In this fifth post in the ‘Successful Search Engine Marketing’ series, I’ll be giving my hints and tips on how to make sure you are paying the correct amount for the traffic you’re buying from search engines. Keyword, ad text and landing page optimization will all help you to run successful SEM campaigns, as will ensuring that you are buying your traffic from the right search engines. However, if you are not bidding the appropriate amounts for the keywords you are buying, you risk either missing out on valuable traffic, losing money on your SEM campaigns, or, in a worst case scenario, both.
The most basic rule of determining what you should be paying for SEM traffic is that it should not be more than the amount of revenue the traffic makes you in return. Failure to adhere to this principle can lead you to losing money on your campaigns very quickly! Having the appropriate tracking to enable you to monitor the revenue you are making and to associate this back to your marketing costs is therefore absolutely vital. Ideally, this link between cost and revenue should be maintained at the most granular level possible, so that you have the ability to see how a specific keyword is performing relative to its cost. However, when looking at cost versus revenue on a keyword level, always remember that you need to have enough data to be able to make a robust decision on the value of traffic from that keyword – 100 clicks is normally a good rule of thumb. This is a topic I’ll be returning to later in this post, and will pick it up again in the sixth and final post of the series.
On a related note, we have had many questions from publishers operating sites where eBay Partner Network is the major source of their revenue regarding how to tie SEM costs to Quality Click Pricing earnings. Under Quality Click Pricing, the lowest grain at which Earnings per Click (EPC) information is available is by Campaign. For most publishers buying SEM traffic, we would recommend setting up your search engine AdGroups to tie in closely with your eBay Partner Network campaigns, so that you can directly compare how much you pay for the traffic you drive to your eBay campaigns with the amount you earn from them. In order to optimize your CPCs for keywords within these campaigns, an additional technique you can use is to assign traffic which has been driven to your site from SEM a unique custom ID per keyword. You can then look in the Transaction Download report at the amount of revenue driven by each custom ID. Though the Quality Click Pricing algorithm contains many factors other than Winning Bid revenue, looking at this should give a pretty good indication of the relative performance of different keywords within a Campaign.
Valuing your keywords
Once you have the appropriate tracking in place, the next step to optimizing what you are paying for SEM traffic is to accurately value the average Revenue per Click (RPC) for each of your keywords. While at first glance this may appear a relatively simple task, it is worth investing the time to get this valuation right, since all other aspects of CPC optimization ultimately flow from this. Without having an accurate RPC valuation for a keyword, it is impossible to know whether you are paying more or less for traffic from that keyword than it is worth.
Valuing keywords for which you have a lot of data is definitely easier than trying to evaluate lower volume or new keywords. Although you can use statistical analysis to work out how confident you can be about the value of a keyword depending on the number of clicks the keyword has generated, as mentioned above the rule of thumb I would use is that if you have a keyword with more than 100 clicks, you can be confident that dividing the amount of revenue earned from those clicks by the number of clicks generated will give you a pretty accurate RPC. You can therefore base the RPC value you give to these keywords purely on their own performance.
For keywords which have generated fewer than 100 clicks, deriving an accurate RPC value is more of a challenge, but by no means impossible. The key for these keywords is to combine the limited information you have about an individual keyword with data on related keywords. For example, let’s say you have 50 clicks for the keyword ‘ipod nano 8gb black’, and those clicks have generated $10 of revenue at an RPC of $0.20. Your other ipod-related keywords have an RPC of $0.10 over thousands of clicks. In this example, you only have 50% of the clicks required on the keyword ‘ipod nano 8gb black’ to meet the 100 click threshold; you may therefore want to base 50% of the keyword’s RPC value on these clicks, and 50% on the average RPC for other ipod-related keywords. This would give you a value for the keyword ‘ipod nano 8gb black’ of $0.15. As you generate more clicks for that keyword over time, the percentage weighting you give to that keywords own clicks when valuing its RPC should increase (e.g. at 75 clicks, 75% of the weighting should be given to clicks from ‘ipod nano 8gb black’ and 25% to other ipod-related keywords), until, at 100 clicks, the keyword is able to stand alone for RPC valuation. Conversely, for a new keyword which has generated 0 clicks, you may want to use the RPC from related keywords entirely to generate an initial RPC valuation, and then revise that valuation over time as the keyword generates more clicks.
Part of the challenge here is determining the correct set of keywords with which to ‘top up’ your data on a low volume keyword. Keywords including the same individual words (e.g. keywords ‘ipod’, ‘ipod nano’ and ‘ipod nano 8gb black’ all include the word ‘ipod’), keywords in the same category (e.g. mp3 players) and keywords which drive traffic to products with a similar average selling price are three of the ‘top up’ techniques I have seen used with some success.
