A considerable portion of paid search advertising expenditure comes from highly specialized PPC marketing companies which co-operate with merchants on a revenue sharing basis. Although revenue sharing is the most advanced performance-oriented search marketing approach, such a pricing mechanism is unable to exploit the full potential of paid search. We suggest a new pricing mechanism, based on profit sharing, that is superior to revenue sharing, and discuss how this mechanism can be applied in practice.
Clicks2Customers Research Series | Clicks2Customers
19 Bree Street, Cape Town, 8000, South Africa
For a merchant there are three different ways to make use of paid search advertising: campaigns may be managed in-house by the marketing department, agencies may be appointed to run a paid search campaign on behalf of the merchant, or the merchant may approach specialist search marketing companies, chiefly through affiliate networks. Today PPC auctions are a highly competitive environment and bid prices are increasing in long term, while they are rapidly changing in the short term. An efficient management of PPC auctions not only requires specialized skills but also sophisticated technology for keyword research, automated bidding and reporting. As merchants’ in-house marketing departments do not usually have core capabilities in PPC marketing, they are unable to manage campaigns efficiently. Therefore merchants frequently outsource this task to third parties – either to a marketing agency and/or to highly specialized search marketing companies.
Agencies provide end-to-end online marketing solutions, but are not specialized in PPC marketing. They tend to work with non-proprietary, standardized industry software and bid on a relatively small number of keywords, mainly brands and trademarks. Remuneration varies from fixed payments to performance-based payments in some cases. On the contrary, specialized search marketing companies usually possess highly sophisticated, proprietary technologies and also bid on millions of keywords. These companies pursue a completely results-driven approach, where the merchant does not bear any risk. A merchant pays a commission, either a percentage of the revenue or a fixed amount, only when a desired action is taken, for example, a product is sold or a lead is generated. In many cases, merchants approach those specialized companies through affiliate networks. This allows merchants access to a multitude of affiliates and lowers transaction costs.
For merchants, the most promising and efficient way to make use of paid search is to outsource it to specialized, performance-based search marketing companies – either through direct contact or through affiliate networks.
In the following section, the profit optimizing bidding behaviour of a merchant will be analyzed. Then, in section 3, we will analyze the current performance-based search marketing business model which is mainly based on revenue sharing deals. Finally, a new business model will be suggested.
A PPC auction is a continuous second-price auction for advertising space on search engine results pages (SERPs). If a user searches for a specific term, search engines such as Google, Overture (Yahoo), MSN or Ask.com display organic results and sponsored listings. Depending on the amount an advertiser is willing to pay, he enters a bid for a specific keyword or a string of keywords. When a customer searches for this keyword, search engines show the corresponding adverts. Whenever a customer clicks on the advert, the cost per click (CPC), the price the advertiser has to pay, is determined by the next highest bid.
The auctioneer – a search engine – sorts all of the bids that participants placed for a certain keyword. Overture awards position 1 to the highest bid, position 2 to the second highest bid, and so on. The positioning mechanism on Ask.com embraces relevance, mainly influenced by the Click-Through-Rate (CTR). Thus, the higher the product of bid and CTR, the higher the position the advert will achieve. Google incorporates a so-called quality score, which is based on CTR and landing page quality. The higher the quality score multiplied by the bid price, the higher the position the advert will achieve. Positions are re-calculated continuously throughout the day and participants may change their bids at any time. An advertiser can submit any bid. Lower bids imply lower CPC, but when the bid is too low, adverts are displayed on lower positions and receive fewer clicks. Two factors determine how many clicks an advertiser will get for a given search phrase: impressions and CTR (Brooks 2005).
An impression is counted whenever an advert is shown to a customer who searches for the keywords for which the advertiser has placed bids. Search engines do not display all adverts on the first SERP. As many users do not click through to successive pages, impressions decrease with lower advert position. In addition, only the first adverts on major search engines will appear on other syndicated partner search engines.
