot. This toolset allows us to "test-drive" alternative versions of our site for a select audience of live customers, and see which option puts more money in the bank. I like A/B testing. It paid for our new margarita machine, so what's not to like? It's intellectually intriguing, too, because sometimes very subtle changes in copy, or images can have an impact on how users interact with the site. For instance: people like pictures of people, not product boxes. So we've learned a lot, which has also been really cool.

And also sort of depressing.

You can't buy software from women, no siree Bob!
Some of the many women who've unsuccessfully hawked our software.
None of you could even get a date with any of these fetching ladies,
but you won't buy software from them? What the hell?
Here's one of those depressing facts: men don't like buying software from women. In the course of several months' testing, we've been comparing how images of men and women perform when placed on commerce-related pages on our site. Thus far, images of men have universally out-performed those of women. Two of the experiments showed differences in conversion rates of 20% and 45%. Granted, the sample set wasn't super huge, but still: that's a big gap.

We've gone for any number of different female looks, from downright beautiful, to tomboy-ish, to sassy, to cute. Blondes, redheads, Asians: you name it. And no matter what we do, our predominately 25-45 male clientele won't buy from women the same way they buy from men.

This bugs the crap out of me. It really does. Where is it written that females are somehow less technically savvy, or less qualified to make recommendations regarding a technology purchase?

When I think of competent technologists in my own life, my younger sister, a data warehousing consultant, comes prominently to mind. She chews through SQL queries like taffy, and can extract, transform, and load that damned data like nobody's business. Or my highschool friend Dima ElIssa, a serial technology entrepreneur, and one of the smartest, savviest people I've ever met. Or my other highschool friend Mary Ann Clark, who is just flat-out brilliant, and rightly reigns as the preeminent goddess of crunching ginormous healthcare-related datasets. I'd cheerfully accept tech advice from any of them over the random blueshirt I get at my local BestBuy, for instance.

Now, maybe I should be happy, in that we've apparently found some sort of exception to the universal trope that "Sex Sells." But it doesn't feel like that. My cohort in crime, Jana, very carefully selected these images (both women and men) to be attractive, but not blatantly "sexy." No bikinis, no sideboob, no six pack abs: just reasonably attractive people of both genders trying to sell you software. But the women apparently aren't allowed to do that.

Hello, I'm a random guy you are comfortable buying software from...
Hello, I'm a random guy you are apparently comfortable buying software from.
What galls me is that I'm pretty certain this bias is unconscious. I'm convinced that the majority of our male customers would never describe themselves as being sexist. They probably don't look at an image of a woman on our web site and consciously say to themselves, "Oh no: I obviously cannot buy this software, because an attractive woman is offering to sell it to me."

But subconsciously we're losing a consistent percentage of those male customers. Maybe they were on the fence anyway. But for whatever reason, they decided to abandon the purchase. And the only data point we have as to why they did so, is that they also saw a picture of a woman on, say, our products page. The numbers don't lie. Something significant is going on here.

And that, in turn, makes me wonder: do I unconsciously evince the same behavior, and not even know it? If I went to another software company's web site, and they A/B tested me, would I be one of those people that abandoned the sale, maybe rationalizing it on the basis of some missing functionality in the software, but actually doing it because at some gut level I just don't feel comfortable buying from the slightly geeky Asian gal on the checkout page? (Just for the record, that would never be true, at least in my case.) But you get the idea. Just how deeply are we all affected, at some unconscious level, by the societal factors which shape our understanding of who is allowed to do what in the world?

Who knows, maybe this current bias against women in technology is actually beginning to correct itself. Maybe if we'd had these same analytical tools back in 2002, the numbers would have been even worse. On the other hand, I have to wonder how long this is going to take to correct itself? Are we talking a few years? A few decades? A few generations? Or maybe never?


About Jon Parshall
Jon has been working in the computer software industry for over 20 years. He joined CodeWeavers in 2002 as COO. Contact Jon at and learn more about his professional accomplishments on LinkedIn.

About CodeWeavers
Founded in 1996 as a general software consultancy, CodeWeavers focuses on the development of Wine – the core technology found in all of its CrossOver products. The company's goal is to bring expanded market opportunities for Windows software developers by making it easier, faster and more painless to port Windows software to Mac and Linux. CodeWeavers is recognized as a leader in open-source Windows porting technology, and maintains development offices in Minnesota, the United Kingdom and elsewhere around the world. The company is privately held.