One final piece of information you should consider is that the value of your keywords will change over time. Seasonal trends and changes in the popularity of particular products will not only alter the number of users searching for products and clicking on related ads, it will also have an impact on the likelihood of those users who do click on your ad to make a purchase. For example, ‘go go hamsters or Zhu Zhu Pets as they were called in the US’ was a very popular keyword during the 2009 holiday season; twelve months earlier it would barely have registered on anyone’s radar, and who knows how it will fare in 2010. It is essential that you keep an eye on the performance of your keywords over time and adjust their RPCs accordingly in order to ensure you are paying the correct amount for your traffic.
CPC optimization basics – average margin-based bidding
Once you have appropriately valued the keywords in your portfolio, you can start to optimize the CPCs you are paying for your keywords based on their RPCs. The simplest method of doing this is to set an average margin percentage that you want to make from your SEM activity, and to set the CPCs for all of your keywords to try and achieve this same margin. For example, let’s say you want to achieve a margin of 10% on all your SEM activity. In this case, if the keyword ‘hockey stick’ had an RPC of $0.10, you would want to be paying a CPC of $0.09 (what I will refer to as your ‘Target CPC’, giving you a margin of $0.01 (or 10%).
One thing that makes this a little more complicated is that, with all the major search engines operating an auction-based system, the Max CPC you submit for a keyword is exactly that – a maximum that you could possibly pay for any one click on that term. What you actually end up paying for each click could be anything between the Max CPC you have submitted and the minimum CPC set by the search engine for your ad on that keyword, and will be determined by the Max CPC and Quality Score of your ad compared to those of other advertisers bidding for the same keyword. Taking the ‘hockey stick’ example above, if you placed a Max CPC of $0.09, in most cases you would end up paying less than the $0.09 Target CPC. While this would give you a margin above 10%, you would potentially be missing out on traffic you could have acquired at an acceptable margin had you bid slightly higher, since your $0.09 bid may have caused your ad to appear lower on the page than would have been the case had you submitted a Max CPC of $0.10 or $0.11 and actually paid your Target CPC of $0.09.
Estimating what Max CPC you need to bid in order to actually pay your Target CPC can be tricky. One simple but reasonably effective method is to look at what Max CPC you have historically bid for a keyword, and compare it to the average CPC you paid for that keyword over the same period of time. The percentage difference between the two can then be applied to your Target CPC to determine what Max CPC you should bid. To go back to the ‘hockey stick’ example, let’s say on average the CPC paid for that keyword was 75% of the Max CPC submitted. To work out what Max CPC you should bid for this keyword, you would need to divide the Target CPC ($0.09) by 0.75 to get to the appropriate Max CPC ($0.12).
Advanced CPC optimization – elasticity-based bidding
Optimizing your CPCs to achieve a flat margin percentage across all keywords is a great technique to use when you are initially trying to get to grips with your Campaigns, because it is relatively simple to implement, does not require a huge amount of data analysis (though it undoubtedly requires some), and compared to a portfolio of keywords with non-optimized CPCs, will get you a significant uplift. However, there are limitations to this approach in comparison to some of the more advanced CPC optimization techniques out there. The most widespread of these techniques involves bidding not just according to the value of the keywords in your portfolio, but also according to how elastic they are. While I am calling this ‘elasticity-based bidding’, you may see it referred to elsewhere as ‘portfolio optimization’, after similar techniques used by stock traders.
The theory behind elasticity-based bidding is pretty simple. When determining what CPC to bid for a keyword, you need to understand not just the value (RPC) of traffic from that keyword, but also what CPC you need to pay in order to get that traffic – in other words, how elastic (or inelastic) that keyword is. For example, let’s assume you have a golf website and want to buy two keywords: ‘Callaway big bertha’ and ‘Callaway big bertha diablo edge driver’, both of which have the same RPC. Many advertisers are bidding on the first keyword, while only one or two are bidding for the second. In this example, if you used average margin-based bidding, you would submit the same bid for both keywords. However, proponents of elasticity-based bidding would argue that this is not the right approach. Instead, they would recommend bidding less on ‘Callaway big bertha diablo edge driver’, since cutting the CPC for this keyword will make relatively little difference to where your ad appears on the page and therefore the amount of traffic (and revenue) it generates; and to bid higher on ‘Callaway big bertha’, where an increased CPC could make a big difference to your ad’s position, traffic and revenue. In other words, the more competitive keyword is likely to be more elastic than the less competitive keyword.
When applied across a whole portfolio of keywords, elasticity-based bidding becomes a lot more complicated. In order to make it work successfully, you need to understand as far as possible how many clicks each keyword will generate at each Max CPC you could possibly bid, what CPC you would actually end up paying for each of these clicks, and how much revenue each of these clicks would make you. Even this last part is not simple, since different keywords show different levels of elasticity in terms of RPC as well as CPC (for example, the RPC for some keywords becomes lower when their ads are shown in the first position on a search engine since the clicks they generate are less targeted than when their ads are shown in a lower position). Once you understand all of this, you can then work out the combination of keywords and CPCs which will generate the maximum possible of revenue you can generate from whatever your marketing budget is.
