How to Advertise on the Web
by Arthur Middleton Hughes


The Internet has grown so fast that virtually all large companies are trying to decide how to use it to reach their customers and prospects. There are more than a million web sites out there. A couple of thousand of them accept banner advertising. If you are interested in reaching prospects, where should you place your banner ad?

In traditional advertising, you use a media expert. This expert studies the demographics of your customers, figures out which media they pay attention to, and places ads based on careful targeting and cost. Media targeting has been a science for the past thirty years. A modern marketer would assume that she should use the same techniques for the Web. That would probably be a mistake.

Much of what we have learned about reaching prospects over the years does not apply to the web. Marketers, using those time honored principles in web advertising, will lose their shirts, and fail.

The basic lesson we have all learned is that direct marketing, whether by direct mail, or print, is expensive. To reduce the costs, we must carefully target our message using the right lists or the right media. So, if we are going to sell sports outfits, the reasoning goes, we will pay a premium to advertise on a web site devoted to sports. This type of thinking is probably wrong, according to George Garrick, CEO of Flycast. On the web, the waste cost is very low. It is worth it to cast your net very wide to reach audiences that you could not possibly address with traditional media.

Flycast is the third largest ad network on the web. Web advertising networks offer a single point of access to advertisers that want to reach millions of consumers quickly and easily. They acquire impressions given to them by their we site "affiliates" and sell the aggregated inventory. This process simplifies the acts of buying and selling for both the advertiser and the web publisher. The largest are DoubleClick and 24-7. Flycast differs from their larger rivals in that it is focussed on response advertising, rather than image advertising. In the last two years it has grown from 150 to 780 sites which carry banner ads for its clients. They sell over 2 billion ad impressions each quarter. One ad banner equals one impression.

Flycast’s rivals, like most advertisers, are trying to create an image in the mind of the viewer. Flycast, on the other hand, is trying to get the viewer to respond. Advertising in the web has focused on banner ads which are placed on hundreds of sites. When a viewer clicks on a banner, she is usually transported to the advertiser’s site where she sees their products and prices, and may place an order. Banner advertising is paid for by CPM, cost per thousand impressions. An impression occurs when a viewer clicks on to a site that contains your banner.

Web advertising differs from traditional media in a number of ways:

Traditional Media Web Advertising
Surrogate Measurements Direct Measurements
High cost of waste Low cost of waste
Single messages Multiple Messages
Delayed feedback Immediate Feedback
Hard to adjust Real time testing & tuning

Traditional media must use surrogate measurements such as awareness, focus groups, coupons, and Neilson ratings, none of which measure actual consumer activities. The Web, on the other hand, permits you to know who is looking at the page on which your banner is displayed, how long they are there, where they came from and where they go afterwards. There is a wealth of information that is not available in any other media.

Traditional advertising usually involves single messages. Thousands of dollars are spent on creating the perfect TV, magazine or direct mail ad. Most of these are single messages. You cannot afford to send four letters to each household to measure which works best. Your tests have to be conducted on different households. On the web, you can change your message every hour every day, and expose one customer to many different approaches. In addition, because of the limitations of the Web, your creative must be relatively simple. Complex graphics and colors which take too long to download may result in poor response. On the other hand, "rich media" ads, utilizing high involvement and interactive formats hold the promise of increasing the impact and overall effectiveness of web advertising.

On the web, you can adjust your advertising on a daily or weekly basis. You get immediate feedback on what is working and what is not working. There are four steps that a customer goes through in making a purchase on the web:

  • Click on a Website that has banners displayed. This is an impression. The advertiser pays from $5 to $40 per thousand impressions

  • Click on a banner in the Website, which transports the clicker to your Website. That is a response. You can measure the response rates and get a cost per response.

  • The clicker views your offerings and fills out a form. This is a lead. You can figure the percentage of responses that become leads, and the cost per lead.

  • The clicker buys the product by giving a credit card number and authorizing the purchase. This is a sale. You can figure the lead conversion rate and the cost per sale.

Advertisers are only responsible for the first two steps. They can get people to visit their site, creating impressions. They can encourage people to click on your banner, creating a response. The last two steps are up to the marketers on each web site. Viewers can and do get lost at each step along the way.

So what’s wrong with targeting – selecting the best possible sites for your banner ad, rather than just scattering your message any old place? As direct marketers we believe in targeting. We use it all the time. Why not use it on the web? The reason is that the situation on the web is quite different.

There has been dynamic growth in web advertising in the last couple of years. Pages are being created very fast. There is a glut of unsold ad space. Much of the space is really inexpensive. And, all of it can be measured and tracked.

There are three categories of web ad space:

  • Run of Network (RON) which is very cheap. It is not undesirable. It is not remnant inventory. There are very good bargains in RON. You pay $4 to $6 per thousand impressions.

  • Affinity Group Space for categories such as Sports, Business, Travel. This space is more expensive, but it can be a very good value. This space is not branded. You pay $10 to $20 per thousand impressions.

  • Branded Space This is the most expensive and often has a very low response. You pay from $30 to $70 for branded space.

Note: the data for the charts in this article were provided at my request by Flycast.

