Lessons Learned

So, we have essentially completed the long-extended transition from Blogger to WordPress. Regular readers over the past few months know that this transition has not gone particularly well. In particular, we suffered severe problems with our server that substantially increased download times and often cut off traffic altogether. This has had a negative effect on our overall readership; we were down by about 50% at the nadir, although we have now recovered somewhat.  Some lessons learned:

1. Unless you are willing to embed/develop expertise within your own organization or pay substantially, going with the “more complex but more flexible” is NEVER a good idea.

We shifted to WordPress from Blogger for several reasons, including most importantly a functionally and aesthetically improved site. We also wanted to move from the blogspot address to our own URL, although we could have accomplished that without the move to WordPress. We hired an individual to manage the transition who seemed to have considerable skills, and who came at a discount. On his advice we went with Bluehost, an inexpensive provider, for our server needs.

We made the decision to push over before we’d worked out all the bugs, in part because we’d only have a full sense of the bugs after undergoing the full transition.  We called this the “Brigham Young” option.  In retrospect, there were certainly some issues that we could have better worked out, but the main problem (the inadequacy of Bluehost) was only revealed by directing the full traffic of the site to the new server.  Unfortunately, the “Brigham Young” strategy put additional pressure on our transition manger to do things that he wasn’t fully prepared to do.

It turned out that the individual we hired had considerable skills with WordPress, and managed to produce a very nice website.  He also managed to transition our feed over without incident, which was one of our most important requirements.  Unfortunately, his advice on the server turned out disastrously; Bluehost was unable to handle our traffic, and unwilling to give us straight answers on why the site was continuously failing.  We also had some problems transitioning over our archives and our comments.  We were not able to save our comments, and unfortunately the earliest archives were also lost, in the sense that we don’t have them in a format that facilitates importation into WordPress.  Our decision to push forward quickly proved problematic, because our transition manager was essentially learning on the job; the desire to avoid serious disruption ran counter to the necessity of learning how to do everything that the transition required.

As our server problems continued, another issue cropped up. Blogger has a very simple, easy to understand interface, even as it imposes some limits on the possible. It’s relatively easy to manage problems and make small changes. WordPress is more complicated; not radically so, but sufficient that it’s impenetrable to bloggers without any html skills. It turned out that even if we had a well designed site, it was difficult for us to manage the blog without assistance. This meant that problem-solving and minor changes came slow, if at all; we didn’t have time to develop the expertise inside the organization sufficient to make the site run smoothly. Similarly, we didn’t have the good sense to either a) note ahead of time that Bluehost would be insufficient, or b) figure out what we needed to do to solve the problem.

By early this month, we were left with a non-functional website, collapsing traffic, and no good sense of how to solve the problem. None of us had time to figure out how to make the site run without assistance. Finally, we decided to turn to the professionals.

2. The professionals know what they’re doing, but it’s hard to know when you need the professionals.

At the recommendation of several other bloggers, we turned to SunAnt, a company specializing in hosting and website design. They figured out our problem, gave us a series of options, and managed our transition to their servers. They are professionals, which of course means that they cost money; nevertheless, they returned our site to good working order with a minimum of fuss and muss.  The archives problem has turned out to be more intractable, as there appears to be no option other than direct reposting of the first eight months of the site.  We’ve have begun this process, although it is (unsurprisingly) going very slowly.  Had we gone with SunAnt from the beginning, we might well have paid more for the website design, but we also might have our full archives, and would likely not have lost so much of our traffic through server problems.

Unfortunately, it’s difficult to evaluate the value of professional expertise without professional expertise.  We didn’t know enough to know what we needed, or to be able to correctly assess the risks and benefits of various approaches.  In retrospect, there are certainly other steps we could have taken that would have improved our knowledge of key issues and prevented the meltdown that eventually occurred.  For example, it was apparently no secret that Bluehost could have problems with a site of this size; a more careful evaluation would have revealed this and perhaps inclined us to a better set of options.  It should also be noted that our own lack of expertise made the situation more difficult for our first transition manager, since we often asked of him tasks that made no sense or that could not be accomplished in reasonable time.

3. Commenters lieare sometimes mistaken about their own preferences.

The number one complaint with the old site, by far, was about the JS-Kit comment system. Accordingly, we shifted to WordPress comments shortly after we moved over. In the immediate wake of this shift, comments crashed by on the order of about 75%. This can be partially explained by the server problems, but not wholly; many regular commenters who used the old system simply never returned after the shift.  None of this can quite recollect why we thought JS-Kit so terrible; it had some issues, but worked well on a number of other dimensions.  FJSKIT became a meme, but in retrospect it’s unclear that transitioning to the new system was worth it.  As I understand it, had we not shifted away from JS-Kit it would have been easier to retain many of the old comment threads, the loss of which we mourn as much as anyone.

And so that’s that.  We have a functioning website, and our traffic has recovered from roughly half of what it was before the transition to about 2/3rds.  For my own part, I can now be relatively certain that when I tweet or otherwise send out links to LGM posts those links will actually function.  The site looks nice, especially with the slightly redesigned header that Melissa at Shakes made for us.  We all wish, however, that the transition could have been accomplished with less difficulty and expense.  Finally, I wanted to thank everyone who’s still reading LGM for their patience during the process; despite a situation that was often intolerable, you stuck with us (or came back when the site worked).  You have our gratitude.

UPDATE: I should add that if anyone is continuing to have problems, please let us know. Someone in comments mentioned a problem with the feed, which I haven’t experienced; please give some details so that we can fix.

UPDATE II: I also would like to direct everyone to this fabulous comment, left by Jamie:

Re: the question of knowing when to seek professional help: that’s a topic near and dear to my heart, as I’ve been doing web development/administration since 1993. I don’t think you need these pointers, but for anyone else with a blog or other web app who is trying to decide between some combination of (a) a professional who bills like a lawyer (that would be someone like me, or I presume, SunAnt) (b) What’s-His-Face’s-Kid, What Knows From Computers, (c) do it yourself, (d) someone who sit between a and b, here are some tips…