Novel Done!!

Sorry I’ve been so quiet. I actually wrote a whole novel. It started with NaNoWri Month, but I didn’t finish it until December 6th and have been furious edit since then. It’s going to my Mom’s agent Dan as soon as I have it ready for him. We had hoped to make that on Friday, but Ruth’s cancer needed some attention. She is still OK, but they will be doing another operation to make sure that the abnormal cells they found along the edges are gone as well.

I’ve been working on my one sentence description:

A black comedy, well grey, about a rock star, his nephew the banker and their eccentric friend’s journey to come to terms with coming to terms for tragic events in the far past.

It is such a rush to know I did something like that. It changes everything. Walking into Barnes and Noble is 100% different. It’s gone from a reverent location of knowledge to the most exclusive club in the universe. Whose membership is now a holy grail and all should bow down and behold the power of the fraternity of published authors. I think it’s changed my writing as well! :-)

PM Q: What causes a project to slip?

Q: What causes a product or project to slip schedule in your opinion? Where have you had the most problems in the past?

Mistake #1:

Everything in life is a tradeoff and software projects doubly so. Oversimplifying a little, the three variables of software projects are the amount of resources (and that typically means people), the product’s features and lastly, the time to market. You can not change one without affecting at least one of the others.

There are limits to this. Getting 100 people to do 100 features in a week would, in fact, create a program, but probably not something that would be a good program.

If time to market is the number one priority, then features and/or resources must be ruthlessly adjusted to keep the schedule on track.

Mistake #2:

Not enough buffer in the schedule. The different team members will give you estimates of how much work the different tasks require. The key word is estimates. There are a lot of techniques you can use to maximize the accuracy of the estimates, but in the end, people are always going to be optimistic and the only way to address this is to make sure there is room in the schedule for mistakes.

My Feelings:

It’s not IF the project will slip, it’s WHY the project slips.

Projects where the date is the most important thing are called “Slogs” or “Death Marches”. You can’t ask a team to do too many of these without repercussions. Projects that allow features to change or new ones to come in willy-nilly never ship and are just as painful to work on.

I like a reasonable number of milestones whose dates are adjusted as they approach. If too many milestones are coming in late, then program management digs in and figures out why, fixes it and then adjusts the rest of the schedule.

And to keep things more fun, besides the normal buffer, put in some time at the end for creative new ideas and market course corrections.

PM Q: How do you prioritize features?

As many of you know, I'm interviewing for a new job. It's going really well. Yesterday I had to write up a screening document and I thought I would share some of my answers here.

I wish I had had a few more days to polish and simplify my answers, but they are still comprehensable. I start the document out with just about my favorite Mark Twain quote: “I'm sorry this letter is so long, but I did not have time to make it shorter”

-----------

How do you prioritize features? What criteria do you use?

At the highest level: Must have, Should Have, Could Have.

Although generally worked out long before a feature list, big items like business priorities, time to market, globalization, industry laws/standards, revenue models, applied laws, and other items at this level are the first stop.

And I like a very tight vision statement or mantra that the whole company believes about a release. It needs to be so clear that it is immediately apparent to everybody on the team if a feature moves the release in the right direction or not. For example if this release is all about making sure we find every document, we would prioritize supporting the Macintosh over features that go very, very deep into specific file types.

I like the feature team itself to make this call. I like for the developer and tester to sit down with the PM and talk about the features in context and make the first stab at it.

I also like good user and market research, based on science if at all possible. Things like usability testing, experimental releases, betas, etc. Periodic customer Strategic Design Reviews are great. I want to know the other products in our market space better than most of their employees do.

I like to do feature level SWOT analysis: Strengths, Weaknesses, Opportunities and Threats. This can keep the most important issues front-of-mind even if they are not that obvious. Sometimes it a feature looks very cuttable until you remember that this feature plugs a hole that could keep the company out of trouble in Indonesia.

If I’m pulling together a team wide prioritized feature list then I typically need to do some balance between the different feature teams’ prioritization. With all of the above data, matching these up is generally pretty obvious. Then I publish the list in an easy to understand manner and gather feedback and make adjustments. If possible, I would love to do a team offsite on this and get bottom-up buyoff and feedback. The more I can get the team on the same page here the easier the rest of the journey is going to be.

And one last thing I like is to have some space at the end of the schedule for a feature to come back. I create a mini-milestone that can be used as extra buffer if needed or for the team to sort through all the postponed features and new ideas and pick a few that are implementable. This takes the pressure off any one feature cut and also gives a way for the team to be more creative or more responsive to any late breaking market pressures.


-------

More soon...

More on the Aston Martin's Green

It's has been demanded that I be more honest about the Aston Martin's green.

The picture I linked to had so many lighting tricks that it really didn't show off the color unless you already knew it. The picture on the left is more accurate.

Yes, it's a lime green, only with more red and yellow. It is a shocking color. In yesterday's picture you could believe that those colors were highlights, instead of the real color.

And that brings us to the second point. It's not just the color, but also the paint. Even a 10 year old girl’s plastic glitter pony cannot compare to the amount of metal suspended in that paint. And there are a lot of layers to suspend the particles, and then they throw in a lot more layers just to be on the safe side. It would not surprise me if they told me that the paint job alone costs more than the average family car.