A 1% response rate on the web is good. A good response in direct mail is 2%, but you have to spend a lot more for it.

  Direct Mail RON Web
Contacts 100,000 100,000
CPM $400 $6
Cost $40,000 $600
Response Rate 2% 1%
Responses 2,000 1,000
Cost per Response $20.00 $0.60

Of course, a response in direct mail may well be a sale, but a response on the web is only a single step on the road to a sale. If you are going to pay a lot more for targeting, you have to be sure that the targeting is going to pay off. If you paying $6 CPM for RON with a response rate of 1% you are paying $0.60 per response. If someone says that you should be on this really cool Website that costs $42, how much better would the $42 site have to be to give you responses at $0.60 each? If you look at the math, you will see that your $42 Website would have to give you a response rate of more than 7% to be better than RON. Few Websites give you that kind of response rate.

So, if you should not target, how should you decide where to place your web advertising to make sure that you are reaching the right people?

Garrick has a very interesting answer. He recommends testing, using RON. He calls it zero based media buying. Do no targeting at all. Just throw ads out there and see where you get response. Within a few days, at the most, you can figure out where you are getting your response. You can group your responses by category (sports, business, entertainment) or by individual site. Or, even better, you can create your own category "People who respond to ads about my product." You cannot do that in any other media. He recommends to his clients that they begin with advertising on all 780 Flycast sites for a week, and compare the response rates. Here is what happened when Casio ran a RON ad for a digital camera:

What this shows is that the best category for Casio digital camera ads is Travel. Sites aimed at women, games and sports are definitely not winners for them. Would anyone have guessed this without this test? Not likely. Furthermore, this test can be conducted in a week. Responses are tabulated from all sites every day – every hour if you want. Rapidly knowing what works, you can immediately shift and place your digital camera ads on the right sites, dropping the less responsive ones. You cannot do that in any other media. It takes weeks to know using direct mail, and you may never really know using TV or print.

Response rate, however, is only a part of the story. You have to factor in the cost of the space. Would concentration on Travel sites be a good idea for Casio?

  Response Rate CPM Resp.Per M Cost Per Response
Run of Network 0.90% $6 9 $0.66
Travel Category 2.40% $15 24 $0.63

This shows that at a cost of $15 per CMP, Travel sites are just barely a better buy for Casio than RON. In subsequent weeks Casio should concentrate on Travel sites, but only after testing whether the conversion rate of prospects that come from Travel is equal to or not worse than those who come from RON.

Which would be the best site to advertise Carpoint? Would you guess automotive? Of course. But how about sports, shopping and travel? Would advertising on these sites be better or worse than RON? We need to look at the CPM to know the answer.



Cost Per Response

Rate CPM Per M Response
Run of Network 1.2% $6 12 $0.50
Travel 3.1% $15 31 $0.48
Sports 1.2% $15 12 $1.25

What this shows that even the best category for CarPoint, Travel, is only slightly better than RON as an advertising medium. All the other categories are much worse.

When a RON test is completed, the software returns complete reports. The following is the top of a long report on a $13,000 test conducted for E*Trade over one week in November of 1998. A total of 94 sites were used for the test. In all, the advertiser paid for 2.1 million impressions during the week. Viewers clicked on the E*Trade ad 6,495 times for an overall response rate of three tenths of one percent. These 6,495 clicks brought the viewers over to the E*Trade web site where only 105 (1.62%) of them filled out a form to become E*Trade customers and thereby became a lead. The cost per lead varied from $3.68 to $542.02 with an average cost of $128.78. This one week test enabled E*Trade to determine which sites were best for the subsequent rollout.

Results of One Week's E*Trade RON Test by Site
Top Eleven out of 94 Sites

Site Impress Clicks Rate Leads Lead Rate Cost Lead Cost
Igive 14,091 1,384 9.82% 23 1.66% $85 $3.68
Lottery USA 43,112 132 0.31% 12 9.09% $259 $21.58
The Onion 126,732 263 0.21% 10 3.80% $760 $76.04
The Huddle 62,722 49 0.08% 6 12.24% $376 $62.72
123Greetings 361,345 673 0.19% 4 0.59% $2,168 $542.02
3DFiles 56,600 57 0.10% 4 7.02% $340 $84.90
Rogue Market 14,932 10 0.07% 3 30.00% $90 $29/86
Netsurfer Digest 23,333 29 0.12% 3 10.34% $140 $46.67
Nordic Music 28,397 34 0.12% 3 8.82% $170 $56.79
ShowBiz 5,063 6 0.12% 3 50.00% $30 $10.13
People-Talk 37,028 103 0.28% 3 2.91% $222 $74.06
  2,166,196 6,495 0.30% 105 1.62% $12,997 $123.78

As you can see from the chart, the best site was IGive. Next is ShowBiz which had a very low response rate, but had a high conversion rate.

Who would have guessed that the two best sites would be IGive and ShowBiz? Probably nobody. This is the beauty of RON advertising. You are letting the customers – the market—decide which is your best media, instead of paying a high priced media expert to make the decision for you. There is a whole new world out there on the Web.

