Skip to content

Calculating the Revenue of Ad Based Business Models

Why Estimate?

People often pitch ad-based revenue models for sites with very small niche markets without acknowledging that the business will only be viable once the market has been greatly expanded. Sometimes this is just a simplification, but often it is because it just isn’t obvious how small the ad-based model revenues can be.

Total Available Market

These tables provide a basis for estimating the revenue through the value of 1000 unique visitors per month. The revenue needed to sustain the business is then divided by the value of 1000 unique visitors to give an estimated necessary number of visitors and the total available market required. The Total Available Market (TAM) is the number of potentially profitable customers who could make use of the product or service. In the case of web advertising, the number of potential users in the target audience of the website. Obviously, there are a large number of variables, so the value is only an indication for a generic business, but should allow a reality check, and in some cases for a more realistic TAM to be planned.

Your entire website and the individual pages are considered by search engines and ad networks to be about a list of keywords. The number of keywords is quite limited, so consider the couple of search engine queries which you would most like to be the recipient of. Then list the four or five distinctive keywords from these queries. Google Webmaster Tools offers a method of viewing keywords associated with your site by Google. These keywords are a factor both in organic search traffic as well as ad placement. Consider how closely they match high-valued purchase decisions. For example, does your site review new cars, a classic high-value purchase decision, or is it about pedestrian safety. Clearly the auto reviews are closer to a high-value purchase decision.

Think not only about the initial purchase but about the lifetime value of the customer to your advertiser. For example if a consumer purchases cable TV every month for three years then the decision has a considerably higher value than the initial purchase. Other examples which need replacing after use are replenishment items such as vitamins, diapers or books which can total a significant value over a period of time.

Visitor Valuation Based on Purchase Decisions

The majority of online advertising is attempting to solicit a direct response from the viewer. A click on an advertisement which leads to a purchase or registration. As a result if your page is a good home for purchase related topics your advertisers are obviously going to have a higher response rate to their placements. The higher value those responses are to the advertiser the more they are going to be able to spend on competing for your advertising space.

Based on your estimate of the lifetime value of customers who are on your site, select a value per 1000 unique visitors below. If your website is not directly related to a spending decision (e.g. pedestrian safety) then use the initial value of 25 cents.

Lifetime value to advertisers V (Value of 1000 unique visitors)
Not purchase related $0.25
< $150 $1.50
$150 – $1,000 $5
$1,000 – $5,000 $10
$5,000 – $10,000 $25

Estimated value (V) of 1000 unique visitors per month based on live time value to the advertiser.

The majority of websites are not directly purchase-related so in most cases you should be selecting the first row of the table above as the estimate. Compare the value of 1000 visitors to your site with the spread of values within the Google ad network. An advertiser may have paid Google AdWords over $1.00 for 1000 individual impressions where as AdSense will have paid out to the average site less than 25 cents for those 1000 impressions. Sites more in demand by advertisers will be paid a higher proportion of the AdWords revenue. The fluctuation of this demand in the auction-like algorithm from time slot to time slot across the week makes predicting the revenue an inexact science.

Affiliate networks will provide a greater return (V) than ad networks, often due to the web site content being more closely related to the advertisers proposition, and as a result providing a more effective funnel for customers. The content needs to be more tailored to the advertiser to get the full benefit. The affiliate model can also be used to avoid the margin taken by an intervening ad network. For example, most marketing on Facebook is priced on an affiliate basis and manages to be targeted using the additional information about the user, including their reaction to other advertising, their networks reactions and memberships.

Note here that V is related to, but different from the CPM, the value of 1000 impressions. There are likely to be multiple advertisement impressions shown to a visitor over a month, and on a single page, however you will not have many effective advertisers for any individual so the benefit quickly diminishes unless the site has time to learn about the user and place better advertising. A single well-targeted advertiser is more valuable than showing twenty different advertisers to the user across the month. Subsequent impressions of the same advertiser have a diminishing chance of being clicked upon or at least have a probability as a sequence, that may build with credibility for a less known brand, but which doesn’t grow linearly with the number of impressions. Here, we are using a simple model based on the number of unique visitors regardless of the number of impressions or different advertisements displayed.