And it’s not just color and paint, but also the car. This is considered one of the most beautiful cars going. They guy who created the Jaguar said that the reason they were so beautiful was because they had an “eye line”. A band of light that drew your eye from end to end and helped the viewer get comfortably lost in the car’s sexy curves. The Aston has several eye lines. Actually, its eye lines have eye lines.

It’s the combination of the depth of this shocking color blanketing this car’s amazing curves adding almost a magical sparkle but keeping an aggressive, road eating feeling that make looking at one of these a mind blowing experience. They are very rare cars, so spotting one of these is like spotting a rare Blue-Spotted, Red Crested Gull would be to a bird lover. We saw it at a car show, but if we saw it on the road Mike and I would pull over and follow it while we spoke in reverent whispers “Did you see how that mud puddle was beautifully reflected on the chrome grill?”.

Putting a coat of the same lime (+red +yellow) green on a pick-up truck would be shocking, but it would not be the same.

Color Science

Don't get me wrong, I really love what I do for a living, but there is a new field that if it had been available when I was in college I might have gone for.

I might have majored in Color Science.

You may be thinking, "What, you didn't get enough coloring books as a child?" and you would be wrong. Well, umm. Actually, you might be right, but that isn't the point.

Think of a 1970s bathroom and there is a certain orange that comes to mind.

The green of the Aston Martin this year is an amazing triumph of Color Scientists. http://www.bergoiata.org/fe/voiture-sport-90/cars_aston_martin_006.jpg, although you really have to see it in person. One more hint of yellow and it would be tacky. One more hint of green and it would be boring. A little lighter and it would hurt your eyes. A little darker and it wouldn’t contrast with the bright work. It is amazing.

The new colors in Vista are good. In many ways, Web 2.0 sites can most easily be identified by their color choices, although that doesn't always hold up.

And that 1970's orange? Make it a little brighter and it's back in style.


I think my next personal project will be to write a side bar application where you can click on a color on the screen and the program will show tints and hues and suggest color combinations for the new scRGB. The math for this isn’t trivial, but it would be so beautiful. When not in use I think I would just have it step through the spectrum.

If you think I’m exaggerating the math, here is a blurb from a color person’s blog:

http://blogs.msdn.com/billcrow/archive/2007/10/25/hdr-and-color-spaces.aspx

"For sRGB, the 1/2 brightness gray image will result in a numerical value of around 186, not the midpoint value of 128. This is based on an sRGB gamma of approximately 2.2. (The actual gamma calculation for the sRGB color profile is a little more complex, but we can use a value of 2.2 for an approximation for our
purposes.) This means that the sRGB profile uses 186 steps to represent the lower half of the luminance spectrum, and only 69 steps (255-186) to represent the upper half of the luminance spectrum. sRGB defines a non-linear luminance curve to provide more detailed information in the darker or shadow areas at the expense of the brighter or highlight areas. Since we're far more likely to see visual differences between each of the 255 total luminance steps, this non-linear representations significantly reduces the chance of seeing those artifacts.

What's become of me?

Jeeze. My last blog was titled "Maximize Ad ROI".

What the hell??? When did I lose my sense of humor??

It wasn't even:




"Clippy says: Maximize Ad ROI."
























It's just a matter of energy. I haven't been putting energy into humor. I did help Digg.com ship Microsoft ads, but it wasn't that which sucked my humor away.

I will fix this and hopefully not make this mistake again. But if you ever see another title that is as dry as the last one, please put a sign on my back that says "Kick me!"

Maximize Ad ROI

A non-ad co-worker asked me if I would give some advice for a company that wanted to do more to maximize it's ROI. This was my reply:

- - - -

Generically, 1000% better than any other strategy, is for them to sell their own ad space. We are seeing a trend where even medium sized sites have at least one ad sales person. Even if they don’t go that far, having somebody spend a little time on this goes a long way. Since they know their specific users and know what they are generally looking for on different parts of their site, they can link them directly to complementary products. This can bring in 2 to 4 times as much money. Even more if you are willing to work on some joint content that highlights their products.

One-off campaigns, often times called roadblocks, can be very lucrative. One example of this is where you allow every ad placement on your site to be taken over by one company. Sometimes with a special event and content. Or even a few pages/contests etc. One site I work with has tripled their monthly income with just one weekend of this.

Targeting is another key. MSFT gets a 15% bonus for every targeting tag that is applied. I figured that for a few top-of-the line placements people can get $20.00 CPM (1000 views).

Next up is to write the code to be able to swap between a bunch of different advertising networks. Selling ad space is much like the stock exchange. Sometimes they will get $8.00 CPM (for a 1000 views) and sometimes 50 cents. If they can swap networks when the timing is right that can make a big difference. Again, this takes staff. One client has two full time people that do this just for Europe.

And lastly sign up with a single ad network that seems to specialize in their area and implement it and ignore it. Yes it won’t maximize profit, but it also won’t cost you way expensive staff time. For small sites I think this probably has a better ROI.

All of the above are generally combinable.

There are other options, but the above is what I’m seeing being done across the web.

- - - - - -

I'm happy to answer email or comments on this subject.