If you are viewing this article on the site, and come here often, they you will notice much of it looks very different today.
We’re in the process of doing a massive upgrade to the entire site.
This will make navigation much easier as we detect if you are logged in, and if you are a free or paid member, and then we’ll show a different navigation bar to make it easier to go between items.
The member pages will be the next overhaul, making membership pages even easier to navigate and visualize.
I’m expecting that the few remaining bugs will take 2-3 days to squash. We know of most of them, but its possible we missed some as well – so if you spot them, please just let us know.
On the ‘lessons learned front’, here’s the story.
We have a dev server that is 99.8% the same as production. The site was fully coded and tested on the dev server except for exactly one issue that could not be tested on dev.
We’ve been working on this upgrade for quite a while, and the team has been very anxious to get this shipped so we can go work on making other pages even more friendly and work on community engagement.
I’m at a conference right now, and got back to the hotel around 8pm and the team really wanted to launch. I asked how long it would take, and the answer was “15 minutes”. Now, I’m a realist so I’m thinking this is a one hour project as no matter how much you try – something will break. But with Pubcon next week and knowing I’d be too busy to even think about a site launch, I wanted to see it shipped as well (hopefulness winning out over a pragmatic approach).
So we launched it. And we did break some items and we fixed most of them (there are still a few lingering issues on very specific pages), but by this time it was 11pm. So I’m thinking this isn’t really that bad as I don’t have to speak until later tomorrow.
To put this in perspective, we have 3 CMS on the site and more than 10 CSS sheets, 30 page templates, 10 headers, 4 footers, and 10 sidebars. I’m sure the numbers are higher and I’m just rounding down. So, doing a 30 minute deployment and 2.5 hours of bug squashing isn’t really that bad.
So we went to put in a test transaction to make sure the shopping cart also worked – and it failed. And we fixed the problem – and it failed. The only difference between dev and prod is that the dev site doesn’t have a merchant account so we can’t test transactions on dev. This has never once mattered in the past, the payments always worked - but this time it did matter – a lot.
Oops, if you can’t take money – you can’t stay in business.
So, we rolled back the shopping cart – so the look and feel of just those pages is the old style until we squash the XML payment errors. So if you go to your membership pages, or go to create a membership and feel as if you’ve just switched sites – that’s why.
We had the option to roll back the entire site to the old look and feel, but internally, that probably would have caused some people to feel as if their work was substandard or that they just failed. So I decided to leave the site split so that they could see what they were working towards, and have even more motivation to just fix the rest of it.
If you plan on a massive upgrade – its best to not be traveling – to do it in the morning (not one thinks clearly at 1am when it comes to XLM code) and probably have more contingency plans in place. It also helps if you dev site is 100% the same as your live site. You can’t always do it – but the closer it is – the better your results will be.
This is the type of upgrade I’d have never done for a client, never done on a really large site (especially ecommerce); but I find marketers are better at working for their clients than themselves
As any website owner, marketer or not, its always best to treat yourself as a client and do deployments, updgrades, etc in the same way.
So lessons learned, the site will soon be much nicer on all the pages, and hopefully, you’ll have a better experience learning about AdWords and using our tools.