CTR is influenced by a number of factors, such as relevance and copy of an advert. An empirical study by the marketing institute Atlas reveals that CTR consistently falls with lower ranks. It is a well known phenomenon in psychology (“list effect”) that individuals tend to focus on the top of a list.
Thus, there is a trade-off between keeping a bid low enough to maintain profitability and high enough to generate volume (Kitts & Leblanc 2004). In order to optimize the profit of a specific keyword auction, the advertiser has to consider the contribution margin of the product he promotes, must collect data in order to estimate the mentioned variables and then determine the expected profit maximizing bid . This profit maximizing bidding behaviour is illustrated in the following example.
In summary: There is a trade-off between keeping a bid low enough to maintain profitability and high enough to generate volume (Kitts & Leblanc 2004). In order to optimize the profit of a specific keyword auction, the advertiser has to consider the contribution margin of the product he promotes, he must collect data in order to estimate the mentioned variables and then he determines the expected profit maximizing bid2. This profit maximizing bidding behaviour is illustrated in an example in the following section.
1. The CTR is defined as the number of impressions divided by the number of clicks.
A merchant wants to sell a product with a price of $100 and a contribution margin (CM) per unit of $50. His main objective is to maximize profit. He participates in a keyword auction and competes with nine other advertisers, who bid on the same keyword. Data about click potential stems from an empirical analysis, conducted by the Atlas Institute. Click potential is the product of relative impressions and relative CTR. Relative impressions are the number of impressions on different positions divided by the number of impressions of the top ranked advert. Relative CTR is the CTR in relation to the CTR of the first placed advert. Thus click potential shows the expected percentage drop in click volume by rank.
We assume that Rank 1 gets 100 clicks. The number of clicks on lower positions drops according to decline of click potential. The conversion rate (CR) is 2 percent, independent from the position on which the advert is displayed . The CPC we choose in the example increases disproportionately with higher positions, something that can be observed in practice.
The merchant maximizes his profit as follows:
Profit = Total CM – Advertising Costs
= Clicks * CR * CM per Unit – Clicks * CPC
In any keyword auction for this specific product, a merchant would spend at most $50. This is the point where marginal advertising costs equal the contribution margin of the product. If the merchant had to spend more, he would suffer a loss and therefore decide not to participate in the auction. However in the scenario demonstrated below, a profit can be realized on every single position.
2. Setting an profit maximizing bid for every keyword auction only optimizes the advertisers overall profit if the budget is not constrained or least sufficiently high.
3. Some in the search marketing community believe that conversion rates increase with lower positions, because those consumers who click on ads that are located on low positions have a stronger buying interest than users who click on the first few ads. However empirical analysis of conversion rates could not prove this effect yet (Kitts et al. 2006, Brooks 2005).
The merchant’s profit maximizing bid is between $0.17 and $0.32; the ad would be displayed on position 5 and the corresponding profit is $26.86. The revenue of the respective search engine is $7.94.
However, merchants typically do not have the capabilities to manage keyword auctions efficiently. Even if a merchant was able to determine the optimal bids in the previous example, the merchant could not exhaust the full potential of paid search, because they do not possess the required technology to generate and to bid on thousands of keywords. So this example is based on the assumption that a merchant has core capabilities in PPC marketing. We will use the results as a benchmark for the analysis of the bidding behaviour of affiliates.
Revenue sharing is currently the most advanced performance-based search marketing approach. Revenue sharing is the prevalent mechanism in almost every affiliate network, but is also facilitated by specialized search marketing companies and in some cases by marketing agencies. A company who manages paid search marketing on a revenue sharing basis will be referred to as a PPC marketing company in this section whether it is an affiliate, a specialized search marketing company or an agency.
PPC marketing companies that receive a revenue share of the product they sell, maximize their profit as follows:
Profit = Revenue Share – Advertising Costs
= Clicks * CR * Price * Share – Clicks * CPC
While a rational merchant would maximize the difference between contribution margin and advertising costs, the PPC marketing company maximizes the difference between the revenue share and advertising costs. Those companies would participate in any auction where the expected advertising costs are lower than the expected share of revenue. Thus the PPC marketing companies' profit maximizing calculations differ from those of merchants.
