Want to Drive Innovation through Crowdsourcing? Learn the 5 Steps Get the eBook ×

Tuning Your Product Development

By mess In Community Stories

Posted December 12th, 2011

What if you need to do more with less? In the past, that usually meant working longer hours and/or skimping on the quality, scope and completeness of work product. Today, I look at that challenge in a very different way.

The TopCoder Platform places a lens on top of your software factory, focusing in on key metrics within its global marketplace. So, when I’m trying to get more done within my budget the first thing I look at is the market and how my world looks within that market. What is my unit cost (cost per competition)? How long is it taking us to complete each unit and the subsequent project? What is our success/failure rate? Then, I make a couple adjustments and monitor the impact.

Unit Cost

Here is a quick example:

In December I have a budget of X. The amount of work I have planned requires us to produce Y units of output. My cost per unit in November was $3300 (average is $2800 for the past six months) so I knew that Y * $3300 could not exceed X (my budget). Based on the amount of work we need/want to get done in December, I knew that $3300 per unit would cause an overrun. There wasn’t much I was comfortable cutting in terms of delaying or dropping features, which is what I would have traditionally done. Or, I would have asked for more budget and/or “borrow” from another group which would cause them to suffer.

But what if I could get my cost per unit down to $2600? Then I could still get everything on our roadmap done for the same budget. So, that is what we did. We tuned our approach to running competitions to lower the cost per unit. This, so far, has also had the side effect of increasing our success rate and lowering our time to produce each unit. A few of the adjustments we made were:

1. Copilots updated the game plans for all projects in Cockpit. This gave us clear visibility for the full month, along with the projected costs and timelines.

2. Encouraged our copilots to pay attention to the market. Identify areas of opportunity. For example, if we’re running high on assembly contest cost compared to the market, then figure out why and adjust accordingly. Also, schedule contests in an opportunistic fashion so we take advantage of positive market conditions. For example, if there are a lot of web design contests running today, but 30% less scheduled to be running 2 days from now then consider waiting a couple days to post your contest.

3. Decompose further. If there are complex portions of a contest that are driving up the cost, then consider extracting the complex portion to a separate contest.

I’ll post more tips and tricks here in the coming months. If you have questions or specific topics you’d like to hear about, please let us know.