Will Marlow - Filed under 'Software development'

March 6, 2010

Don't make assumptions

When it comes to building a website or any type of software, it's important to make as few assumptions about how you think people will use your product as possible. (This is especially important if you want to be taken seriously by veteran software developers and Internet entrepreneurs, who are very big on the concept of split testing everything (read about that here).

But what do I mean when I say that you shouldn't make assumptions about user-behavior? Here's an example of funny user-behavior that I wouldn't have predicted, but that I learned about today. Apparently there is a not insignificant number of people out there who think that every link on the Internet requires a double-click - in other words, they treat all links as if they were icons on a desktop, and they rapidly click them two times whenever they want to use them. This is harmless behavior, of course, because it doesn't prevent anyone from accomplishing their goals on the Internet. But there is an nearly infinite number other behaviors, some of which are harmful, and if you tried to guess them all in advance you'd fail miserably.

Here is an example of a more harmful behavior. Google redesigned its Chrome Internet Browser because they noticed that a lot of people were attempting to type search terms into the URL field. With a normal browser, that would be a very frustrating experience. Rather than trying to educate those folks on how to use the browser properly, Google simply enhanced their browser so that people could type search terms into the URL field.

The point is, people will use your website the way that makes sense to them, not you. And if you listen to what they have to say, you can make your website a much nicer place to be.

Will Marlow co-founded AlumniFidelity to help his clients reposition their fundraising to benefit from Web 2.0 technology and marketing techniques. He’s working with clients such as UVA, the College of William & Mary, the University of Oklahoma, Bowling Green State University, Randolph Macon College, and he loves nothing better than a thorny marketing challenge. Read more about Will Marlow here, or email him at will@alumnifidelity.com.

Loading mentions Retweet

Filed under // Software development Web development Website development

Comments [2]

February 5, 2010

Facebook Is Doing "Split-Testing" Testing Right Now

Split-testing, or A/B testing, is one of the most important concepts in designing new features on a website. Anyone can do it for a very small amount of money (if you know how), but I wanted to share a real-time example of split-testing in action.

Click here to download:

Facebook_Is_Doing_Split-Testin.zip (115 KB)

Facebook is currently testing a new user-interface, which means they may be gearing up for a major overhaul. My fiance (Photo A) logs into Facebook and sees the new layout, while I (Photo B) login to Facebook and see the old version. Facebook's analysts will measure the clicks of group A and compare them to group B, and if the analysts determine that group A is more productive or optimal against its metrics, they'll implement the changes for all of Facebook's users. Click the images above to see the different layouts that Facebook is testing.

Will Marlow co-founded AlumniFidelity to help his clients reposition their fundraising to benefit from Web 2.0 technology and marketing techniques. He’s working with clients such as UVA, the College of William & Mary, the University of Oklahoma, Bowling Green State University, Randolph Macon College, and he loves nothing better than a thorny marketing challenge. Email him at will@alumnifidelity.com.

Loading mentions Retweet

Filed under // A/B Testing Facebook Software development Split-testing Website optimization

Comments [0]

January 2, 2010

Usability Testing on a Shoestring

A lot of times we get really fancy when we need to do things like "usability testing."

Someone I trust recently described a great way to do usability testing for your website for just $30.00.

1. Pick three of your friends who you don't work with.

2. Give each of them $10.

3. Sit down with each of them in turn, and ask them to make a donation on your website, but don't give them any instructions.
4. Watch what they do, and take notes.

How long does the process take them from start to finish?
Do they find you by searching in Google?
Do any of them fail to make a donation?
Do they get sidetracked by anything?
Do they look frustrated?
Does any part of the process confuses them?
Where do their eyes go when they reach your website?

You can learn more by spending $30 in this way than you can by spending weeks or months thinking about the way that you think your website should work. And even if your web site doesn't have anything to do with donations, you can still learn a lot from an informal focus group like this.

Because when it comes to the user-experience of your web site, what you think is just a guess until you ask.

Will Marlow is the co-creator of AlumniFidelity, which helps schools and nonprofits improve their online fundraising results with peer-to-peer, social-based fundraising. Email him at will@alumnifidelity.com.

Loading mentions Retweet

Filed under // "Usability Testing" donations higher education modular modules nonprofits Online Fundraising schools Software Software development

Comments [0]

December 31, 2009

What is A/B Testing?

A reader pointed out that I referred to something in a recent post called “A/B testing,” but I did not define it. Essentially, A/B testing is something you do when you are re-designing a web site or building new software, in order to find out if a new change will be popular among your users. You set up an A/B test in which one visitor to your web site (visitor "A") sees the original feature, and another visitor to your web site (visitor "B") sees the new feature, and then you compare the behavior of Group A and Group B. The results of this type of test frequently are among the strongest factors in determining which features become a part of a web site. (When it comes to building a company like AlumniFidelity, which has products that live exclusively on the web, A/B testing should be done as often as possible to help you make sure the product fits the market the way you want it to.)