Table 2 shows the profit maximizing bidding behaviour of PPC marketing companies and the corresponding profit of the merchant and the affiliate. The following example is based on exactly the same data as the previous example. In addition we assume a realistic revenue share of 10%.
The profit maximizing position for the PPC marketing company is position 8. The optimal bid lies between $0.06 and $0.10. The maximum profit of the PPC marketing company is $2.44 and the merchant makes a profit of $16. Compared to the first scenario, where a merchant with core capabilities in PPC marketing targeted position 5, the PPC marketing company bids only for position 8.
Looking at the total profit, which is the sum of the profit of both partners, it becomes clear that the profit optimizing behaviour of the affiliate does not maximize the total profit. By bidding for position 5 the overall profit would be considerably higher. However, this would never happen in this scenario because the PPC marketing company would suffer a loss. The maximum possible profit will not be exploited because the PPC marketing company has an incentive to maximize their profit and not the profit of the merchant, or the overall profit.
In the above presented revenue sharing model, the merchant does not bear the risk of losing money when customers click on adverts and then do not follow through to purchase. In addition, the merchant takes advantage of the capabilities of specialized PPC marketing companies who realize the economies of scale. The merchant is therefore are able to acquire customers at a lower cost than by utilizing in-house campaign management.
Outsourcing paid search to highly specialized search marketing companies is, in most cases, better than in-house management. However, revenue sharing agreements are unable to exploit the full profit potential of the merchant as well asthePPC marketing company. In this section we suggest a new pricing mechanism which we call profit sharing. A PPC marketing company – in this section – is a company who co-operates with merchants on a profit sharing basis.
In a profit sharing model, PPC marketing companies maximize their profit as follows:
Profit = (Total CM – Advertising Costs) * Profit Share
= (Clicks * CR * CM per Unit – Clicks * CPC) * Profit Share
So in a profit sharing model, PPC marketing companies would, in the same way as a merchant with an in-house campaign, maximize the difference between contribution margin and advertising costs. With such a mechanism, these companies would participate in any auction where the expected advertising costs would be lower than the total expected contribution margin.
For the following example we again use exactly the same data as in the previous examples, and in addition assumed a profit share of 20%.
As in the first case of a merchant’s in-house campaign, position 5 is targeted; the profit maximizing bid is between $0.17 and $0.32. Compared to the example based on revenue sharing, the overall profit is higher. Depending on the percentage of profit sharing between the PPC marketing company and the merchant, a win-win situation would result. In the chosen example, both now earn a higher profit. Table 4 provides a comparison between both business models in this example.
In the profit sharing model, the profit maximizing calculation of the PPC marketing company maximizes its own profit as well as the merchant’s profit. The PPC marketing company bids on the same position as the merchant would have done so in an in-house campaign. In addition, more money is spent on advertising, so that search engines take advantage of the new business model.
It should be noted that not every auction behaves like the above examples. For instance, there are auctions for very uncommon or specific keywords, where only one bidder participates. In such a case, it usually does not matter if bidding occurs on a revenue sharing or on a profit sharing deal. There are other scenarios where the current model is equal to the suggested new model. However, there is no case that we have found where the current revenue sharing model is superior to the profit sharing model. Thus the profit sharing model is superior to the revenue sharing model.
In a revenue sharing business model, the whole risk is carried by the PPC marketing company. When affiliates start new campaigns, they usually need several weeks or even months to make them profitable. During this period they do not make profit. Instead, they lose money. In such cases, the bids of the PPC marketing company are too high to be profitable. However, this can cause a high sales volume which benefits the merchant who makes a profit for every product sold, whether the PPC marketing company loses money or not.
The risk distribution in a profit sharing model differs. Should the actual position deviate from the optimal positions (too low or too high), the overall profit decreases. While in the revenue sharing model, sub-optimal bidding (for example, in a campaign testing period where the bids and thus sales volumes are too low) only affects the merchant, in the profit sharing model both parties are affected whenever the PPC marketing company’s bids are sub-optimal.
