Always test, test, test whenever you make any change to your web site.
A week ago, I was concerned that my newest web site was drawing traffic, but not turning the traffic into a reasonable number of clicks on product advertisements at my site. I couldn’t see anything wrong, and I feared the worst: somehow, the project that I’d worked on for the past 10 weeks just wasn’t going to work — even though it was drawing decent traffic, the visitors were clicking on ads at about 10% the rate I’d expected, meaning I’d never earn much revenue.
Finally, dejected and deflated, I decided to do what I should have done all along: start working through a “web-site-launch checklist” that I’d started but never finished a few years ago. One of the first steps was to view the site using five different browsers. While I was fine-tuning the site, I had only viewed it using FireFox and Safari, and it looked fine.
When I viewed my new site using MS Internet Explorer, Chrome, and Opera, all of the link text was visible on my pages, but NOT as hyperlinks.
I quickly discovered a typographical error generated by the script I’d written to generate my HTML pages (an extraneous quotation mark inside the <a> tag, so the tag (simplified) read <a “href=”url”>). Somehow, my versions of Firefox and Safari “knew” how to interpret around this mistake, but the other browsers could not.
Of course, the error was replicated for tens of thousands of links that appeared on all 2,000+ static content pages on the site, all generated by my script (which took about 12 hours to generate all the pages).
I fixed the typo, generated some test pages, and verified that they worked with all 5 browsers, and then I started the process of re-generating the full set of files and uploading them to my web server.
Within a few hours (after I’d uploaded about 20% of the corrected pages), I saw a sharp increase in clicks from my site to my advertisers, and then I saw an advertising fee reported (thank goodness for ShareASale‘s real-time reporting). The next day, I saw that I’d also earned advertising fees from Amazon for the new site (which is now performing within the range I’d projected more than two months ago).
Test, test, test. This experience led me to take a fresh look at that preliminary “web-site-launch checklist” that I’d started working on a few years ago. I “knew better” but still made the same mistake — so it makes sense to spend some time now to refine that checklist, and then methodically use the checklist (which will likely save me much more time in the coming months).