Website Optimization Techniques

Deciding the pricing of a product can be a tricky situation to be in. There is an optimum price tag for any product where the profits are maximized, but finding this sweet spot is the problem we need to solve.

Supermarkets are experts at this since they carefully analyse the data from millions of shoppers to finely tune their prices. But checking out the price of a can of beans in the supermarket does not help us much on the web I think.

So we have to do our own testing. It’s not efficient to guess, since if you get it wrong then you will make no sales. The price has to be dynamically adjusted to continually zero in on the most profitable price point for your product at any period throughout the year.

But to test with too many price points may be a problem if you have no idea of a suitable price and/or you have low visitor traffic to your sales page.

So it helps if you can get some knowledge of what digits to use in your online sales prices. Then you can produce a manageable number of variants of your sales price to test out.

To cut a long story short, the most profitable numbers to use are: 0, 1, 3, 5 to produce prices such as $15, $35, $100 etc. But this is not my data (it’s from the man from Costa Rica who shall not be named on Forums).

I am currently setting up a test with my own product and look forward to revealing results when I have them. Before this test, I did make multiple sales at $5 and one sale at $25

A while back I asked a question on a marketing forum why everybody was obsessed with using 7’s in their pricing since I couldn’t find any logic to it. And nobody out of the circa 100,000 members answered my question.

So that makes me suspect that the guru’s have programmed their automatic lemming buyers to respond to the “7″ digit. Of course, the guru’s do not reveal these kinds of secrets to their followers. Only those in “inner circles”. But let’s not get too mystical here.

p.s. I just found an option in my PayPal payment acceptance preferences that is set to “must have PayPal account” by default which I think would reduce sales, so I switched this setting, I don’t care what credit card processing system customers use as long as they buy.

If you have a selection of Ads to test, you are inclined to rotate them and see which ones work best. In my previous post I included some code to do this but to take it a stage further, there are some optimizations we can do to speed up the process of optimization of the Ads.

First of all you can monitor the clicks on the Ads. This is the first goal of a banner or text link Ad, to get the click. The second goal is to achieve a conversion to a sale or action such as a subscribe request.

As Ads get clicked, you can adjust how often they are displayed. The Ads that get clicked most should obviously be displayed more often but only after you have a statistically significant number of clicks to evaluate.

So you can give each Ad an exposure life time where they either perform or get binned. For example, 100 exposures. If the Ad gets exposed 100 times and there are no clicks on it then you may want to expire it. If most of the Ads in your rotation have similar dismal results, then you can start again with a higher exposure allocation and also, consider your web page content for improvement to get better relevancy to your Ads.

To further refine this process, we can track our visitors based on their IP address and attempt to give them a cookie. This way we can show the same Ads to this visitor so that we are testing the same set of conditions with them. What do you think? Is this the correct approach or should the Ads change on each hit on your site by the same visitor?

Also, we need to differentiate between real human visitors and robots. This is vital since it is very challenging to make a sale to a search engine robot.

The way I do this is by checking for common browsers such as FireFox, Internet Explorer and Opera. Since most of your human visitors will be surfing with just a handful of common web browsers, it is easier to check than trying to detect a bot. Some bad bots or content scrapers may fake the browser details but they should be in a minority, so this method should be accurate I think.

So now we have visitor data and click data for all our Ads. This allows us to calculate a click-through-ratio (CTR) for each of our Ads.

As the exposures build up, we can decide to make use of our CTR calculations to govern how often our Ads are displayed.

Say there are 3 Banner Ads in rotation: Ad1, Ad2, Ad3 and they have CTRs of 3%, 2% and 4% respectively.

3% CTR means 3 clicks in every 100 displays.

Reader exercise: can you think of the best way to display these Ads for maximum conversions given that conditions may change?

If you think that we should display the 4% CTR Ad all the time, it is the wrong answer I think since conditions can easily change such as your source of website visitors and anything else in the news etc.

If you do have wildly fluctuating sources of traffic (lucky you), you could look at creating separate landing pages for each traffic source.

So it is best to balance the exposures of each Ad according to their CTR. So in this simple case we could give the Ads a bias of 3:2:4 in our allocation of visitor traffic.

Now, the best performing Ads get the most exposure, but the under-performers still get a chance to improve or will die off.

But we have to continually adjust the settings as we recalculate the CTR values. So we need software to do this automatically. Stay tuned for my software

You may wonder why we are going to all this trouble just to rotate a bunch of Ads. The answer is to achieve optimum results in the shortest time possible and ideally on autopilot. Since in marketing, time is money and why waste time maximizing profits?

Wow, this is an interesting subject don’t you think?

To your success,

Mr MultiVar