When A Website Becomes Too Popular

Ho Kee Soon

Read an article on “Top 10 Internet Startup Scalability Killers” which highlights the serious faults on why many websites start to choke when the traffic goes up.

That reminded me of a post summarizing a lecture on Facebook’s massive scalability , on what Facebook did right.

A common problem is that the startup did not anticipate their website’s popularity, they were taken completely by surprise. Their project ended in failure, when users start abandoning them due to “slowness”, “errors” etc.

It is understandable that sometimes the project might not have the resources (hardware, the developers etc) to have a perfectly scalable solution.

So the next best thing is to be prepared for the “possibly inevitable”, to be able to “fail gracefully” or “soften the impact”.  Or like how Hal Helms called it, “The Pre Post-Mortem“.

8 Comments. Leave new

Project teams or vendors are typically so pressed to roll out VISIBLE and FUNCTIONAL software by a certain date that it not practical to factor in scalability.

For an internal team, why worry about a POTENTIAL future problem when the current visible work is already pushing the limits?

For a vendor, why prematurely think about a problem that may arise long after I’ve been paid and moved on?

Similar to personal insurance, only by actively appropriating money and/or time to “prematurely” plan for scalability can a business owner protect his investments over a longer term.

Some of the biggest social websites on the internet never anticipated their growth spurt. It just sort of happened. The websites that pour in great deals of money with tight deadlines usually don’t even come close.

Websites with huge growth offer what their customers – the users – want. Websites developed by corporates usually have the wrong customers – the bosses.

What customers want – Apple App Store. What bosses want, after realizing the competition just blindsided them – Nokia Ovi.

Scott’s response definitely hits the nail on the spot. We frequently see this scenario happening again and again where the final approval has to come from no one but the bosses.

But I also like to throw a question back by playing the devil’s advocate.

If today you are an internal project manager handling the development of a particular website, what would you do to improve the situation? what and how would your actions be to
1) convince your bosses that it’s the users that matters
2) convince your bosses that whatever their opinions may not be representative of the target users?
3) verify and convince which of your bosses opinions are ‘true’ to that of the target users and which are not? To bring it further, how would you even know if your own opinions are ‘true’?

For a project manager, his job depends on his performance in front of the “boss”. Long term benefits may be too far for him to consider.

To answer your question, we can propose running experiments (e.g. using Google Website Optimizer) to get statistical support. Of course, it’s additional work 🙂

Pointing out successes that puts the Users first helps to build a case. But this won’t work if the boss already knows, but chose to ignore it due to other reasons.

The most common not-in-the-best-interest-of-Users alarm is ironically, agreement from the boss. E.g “Yes it’s a great idea. I sure would love to use it. But I can’t see how we can monetize this”.

In fact, if a boss can explain his logical reasons why something isn’t in the best interest of the User, the staff should at least be receptive.

I think the conflicts usually occur because of the difference in criteria used.

doing a/b tests are good. but that’s only possible if we have something to test on. in most cases, i would think most organizations are struggling to get a consensus on that ‘something’.

You must be logged in to post a comment.
Receive FREE updates twice a month on how to improve your online marketing efforts

 Especially for SMEs and Marketers

 15,000+ people have already subscribed. Join them today.

Unsubscribe anytime. We will never sell, rent or share your email.

⇧ Subscribe To Our Newsletter