In the profit sharing model, the affiliate bids for position 2 and not the optimal position 5 (see Table 4). This causes the affiliate a deficiency of $1.36 ($5,37 minus $4.01) compared to the profit on position 5. The decrease of profit for the merchant is $5.46 ($21.49 minus $16.03). So the risk of sub-optimal bidding is shared between the merchant and the affiliates. The party who earns the bigger profit share suffers greater damage from sub-optimal bidding.
If the affiliate targets position 5 in a revenue sharing model (also 3 positions over the optimal position), the merchant’s profit increases from $16.00 to $27.94 in positions over above the optimum. The PPC marketing company, however, would make a loss of $0.98 compared to a profit of $2.44.
The costs of sub-optimal bidding are shared between the PPC marketing company and the merchant up to the point where the advertising spending equals the total contribution margin. From this point on, the merchant does not lose or gain money, but the PPC marketing company bears the all the costs that exceed the total contribution margin.
So, does this mean that the merchant should prefer a revenue sharing model?
There are several reasons why a profit sharing model is still better:
- PPC marketing companies usually have lower financial resources than merchants. However, in the current revenue sharing model these marketing companies carry the whole risk. This implies a moderate and cautious bidding that tends to result in lower than optimal bids and corresponding positions. As shown, less than optimal bids affect the profit of both the PPC marketing company and the merchant.
- The effect of PPC marketing companies targeting a higher and possibly profit-maximizing position (and corresponding higher sales volume) can compensate for the merchant’s lower profit during the campaign testing period.
- Automated bidding systems are constantly improving. They allow approximately optimal bidding, which keeps the risk for the merchant at a low level.
- When the PPC marketing company’s loss during the testing period is expected to be too high, the marketing company stops managing campaigns for the merchant, even when the campaign could be profitable for both parties on a profit sharing basis.
A business model has to be found, where a PPC marketing company bases their decision on the same profit maximizing calculation as the merchant. This happens in every case where both the merchant and the PPC marketing company maximize the profit as the difference of contribution margin and advertising costs.
The sine qua non for profit maximizing is the alignment of advertising costs. The PPC marketing company would have an incentive to cheat the merchant by declaring higher costs than they actually had to bear. This would benefit the PPC marketing company at the cost of the merchant.
There are different ways to solve this problem. For instance, the merchant and the PPC marketing company could integrate reporting systems, so that the merchant has insight in the PPC marketing company’s advertising spending. In all cases where merchants have a direct relationship with PPC marketing companies this would not be problematic. But many merchants, such as Wal-Mart, Amazon or eBay, work with many different PPC companies which are approached through affiliate networks. In these cases, it would not be possible to integrate systems with every single affiliate. Another, simpler mechanism has to found.
In the following section we first provide a short analysis of current PPC search affiliate marketing business model. Then we explain how to adjust this business model to make it suitable for profit sharing.
Business models are perhaps the most discussed and least understood terms and aspects in the areas of eBusiness, eCommerce and eMarkets (Alt & Zimmerman 2001). Authors writing about business models have sometimes a completely different perception of what a business model is. While Magretta (2002) sees a business model simply as a story that explains how enterprises work, Osterwalder (2004) and Gordijn (2003) emphasize the model aspect of a business model. The probably most cited definition is from Timmers (1998, p.4), who sees a business model as “...an architecture for the product, service and information flows, including a description of the various business actors and their roles; and a description of the potential benefits for the various actors; and description of the sources of revenues.”
However, for the purpose of this paper it is not necessary to analyze every dimension of a business model. We will provide a brief description of the basic functioning and will mainly focus on the business model actors and their interrelationships.
Figure 1 shows the business model actors and interrelationships between them. Continuous lines indicate a strong relationship between two actors, dashed lines stand for a weak relationship. A strong relationship exists when two actors regularly exchange goods, services and money. In cases where those exchanges do not happen regularly or in cases where only information is exchanged and no money, we talk about a weak relationship.