Most websites have an average of less than five page views per unique visitor per month as approximately half of visitors will view only a single page. If you can command your user’s attention for periods of time – or even a few hours per month – then your value (V) from the table above may be a multiple greater than that shown. Facebook, for example, made a revenue of $700 million in 2009 from approximately 400 million registered users. If we assume an estimate of 350 million unique visitors per month then their value per 1000 unique visitors was approximately ($700 million / 12 / 350 million) * 1000 = $167 per month per thousand unique users (V). This is due largely to the fact that they achieve many hours of usage per month from their median user. This is extremely unusual – if not unique, but still only $2 per user per year. An average site as generic as Facebook is receiving below a dollar as the median user leaves within a few seconds and does not often return. Many parked domains actually have a better defined audience than normal websites and can receive $5 to $10 per 1000 unique visitors. The lack of content makes following an advertisement actually more likely.

Value estimation based on the duration of user engagement

Sites which have users returning to them rather than being driven by search results can have heavy monthly usage per user and as a result require a second method of establishing the value of 1000 unique visitors based on the number of pages viewed or the amount of time for which they can be held on, or caused to return to, the site. Having the user on the site for longer allows greater opportunities to gather data on the user and to fine tune the advertising. It also implies a stronger bond with the site and allows the sale of advertising space in greater quantities to more significant advertisers and possibly by a more direct sales process. For sites with significant user stickiness, which implies loyal readers, or addictive games, there may be significantly greater revenue opportunity, and it would be fair to use the greater value for V of those shown in the tables above and below.

Median time on site per month V (Value of 1000 unique visitors)
< 5 minutes $0.50
5 to 10 minutes $15
10 to 30 minutes $40
30 to 60 minutes $50
60+ Minutes $80

Estimated value (V) of 1000 unique visitors per month based on total monthly stay.

Very few sites have a median visit of greater than five minutes but it does occur for popular newspapers and for the most popular online games. For the largest of purchase decisions, of a house, car, medical plan, or cell-phone plan, it may be reasonable to assume that a site may have significant duration of reading, in addition to being related to the purchase decision giving a high valuation. The difference between obtaining value from effective strategies, such as the duration of usage or closeness to a purchase decision, rather than less effective ones, such as large numbers of transient users, is key to debates such as Rupert Murdoch’s preference for loyal readers and shunning of search-driven momentary visitors. Typical newspaper content isn’t directly focused on a purchase decision but it also illustrates the ability to play both the high duration and high value purchase decision strategies simultaneously. It isn’t a coincidence that major newspapers have supplements for real estate, automobiles and high-end travel as these are all high-value purchase decisions.

So you have, by the use of the life-time value table, or the total monthly stay method, a value of V, the amount that an ad model might extract for 1000 unique visitors. We need also to consider how much revenue we need from the business. Some possible expectations for your business’s monthly revenue might be:

Expectation R=Monthly Revenue (Before tax) Unique Visitors Needed
Part Time $5,000 500,000 to 20,000,000
Cottage Industry $10,000 1,000,000 to 40,000,000
Commercial Office $25,000 2,500,000 to 100,000,000
VC Financed $1,000,000 100 million to 4 billion

Monthly revenue expectations.

The higher end of the range of unique visitors needed is more typical but the range is based on the value (V) of 1000 unique visitors. The model also assumes that your visitors are organic and that you have no other significant costs other than your sweat equity. Obviously if you are, for example, using paid search marketing with costs typically of $0.25 to $2.50 per visitor you will be paying out far more than you are receiving.

Market Size

Based on your expectations above, take the value of revenue (R) above and divide it by the value of 1000 unique visitors in the first table (V) to achieve an estimate of the number of thousands of unique visitors needed (U). Say $25,000 for a commercial office sized business / $0.25 per thousand to give 100 million unique visitors per month being needed. There are sites which achieve this, but not many. In fact only Facebook, Google, Yahoo and MSN. You need more traffic than e-Bay! Take a look at Compete for a list with unique visitor counts. (Unfortunately they now charge for the full list but you can also query individual competitor sites).

U = ( R / V ) * 1000 Unique Visitors per Month

Remember that it isn’t reasonable to expect to have every member of your chosen market demographic. It gets increasingly expensive to obtain the more reluctant members of any demographic. For arguments sake, lets say, that you are very optimistic and believe that you can reach 5% of your chosen demographic. Your demographic had better be small and reachable with your message. Lets say bicyclists in Brooklyn. We need to multiply the number of unique visitors (U) by twenty to see the Total Available Market (TAM) size required to support your site. Do remember that not everybody is like the people you know, over 99% of people on Earth do not own an iPhone, the majority of people are not between 18 and 30 years old, etc., etc. For our example of a business wanting to pay for a commercial-office sized business we would need a TAM of 2 billion people. Since the number of bicyclists in Brooklyn, is significantly lower you may conclude that you need to plan to target a broader market. There may, barely, be 2 billion cyclists on Earth. Alternatively you could accept that your business will be part time unless you are able to finance going to a wider geography or develop a different business model.