The next step is to take the raw data from the test, and arrange it by category to determine which categories are working the best for your particular product and offer. The E*Trade successful categories look like this:

E*Trade Results Arranged by Category

  Sites Impr. Clicks Response Rate Leads Conversion Cost Cost/Lead
Igive 1 14,091 1,384 9.82% 23 1.665 $85 $3.68
News & Information 2 98,626 203 0.21% 12 5.91% $592 $49.31
Busines & Finance 12 99,739 590 .059% 11 1.86% $598 $54.40
Automotive 7 19,610 43 0.22% 2 4.65% $118 $58.83
Learning 2 24,105 77 0.32% 2 2.60% $145 $72.32
Sports & Outdoors 5 74,706 128 0.17% 6 4.69% $448 $74.71
Consumer Technology 15 270,169 502 0.19% 14 2.79% $1,621 $115.79
Music 2 22,215 31 0.14% 1 3.23% $133 $133.29
Entertainment & Movies 15 461,936 909 0.20% 20 2.20% $2,772 $138.58
Search Categories 2 49,728 79 0.16% 2 2.53% $298 $149.18
College 3 74,936 92 0.12% 1 1.09% $450 $449.62
Shopping 5 484,580 882 0.18% 6 0.68% $2,907 $484.58
Total 94 2,166,194 4,920 0.23% 105 2.13% $12,997 $123.78

After reviewing all the data, using software and intuition, it is possible to optimize E*Trade results into a few sites that have the lowest cost per lead.

Site Category Impr. Clicks Resp. Leads Conv. Cost Cost/Lead
MarketPlayer Business & Finance 555 4 0.72% 1 25.00% $3 $3.33
Igive Charity 14,091 1,384 9.82% 23 1.66% $85 $3.68
Stock Investor Business & Finance 1,145 8 0.70% 1 12.50% $7 $6.87
ShowBiz Business to Business 5,063 6 0.12% 3 50.00% $30 $10.13
Regent Commerce Business & Finance 2,037 37 1.82% 1 2.70% $12 $12.22
CyberInvest Business & Finance 2,442 13 0.53% 1 7.69% $15 $14.65
Critical Path Automotive 3,520 7 0.20% 1 14.29% $21 $21.12
I-Escrow Business & Finance 7,051 15 0.21% 2 13.33% $42 $21.15
Lottery USA News & Information 43,112 132 0.31% 12 9.09% $259 $21.56
Rogue Market Entertainment& Movies 14,932 10 0.07% 3 30.00% $90 $29.86
ezines Database Automotive 6,014 18 0.30% 1 5.56% $36 $36.08
Netsurfer Digest Consumer Technology 23,333 29 0.12% 3 10.34% $140 $46.67
Nordic Music Consumer Technology 28,397 34 0.12% 3 8.82% $170 $56.79
Your New House Learning 19,591 61 0.31% 2 3.28% $118 $58.77
The Huddle Sports & Outdoors 62,722 49 0.08% 6 12.24% $376 $62.72
People-Talk Business & Finance 37,028 103 0.28% 3 2.91% $222 $74.06
Total 817,123 3,108 0.38% 95 3.06% $4,903 $51.61

Results Optimized by Cost per lead

The majority of the successful sites are, as you could imagine, sites devoted to Business and Finance. There are a number of other successful E*Trade sites however, that would have been overlooked if RON had not been used first as a test. From the above, you can see that in subsequent weeks, E*Trade could spend about $5,000 per week on web advertising, obtaining about 95 successful conversions at an average cost of $51.61 per lead.

Once you have decided where to place your ads, you must decide when to run them. You have a choice of day of the week, and hour of the day. Lets take a look at two different advertisers response rate based on hours of the day.

Bank of America ads peak in the afternoon and evening. They are below average in the morning. Would this mean that B of A should advertise in the evening? Not necessarily. Perhaps their conversion rate from responses to leads is greater with people who call earlier in the day. This can be easily measured. Or, more important, perhaps the lifetime value of early day responders is higher than that of evening responders. Now we are getting very sophisticated. But that is what database marketing, combined with web response advertising statistics permits us to be.

If we look at CarPoint and Casio responses by hour of the day, they are almost the opposite of the Bank of America. Their best responses come in the morning hours, and tend to drop off in the afternoon and evening. Here is another area for rapid calculation and revision of our advertising mix.


What can we conclude from what we are learning about Web advertising?

  • Don’t assume that our conventional beliefs and practices hold true on the web. Start from scratch and let the market tell you what works and what does not work.
  • Continually track, test and tune your program based on the wealth of data that is now available.


Arthur Middleton Hughes is Vice President of The Database Marketing Institute. Ltd. ( which provides strategic advice on relationship marketing. Arthur is also Senior Strategist at ( which provides precision e-mail marketing services for major corporations worldwide. Arthur is the author of Strategic Database Marketing 3rd ed. (McGraw Hill 2006). You may reach Arthur at (954) 767-4558 .

The articles on this web site are available to the general public to read, enjoy and for limited business use. If you want to reprint more than one or two of them for resale or use in a business or educational environment, send an email to Arthur Hughes at He will give you permission by return email. The cost, depending on the number of copies you want to reprint, is very inexpensive.