I got interested in web analytics when I was working as an internet investment analyst for a venture capital firm in London during the dot-com boom. I read 200 business plans, and was part of a small team with $400 million to invest. One thing became apparent; very few people had planned the path between initial site launch and success. Most people could tell you what they were going to build, and what they hoped to achieve. Few could tell you how they could control what happened in between.
This is like setting off in a yacht without a rudder or compass. You know where you want to get to, and if the winds co-operate you'll be OK. But you won't know if you're on track till you bump into land, and even if you knew you were off-course, you wouldn't be able to do much about it.
Before a website is built a number of key metrics need to be planned. This is similar to the core financial model one creates when planning a new business. The numbers which come from the planning stage determine targets and budgets. They also tell you if your online business is viable at all. Basic viability is not a given--most online businesses fail, and never stood a chance to begin with.
The first thing one needs to do is set the key targets for the site. These targets fall into two categories; physical components, and financial targets.
People don't passively view websites, they interact with them. Physical components are the things you want someone to do on your site. With physical shops the thing you want customers to do is give the cashier money. Everything a retailer does is geared towards getting people to do this. Your website is similar-- there are certain pages you want people to see. These are your "target actions." If you're selling goods online, you want them to see the page which says "credit card processed OK." If you want them to make an enquiry which you can follow up with a phone call, you want them to see the "contact details received" page. Most sites legitimately want both.
In order to assess whether this is happening, you need mechanisms to report how many people do this. These people are said to have converted from visitors to customers (or prospects). The percentage of total visitors who convert is your "conversion rate."
When your site is running, you'll need mechanisms in place to change things if you're not meeting your targets for converts. Everything you could conceivably do to increase converts, as an absolute number or percentage, will need its own mechanism for assessment.
Determination of conversion targets is a financial planning exercise. Here you are setting absolute numbers, not the conversion rate. Conversion rates are fairly uniform. Aim to break even at a 1 percent conversion rate, and have a thriving business at 2 percent. A few sites do better, but they are the elites of this business. A few others make a profit at a conversion rate below 1 percent, but other aspects of their business are super-streamlined.
If you complete the planning exercise and you can't realistically make money at 2 percent, the business is not viable. Don't despair-- count yourself lucky you found this out before you built it.
Profit per sale
The critical determinant of how many converts you'll need is your online profit per sale. This is your per-sale profit after you remove all off-line costs. Divide this by 50 - 100 to determine profit per visitor. This is the absolute maximum you can afford to spend per visitor. Period. This includes site construction, maintenance, online marketing-- everything. The reason I am emphasizing this so much is it is usually much lower than people expect. There's nothing you can do about it-- that's the hard fist of reality hammering on your dreams.
You can now determine how many visitors you need. This is the overall profit you want to make from your site divided by the profit per visitor. You can use this number to plan your marketing performance targets and budgets. Remember as you do, the maximum you can spend on any online marketing activity is now set by the profit per visitor. You will find that rates for many pay-per-click ads exceed your per-visitor budgets by a large margin. If that is the case, find another ad outlet, or some form of pay-per-acquisition marketing, such as affiliate marketing. Most people are losing money on pay-per-click advertising, they just don't know it.
Performance enhancements components
The path from entry to target action is your "conversion funnel." You will get bleed along it as people drop out. Once the site is running most of your effort will be focused on reducing this. The mechanisms you will use to reduce the damage will either be to change marketing or change the pages in the site. You therefore need to plan how you will assess this bleed so you can determine what to do about it and how you can assess the impact of what you do.
Key sales analyses will be how many fail to complete the credit card form, levels of shopping cart abandonment and of product reviews failing to go to the cart. You need to ensure you can measure all these.
You will also need to know how many people viewed your contact form versus completed it. If you use a phone number on the site, have a number which is unique to the site so you can assess how many people ring as a result of seeing the site.
Another critical assessment is product searches which fail to result in a product review. This tells you how well your search results and/or product category pages perform. You also need to ensure you can measure bounces from major navigation junctions, such as section introductions. Landing page bounce rates tell you how well your marketing is interacting with your site-- the problem could be with either. The home page bounce rate tells you about the page design of this critical component and less about marketing.
If people can download documents, ensure you can measure how many consider this but do not, and how many who do return to the site. If you can get hyperlinks in the documents, and ensure you can measure click-throughs from them directly.
Most sales come from repeat visitors. Since getting new visitors is more expensive than holding existing ones, almost all profit comes from repeat visitors. A critical step to success is therefore to have mechanisms for enhancing your customer retention. In order to improve in this regard you must have mechanisms for measurement. These can include cookies, registrations, and analysis of usage of retention services such as wish lists.
Mechanisms for assessing traffic generation are similar for most online marketing channels, whether it be pay-per-click search advertising or affiliate marketing.
Since you are paying other people, and the content is in their sites, they will have their own assessment mechanisms. The first thing to plan for is discrepancy between what you see arrive on your site and what they claim to have sent you. There are three possible causes for this; click fraud, lost traffic, and differences in the technical means by which this is counted. Lost traffic should account for around 2 percent of what they send you. Click fraud may or may not be a problem. In my experience most discrepancies come down to different measurement technologies. Your technology is, of course, always more accurate than theirs.
Marketing assessment mechanisms
Assessing the performance of online marketing is primarily about conversion analysis. You need to plan for analysis of bounces on landing, and critical points in your conversion funnel, on a per-channel (or per-ad) basis. The more granularity the better. I have seen sales conversion rates vary from 0.5 percent to 40 percent for different ads to the same site. This permitted switching all the ad funds to the 40 percent ads, with incredible results for the business. Without mechanisms for assessing conversion rates for every ad individually, this would have been impossible. The final step is to be able to undertake financial profit and loss analysis of each ad. A loss doesn't automatically mean dumping the ad-- it may be that there are critical junctions in the conversion funnel which can be tuned.
In the case of direct email and newsletters you also need to assess time to respond and how many people pass the item on to others.
When you do your financial planning, don't forget to factor in the cost to alter the site You also need to plan the mechanism by which this will be done and the speed of alteration. This enables you to plan your mechanisms for analysis and response. The key step is to ensure the timeframe for response matches timeframes for analysis. I once saw a low cost airline collapse simply because they couldn't update their site fast enough. You should also ensure you have mechanisms for evaluating how your responses affect things, or you'll never learn anything. You might also want to think about what reports will go up the management hierarchy above you, and where the data will come from for that.
Finally, don't forget the general cost of ownership, including hosting, bandwidth, staff, and some form of amortization of creating the site in the first place.
Most website planning rarely gets beyond the construction stage, but construction is just the beginning. Managing websites is an active web analytics process you need to plan for as you plan the site. If you go through all the steps I've outlined above you'll have a much clearer understanding of what you're in for once the site is built. Only once you've done all this are you ready to start looking at pretty pictures with designers.
Finally, remember the ultimate words of wisdom from Bill Gates: "most people are overly optimistic about the short term future, and overly pessimistic about the long term." Don't expect overnight success-- plan for the long haul.
Brandt Dainow is CEO of Think Metrics, creator of the InSite Web reporting system. Read full bio.