Wednesday, March 18, 2009

Web advertising

Above: Graph from US comparison site showing with up to 20 times the visitors of BBC Homes & Antiques, Antiques Trade Gazette, The Art Newspaper and Period House magazine websites. Yes Salvo rocks rather.

'THERE are lies, damned lies and statistics,' a Victorian wag once said, and how right he would have been when it comes to making decisions to pay for ads based on a website's visitor traffic figures. But traffic is only part of the story, because a website could have high traffic 10,000 visitors with 10 or fewer who buy, or it could have low traffic of 100 visitors with a high strike rate of 20 who end up buying. Sorting out which customers who buy something found you through a particular web ad is devilishly difficult, not least because the customer often does not remember.

Each time a visitor to your website clicks a link to load one of your pages an entry is made in your internet service provider's (ISP) server logfile. To measure the traffic your website gets a logfile analyser is used. Another method is to place tracking code into each web page and let an external agency, such as Google, measure and analyse traffic.

To measure the traffic of someone else's website there are ranking systems which produce browser bars which are installed by millions of people on their web browsers which send reports back to the ranking company about every page that is visited by each these people. Ranking systems produce an approximation of the amount of traffic each website gets. The best way to use them is to compare websites with ones whose performance you know, such as your own. Most ranking systems will not rank very small sites that accurately. See the graph (above) for a comparison between the websites of Salvo, BBC Homes & Antiques, Antiques Trade Gazette, The Art Newspaper and Period House magazine.

There are paying and free systems, but since none are perfect, Salvo tends to stick with free methods:
• Webalizer analyses web logfiles. Duplicate visits and visits by search engines or other bots should be discarded. Ten per cent of SalvoWEB's traffic comes from automated bots sent by search engines, email harvesters and hackers.
• Google Analytics uses measuring code built into every one of the 60,000 or so Salvo web pages.
• Free ranking systems such as,, and These are all USA-based.

The three factors to take into account when making decisions on buying web ads are
• the traffic a website claims to get
• comparisons with websites whose stats you know
• that the website's own clickthru stats for your banner are the same as the referral stats from your own logfiles. There could be discrepancies which depend on inclusion of search engine and bot clickthrus.

If you are paying for web advertising it is important that you can trust the stats given and that the web site has a system in place to avoid duplicate clickthrus and mutliclick fraud. Salvo currently charges 5p per clickthru on the small DIY build-a-buttons on inside pages, a fixed rate of £42 a month for single home page buttons and £84 for double size buttons, which attract between 100 and 700 clickthrus a month. This gives clickthru rates of between 5p and 80p. Google ads cost between 40p and £4 a clickthru. Salvo has built its own software to detect and discard duplicate clickthrus. Salvo also gives users access to the IP address of each clickthru so users can do their own checks that there are no duplicates. Salvo has never had a complaint about multiclicks or click fraud.

Google ad users argue with Google about duplicate and multiclicks (see ) and suggest that Google should allow access to IP addresses as Salvo does. Some savvy users install click fraud software such as which can be used to check for click fraud on banners or button ads on any site, as well as Google ads.

Counting clickthrus does not give an indication of advertising effectiveness. Mail order traders increasingly use Google Analytics to monitor whether clickthrus result in a purchase from their site. Google Analytics will track a visitor's IP address to flag up a 'goal' if that visitor ends up buying something online.

If you do not sell by mail order online how do you assess the impact of your advertising on sales? The only way is to ask real customers visiting your yard where they heard of you from.

Try compiling a questionnaire with questions such as:
Where did you hear of us from? Web site, Web ads, Magazine ads, Word of mouth, Local ads, Yellow Pages, Already knew
Are you going to buy something from any salvage yard today or in the near future? Yes No
Are you going to buy something from us? Yes No (if not why not, if yes how much are you spending?)
How far did you travel?

Try using the same questionnaire at different times of year, and on different days of the week. Employ someone for a whole day to ask everyone who visits your yard or showroom on that day. You will probably find that only a small proportion of your customers will know where they heard of you from. The information from such customers is priceless, and this can drive your marketing spend. Most successful businesses spend around 10 percent of turnover on marketing which includes advertising, personal selling, entertainment, sponsorship, branding and web site, although at least half should be straight advertising, and ads these days must include web advertising.

Oh, and by the way, don't forget to include SalvoWEB in your web advertising so you can get to some of their 200,000 visitors a month.

SalvoWEB advertising


  1. It's a shame you couldn't alter the axis on the left so we can see what the lines are doing more easily.

    BUT yes Salvo does rocks and we need more proof like this!

  2. The reliability and credibility of the Alexa Rankings have often been subject to differences. There are opinions that the Alexa ranking is far more credible and true for the sites below 10,0000 than that for the ones above it. Another factor that proves to be a major drawback of the Alexa rankings is that the rankings are governed by the Alexa toolbar and the Alexa toolbar users community. All the browser types are not taken into account as far as the rankings are concerned. As for example the Alexa rankings does not work in Windows Vista even though the latter has a huge user base and is highly popular.

  3. Thanks Martin and I tend to agree but I did try to make the point that absolute ranking figures are unreliable compared to comparative figures which are more reliable:

    '. . . Ranking systems produce an approximation of the amount of traffic each website gets. The best way to use them is to compare websites with ones whose performance you know, such as your own. Most ranking systems will not rank very small sites that accurately . . .'


Note: Only a member of this blog may post a comment.