A PPC affiliate subscribes to an affiliate network, through which he gains access to a wide range of different merchants whose products can be promoted via paid search. The affiliate selects a product or a product group and bids for the respective keywords on a search engine. When a customer clicks on an advert, the customer is directed to the merchant’s site, where he is tracked. In the case that the customer buys a product or a service, the merchant receives money from the customer and pays a share or a predetermined amount per purchase to the affiliate network provider, who again pays a share to the affiliate.
The affiliate makes a profit when the share of the revenue that is generated by his adverts exceeds the amount that he has to pay to the search engine for the clicks on the adverts; otherwise he loses money. The merchants makes a profit whenever a product is sold, given that the share he pays to the affiliates does not exceed the contribution margin of his products.
Despite many advantages of a revenue sharing pricing mechanism compared to Cost per Thousand Impressions (CPM) or Cost-per-Click (CPC), revenue sharing is not profit maximizing for PPC.
With direct contact between a PPC marketing company and a merchant, advertising costs have to be aligned for the functioning of a profit sharing business model in affiliate networks. System integration between merchants and many different affiliates could cause high transaction costs and therefore weaken one of the most important advantages of the PPC affiliate marketing business model. Far better would be to create a direct relationship between the affiliate network provider and search engines.
Google for instance could send invoices that correspond to certain affiliates and merchants to the affiliate network provider. The search engine may be considered as a neutral party without an incentive to declare higher costs. When the merchant specifies the contribution margin for every product sold, the affiliate network provider would be able to calculate the profit as total contribution margin less advertising costs and could pay a share of the generated profit to the affiliate. All the other relationships of the business model remain the same.
Thus, the implementation of such a mechanism requires only a slight modification of the current PPC affiliate marketing business model described in section 3.1. The main difference would be that search engines would have to create a relationship with affiliate network providers (see Figure 2).
The current revenue sharing pricing mechanism is not profit maximizing for none of all actors operating in the paid search industry. A profit sharing model is superior and allows a win-win situation, in which merchant and PPC marketing company are better off.
Applying profit sharing in large-scale to affiliate networks would have an noticeable impact on the overall PPC advertising spending and would benefit search engines who recently face decreasing growth rates. The current PPC affiliate marketing business model was originally designed for affiliate website publishers, who direct customers from their site to the sites of different merchants.
In such a case, it is simply not possible to adopt a profit sharing business model, because incurring costs of web design, programming, advertising, etc. cannot be assigned to a specific merchant. Afterwards this same business model was applied to PPC affiliates, where advertising costs can easily be assigned to specific merchants. A profit sharing model would be pareto-superior: in some keyword auctions it delivers the same result as the current revenue sharing model, in other cases it delivers a higher profit, but in no case it leads to a lower profit.
Figure 3: Evolution of Affiliate Marketing
Figure 3 shows different pricing models on a timescale. The suggested business model carries forward the trend to performance-based marketing.
Brooks, N. (2005):
The Atlas Rank Report: How Search Engine Rank Impacts Traffic, http://atlassolutions.com/pdf/RankReport.pdf
Gordijn, J. (2003):
Why Visualization of e-business Models Matters, 16th Bled Electronic Commerce Conference, Bled 2003
Kitts, B. et al. (2005):
A Formal Analysis of Search Auctions Including Predictions on Click Fraud and Bidding Tactics, In: Workshop on Sponsored Search Auctions, ACM Electronic Commerce, 2005
Kitts, B. & Leblanc, B. (2004):
Optimal Bidding on Keyword Auctions, In: Electronic Markets 14/3 (2004), 186-201
Magretta, J. (2002):
Why Business Models Matter, In: Harvard Business Review, May 2002, 3-8
Osterwalder, A. (2004):
The Business Model Ontology, Diss. HEC Lausanne 2004
Timmers, P. (1998):
Business Model for Electronic Commerce, In: Electronic Markets, Vol. 8/2 (1998), S. 3-8