The following comments are owned by whoever posted them. We are not responsible for them in any way.

New Comment

Ryan V. Bissell
Posted 2015-06-03 11:20
Hi, cis-male here.   Choice of model isn't going to factor into my decision to purchase software-- I have to have an actual reason to consider buying it in the first place-- but the female models will certainly hold my attention longer.  Hence my surprise (and I guess, everyone else's.)
But, I notice that of all the models in the article, only the male is presenting a computer with a visible display that is ostensibly showing the product.  I feel like he is showcasing the product, and they (the women) are demo'ing their laptops.  Maybe that wasn't consistently true throughout your campaign, but it is true of what you've represented here in this article.
It might also be that holding my attention on the model is counter-productive because it is actually a distraction for me; my thoughts are-- at least for a second or two-- on subject matter not related to your product.  Maybe that is just enough time for contrary thoughts to override any buying impulse I might have otherwise had.
Still: this is an interesting observation, thank you for sharing.
Lasse Nielsen
Posted 2015-06-04 04:16
For me, it's not that I don't trust females to make recommendations, but that I don't trust companies that uses pretty women to make me want to buy their software. And yes, all the female models were pretty, young, smiling women.
Now, as a male, I might be blind to the male model being equally pretty to the right eyes, but I don't feel manipulated by his image.

That said, I can obviously not rule out that an unconscious bias is also at work.
Todd Vierling
Posted 2015-07-09 22:39
This data point disturbs me as well, and I don't think it's just an aversion reaction to the "sex sells" trope.

It feels more like a reinforcement of known hostile attitudes towards women in IT -- and in STEM fields in general. This is a late-20th-century work environment attitude that can partly be traced back to the flood of returning men from WWII, basically pushing their way back into jobs which had been filled by women while the men were off at the front lines. During the war and for about a decade after, the U.S. and UK had a sizable proportion of women in technical jobs, but it all faded quite rapidly after that.

Grace Hopper in particular comes to mind. She pioneered high level programming languages at a time when no such thing existed. We wouldn't have seen the rapid rise of COBOL, much less Algol, FORTRAN, BCPL, B, C, or their descendants without many of the concepts she invented. During the war and for some time afterwards, "computer operators" -- those who made and ran punch card decks or tape spools, and often also maintained and repaired the systems -- were actually women, not men. They were the ones who really understood the hardware the best.

Today, I work for a large, multinational software and hardware company, and despite a lot of its progressive efforts to champion the need for women (for that matter, diversity in many facets) in our field, there are just not enough women in that company's technical divisions today, a direct mirror of the larger IT world.

Worse, there are still women who work very hard to make their own mark in IT, and end up being harassed (in many cases dangerously so) for their efforts. Not just the abhorrent "brogrammer" culture, mind you, but truly documented incidents which have been getting more visible in recent years. There are men who really are intimidated by the concept of intelligent and strong women, and lash out in reaction.

Worth a read, will probably get you angry, but it's important to know what's out there:

I have seen signs of these social attitudes finally changing in those who are just coming out of high school and college about now, and that's a good thing. But we have a long way to go still.

Anyway, this particular post's issue in context: The one data point that would really make an A/B test like this stand out would be a way to correlate the gender identity of the visitor with the conversion rate for a given test image. I have a sinking feeling that the subset male visitors would have an even wider divide between the two test cases. And that's what really pains me.

(BTW, I don't know why this post took almost a month to show up in the blog's RSS feed, but I just saw it a couple days ago.)
Dominik Reichardt
Posted 2017-11-30 19:16
I do wonder what's up with that. Personally I do have a problem with ANY of those pictures including the male one.
A site using person being shown like that to advertise for software looks to me like a bad site selling their awful softwarewith strange pictures. I do associate those sites with software that has some adware and toolbars bundled with it.
These all look like models selected to push sales not like people using crossover (and no, using pictures of users like me is more than likely not going to push sales, either).
And using young attractive women, with a bit of cleavage and/or a sexy post makes it even more dubious for me.
Didn't stop me from buying a lifetime licensebut I did that as a longtime user of Crossover in need of renewing my license.

And I have to admit that there is the possibility that unsubconscios I do think that women are less likely to be using Crossover because of stupid sexist things being hammered into me over the years. I can not be sure that this is not the case. I hope it's not, though.

I'm not in advertising so I'm certainly not an expert but using those pictures of people makes software sites  unappealing to me.
Please Wait...