The aim here isn’t to be exact but to have approximately reasonable expectations and to select a business model and total available market which are plausible. You can expect to have difficulty obtaining funding if the TAM required exceeds the available market which the business can be expanded into. Wider understanding of the limitations of the ad model are an important reason why other business models such as subscription services or data monetization have become increasingly prevalent.


Your business’ growth is also restricted by the fact that there are just under 2 billion internet users – with only about 300 million preferring to read English – and you can’t generally assume that you have more than a fraction of 1% of them visiting your site any time soon. For websites with longer time being spent on the site, the amount of available free time on the part of users is a limitation. It would be possible to show that currently, globally, less than 2000 sites are likely to have over ten million people spend more than an hour per month with them. The public usage statistics would appear to show that fewer than fifty are actually achieving this, and that usage is far more fragmented down below a level at which a VC would find it attractive to fund any but the most popular sites based on an advertising model.

As a result, the most viable advertising model based sites would appear to be those with content closely related to purchase decisions. They can have modest success with advertising and be able to sustain themselves as corporations if they can reach over ten million people per month. Greater success, of the size that might attract VC funding, (over ten million revenue per year) would appear to require very broad usage indeed. Reaching over ten million people for more than an hour per month. This is currently achieved by sites such as news outlets CNN and the New York Times, and more general sites such as Facebook. Not that these news outlets could be sustained by these amounts of money. Facebook is also making considerably more money with its exceptional usage levels. Sites with addictive games and entertainment with a broad appeal might also achieve the same usage at a modest cost. The long tail of general content is unlikely to be supported, as a corporation, by advertising other than by very large, low cost, content farms that can achieve the number of visitors necessary to cover their overheads. Advertising isn’t a model which can sustain long tail sites. It is only going to be significant for the largest of sites and almost none of those will be large enough to give a sufficient return on a classical VC investment.

P.S. I would be very pleased to hear of additional data points which can be used to verify, adjust or disprove the tables in this post.

PDF24    Send article as PDF   

Posted in Business Models.

Tagged with , , , , .

Warning: Division by zero in /home/waves/public_html/wp-includes/comment-template.php on line 1338

4 Responses

Stay in touch with the conversation, subscribe to the RSS feed for comments on this post.

  1. Ani says

    What data are your tables based on? I’m not sure how you arrived at these figures. Could you provide some reference or source from which, for instance, you would be able to derive that 5-10 minutes of activity on a site equates to $15 in revenue? Thanks!

  2. Owen Brunette says

    The data are based on real sites which I am involved with or operate. I have used the cases with the higher levels of return and where a lot of optimization has been done on the performance.

    Thanks for the question. The 5-10 minutes is the median time the user spent on the site not the time the $15 is generated over. The $15 is the revenue for 1000 unique monthly visitors to a site where the median user is spending 5 to 10 minutes on the site. Only a very successful site holds the median user’s attention for more than 5 minutes. If you had say 30,000 unique visitors per month and retained the median for more than five minutes you would be making only 30x$15= $450/month based on this table.

  3. nickk says

    Hi Owen,

    This article is very useful and I am using it to put together a market size and a pitch. However, the article is a couple years old now and I was wondering if you had an updated version of these estimates. It may be a little too much to ask but if you have any documents that demonstrate how you got these figures it would help me out immensely.

    Thank you for your help.



  4. Owen Brunette says

    The numbers are based on real sites but I can’t really disclose more of that than I have in the article. I would really make sure that you think about the various points with regard to your market and product more than worrying about precision of the estimated revenue. Hopefully the recipient of the pitch will understand that too. Every market and product are so different and market timing is also difficult. Every quarter is a different market so there is a limit to estimation by analogy.

    One thing I would add now is that it is also about attracting the right people not just having a site that is directly linked to the high value purchase decision. If people making high value purchases also are interested in fine dining then fine dining may be a perfectly reasonable way to reach them at scale. You are hurt by lack of proximity to the purchase decision but you are helped by the fact that it is now easier to identify the user across websites and get their true value. e.g. know and demonstrate that this fly fisherman is also an Economist reader etc. and sell the ad space as such.

Some HTML is OK

or, reply to this post via trackback.