I've been updating our website the past couple of weeks (which explains the lack of blog posts). I'm definitely not a web developer so I'm having to learn as I go along. Here's a couple of tips I've come across.

Relative-Absolute Positioning

Say you want to position some text exactly (ie absolutely), but with reference to a parent block instead of the page.

(Example: http://www.feedghost.com/UserArea/Help/Introduction.aspx)

The trick is to make the parent block relative positioned. (Assuming you're happy with where the parent block is, don't specify left or top offsets).

<div style="position: relative;">
  <img src="..." style="position: absolute;" />
  <div style="position: absolute; left: 20px; top: 20px;">
    Positioned Text
  </div>
</div>

Why this should be the case I have no idea. As far as I'm concerned, CSS is conflating two different concepts there. The position attribute normally controls the position of the block and all its children; making children relative to the block vs relative to the page is a different concept.

Flicker-Free Image Roll-Overs

Quite often you want a link to be rendered as an image that changes subtly when the user moves the mouse over.

(Example: http://www.feedghost.com - the tabs, secondary navigation, and "stars")

The simple method works, but suffers from a flicker when the mouse is first moved over the image, as the browser loads the mouse-over image:

.MyButton a
{
  display: block;
  width: 50px;
  height: 50px;
  background-image: url(Normal.png);
}
.MyButton a:hover
{
  background-image: url(Over.png);
}
 
<div class="MyButton">
  <a href="..."></a>
</div>

A much better technique is to combine the two images into a double-height image, with the normal one at the top, and the roll-over one at the bottom:

.MyButton a
{
  display: block;
  width: 50px;
  height: 50px;
  overflow: hidden;
  background-repeat: no-repeat;
  background-image: url(Combined.png);
}
.MyButton a:hover
{
  background-position: 0px -50px;
}
 
<div class="MyButton">
  <a href="..."></a>
</div>

OK, here's yet another thing that bugs me: MSDN Just Published (RSS). Not because of the content, or anything like that, but the method of delivery: you get nothing for a month, and then all of a sudden you get fifty-odd articles, published at two-minute intervals. Had they released them spread out over the month, I'd read them, but fifty at once? They all just get marked as read. Maybe they still think they're running MSDN Magazine or something...

PS Click the image to the right to see our upcoming true river-of-news display, ETA one or two weeks.

( Part 1 - Advertising )

( Part 2 - Web Tracking )

Apologies for the delay in this series, I had a bad bout of flu, and then of course I've been trying to catch up with work.

We've been running the adverts for about a fortnight now, and tracking the web stats, so what have we learned?

1. Adverts Sort of Work

We saw a huge spike in traffic once our Google ads started running. With that came a smaller spike in downloads.

We need to improve our front-page: it's not inspiring people.

2. Adverts Sort of Don't Work

We didn't get much of a spike in first-version-checks or registrations. There's some sort of problem there. Why would someone download FeedGhost and then not run it? Perhaps we have a serious bug. Perhaps people just fiddle with it for thirty seconds and then discard it.

We need to improve our help. We need to get people hooked on using FeedGhost.

3. Adverts Need Watching

Google ads run on search, and also on their "content network" (i.e. other web-pages). We found that by midday, our budget was all but consumed, primarily on the content network. We turned that off, and our ads ran all day, with money to spare, and the quality of clicks improved (that is, people were more likely to download after hitting the front-page). Why is that? Perhaps people searching are actually searching for a product? Perhaps the content network is being gamed?

FeedBurner has been a real disappointment to us. We spent roughly equal amounts on Google, Pheedo, and FeedBurner. A quick look in the logs show that Pheedo produced ten times as many hits as FeedBurner, and Google produced three times as many as Pheedo. Now perhaps we chose the wrong target for FeedBurner: but it doesn't matter. We simply won't use them again. Google went out of its way to help us choose appropriate keywords; FeedBurner gives you none. Google and Pheedo scale nicely as you add more money: with FeedBurner, to spread ads around multiple blogs, costs tens of thousands of dollars. The low returns and difficulty of spreading ads means we won't be going back there for a while.

We're going to let our adverts wither. We're moving onto community engagement.

4. People Vary

We've had some fantastic feedback from users while FeedGhost has been publicly visible. We've had great reviews, great constructive criticism, chats, emails, bug reports, the works. Unfortunately, that's the good half of the coin: most people don't get involved. (And I'm not blaming them: we're all busy. I'm just looking at it from my selfish point of view). We get a handful of canceled installs, but we've only had a one or two actual messages explaining problems in installation, for example.

Feedback forms are useful - but mainly for the web stats.

5. Blogs Really Work

Other people's blogs that is. A quick and completely unscientific scan of the logs indicates that putting a targeted comment on a blog is worth about ten hits, and getting a review is worth about fifty. (And I'm not talking A-listers here; I'm talking blogs with a couple of thousand readers; the sort of blog you read because of the great content, not just because everyone else reads it). At Google prices you've saved yourself two or ten quid respectively.

Once we've got our updated river-of-news control in place, we'll start asking bloggers if they're interested in reviewing FeedGhost.

I have to take my developer hat off, and put my marketing one on.