It used to be that A/B testing was the type of thing that only major companies like Google did. But now it takes know-how more than money to make A/B testing a part of your operation. The cost is really the time it takes you to learn how to use the free tools (or the fees you pay to the consultant you hire to do the work for you). You can begin with two free tools in Google Analytics and Google Website Optimizer, and you can get started with A/B testing on your web site.

Why do you want to know about A/B testing? Because, people, in another day it will be 2010, and staring at shapeless, uninterpretable masses of data hasn't been cool since the 1990s. :)

Will Marlow is the co-creator of AlumniFidelity, which helps schools and nonprofits improve their online fundraising results. Email him at will@alumnifidelity.com.

Loading mentions Retweet

Filed under // A/B Testing Marketing site development Software development Split-testing web marketing Website optimization

Comments [0]

December 26, 2009

The Cost of Building a Business in 1995 vs. Today

I was just reading the 2005 blog post by Ryan McIntyre about the plummeting cost of doing a startup, and thought I'd share the comparison of prices for hosting services in 1995 versus 2005.

Bandwidth: $1100/megabit/month in 1995 vs. $128/megabit/month in 2005
Cage Space: $175/sqft/month in 1995 vs. $25/sqft/month in 2005
Disk Storage: $1,300,000/TB in 1995 vs. $3,300/TB in 2005 (SCSI RAID)
1-CPU Server: $25,000 in 1995 vs. $1,000 in 2005 (web server class machine)
4-CPU Server: $360,000 in 1995 vs. $38,000 in 2005 (with 16GB RAM)

Most investors and entrepreneurs LOVE talking about the low costs of doing an Internet-based startup these days.

Which is why I was very glad to read Bill Burnham's great article about the true costs of building an enterprise startup, and why plummeting software costs don't totally change the game for enterprise startups.

Burnham has three main points (all true) as to why "enterprise startups" (think Blackboard or SalesForce.com) are different from "consumer-startups" (think Google or Twitter):

(1) enterprise customers don't do "betas," -- that is, they won't buy anything that's not ready for prime time;

(2) enterprise customers need customer support and sales attention (which Google is learning as it begins to enter the enterprise space for the first time); and

(3) enterprise customers want an inherently stable infrastructure, which costs money to build and maintain.

The only PS I would add to Burnham's post is that cloud computing has a fundamentally different value proposition for enterprise businesses than it has for consumer businesses. Enterprise businesses use the cloud to deliver better software, not simply cheaper software. An enterprise product (whether cloud-based or not) needs to be fantastic to be viable, because being cheap isn't good enough.

And to make a truly fantastic product for enterprises, you need to follow an iterative development process to find the right product-market fit. That is, you would never dream of building a software package alone in a room with four engineers and unveiling it to your enterprise customers at a grand opening. You need to build a non-beta product that is ready for market, make sales, get customer feedback, do A/B testing, get customer feedback, expand your reference-base, incorporate customer feedback and A/B test results into new features, and repeat until the product fits the market, and you're ready to scale. (Of course, this is similar to the process you follow when building consumer software, too, but in this case you need to factor in the higher costs for doing it (1) to a higher non-beta standard for each iteration, (2) doing it in a stable environment, and (3) incorporating sales and support resources starting on day one.

In other words, falling hosting costs may have reduced the overall price tag (with those cost savings coming primarily early in the process), but bandwidth and "cage space" aren't the only things that you need to pay for when you set out to build a fantastic enterprise company.

Will Marlow is the co-creator of AlumniFidelity, which helps schools and nonprofits improve their online fundraising results. Email him at will@alumnifidelity.com.

Loading mentions Retweet

Filed under // Bill Burnham consumer business Enterprise business Ryan McIntyre Software development Startups

Comments [0]

December 9, 2009

How to Learn from Google's Beta Tag

The beta tag is one of Google’s brilliant hallmarks.

The beta phase (it can actually last forever) solves one of the key challenges in software product development. The first challenge (unsolved by the beta phase) is to build software that solves a serious consumer or business problem, which is no small task. But the second challenge (once your software solve a serious customer problem) is that you can never figure out how human beings will actually use your software once they get their hands on it. That is, unless you have a strong beta program and you watch a large number of your users in action, and you learn from them.

Google is the master of using emergent data. Don’t design Gmail. Let your beta testers play with your new email system until the right patterns emerge, and let them design Gmail for you. Same with Twitter. Let the right features emerge, don't airdrop them in from corporate HQ.

Having a strong beta culture is all about knowing when to stop inserting your own preferences and opinions into things, and letting the end users tell you how the product should be designed. That’s when the software gets designed right.

Loading mentions Retweet

Filed under // Beta Google Software Software development Startups

Comments [0]