For most individuals running small SEM campaigns to drive traffic to their niche websites, the drawbacks to this approach are prohibitive. Firstly, you require a huge amount of data on your keywords to be able to get anything like an accurate picture of how elastic on both the cost and revenue side each keyword is. Secondly, being able to process, store and regularly update all of the necessary information is time consuming and can be expensive. And finally, performing the analysis to determine the optimal CPC for each keyword in order to maximize the return from your budget is far from simple! That said, for publishers buying a large amount of SEM traffic who have the data and resources to support it, elasticity-based bidding offers the considerable advantage of delivering higher revenue for a given level of spend. Indeed, if you can capture the data required for elasticity-based bidding, you can also calculate the incremental revenue you will get from each extra dollar spent on each keyword. This in turn allows you to set your Max CPCs for each keyword to where incremental revenue is equal to incremental spend, thus setting the optimal overall spend level for your SEM activity. Given the difficulty of implementing elasticity-based bidding successfully, if you are considering whether this approach might be right for you I would recommend speaking to some specialist SEM agencies, many of whom offer technical solutions to enable this type of optimization.
Bidding by Time of Day and Day of Week (ad scheduling)
Whichever bidding method you decide to utilize, one relatively simple technique which should help you to optimize your SEM spend even further is bidding different amounts for your keywords according to the time of the day or the day of the week, which Google terms ‘ad scheduling’. Depending on the content and audience of your website, you may find that clicks on SEM ads displayed on a particular day of the week or at a particular time of the day have a higher or lower propensity to convert than average. In my experience, these variances can be quite extreme (in many cases variations of more than 50% compared to the average). Varying the amount you bid for your keywords according to these day of week and time of day differences can therefore have a huge impact on the profitability of your campaigns.
Google’s ‘ad scheduling’ tool allows you to vary your bids by applying a multiplier for each time period and day to the CPCs already set for your keywords. For example, if you discover that between the hours of 10am and 2pm on Mondays, clicks from your SEM ads are worth 20% less than the weekly average, you would set up an ad schedule for that period of 0.8. In this instance, the Max CPC of a keyword normally bid at $0.10 would be adjusted down to $0.08 for that four hour time slot.
Controlling your budget
One final tip to help you get the most out of your SEM budget relates to how you control your overall SEM spend levels. This is particularly important if you have a fixed budget that you are looking to invest in SEM. Fluctuations in the number of people searching, the Quality Score of your ads and the amount of competition from other advertisers can all have a significant impact on the amount of money your SEM activity costs you, which can make keeping your spend in line with your budgets difficult. One of the most frequently-used options to keep this under control is the Campaign Daily Budget feature. For each of your Google campaigns you must set a maximum amount you are willing to spend per day. Once your spend for that day has reached this threshold, your ads will cease to show until the following day. Many advertisers choose to set these campaign budgets relatively low in order to reduce the chances of over-spending.
While setting a low cap on these campaign budgets is relatively simple and can be effective from a spend control point of view, it is definitely not the optimal method if you want to maximize your revenue from a given budget. To take a slightly exaggerated example – let’s say you have a campaign with only one keyword in it, and want to spend up to $100 per day on this campaign. One way to do this would be to set a Max CPC of $0.20 and limit the campaign budget to $100. Assuming there are enough searches for the keyword and that your ad is attractive enough, this might lead to your ad generating 667 clicks at an average CPC of $0.15, but your budget being entirely spent by 10am. Though Google gives advertisers the option of having their ads displayed evenly throughout the day, which would get rid of the issue of having your ad completely down for 14 out of 24 hours, assuming the same average CPC you would still be generating only 667 clicks per day.
If your main constraint is a fixed spend limit, a better method of managing to that limit is by adjusting your CPCs. Under this approach, you would set daily budgets for each of your campaigns high enough that you will never realistically reach them (in the example above, for instance, you may want to set the campaign daily budget to $10,000). You would then set your CPCs at a low enough level to ensure that you do not exceed your budget limit. In the example, above, you could probably cut your CPCs roughly in half and still spend approx $100 in a day, but generate twice as many clicks. This is something of a matter of trial and error, but after a few days of adjusting your CPCs, you will quickly get to the right balance to enable you to stay within your budgets, always have your ads online, and generate far more clicks at a lower CPC than you would by using the campaign budget function to throttle your spend.
Stay tuned for the final post in the ‘Successful Search Engine Marketing’ series, in which I’ll be talking about a theme which underlies all of the different topics covered in the first five posts – testing and optimization.
Chris Howard European eBay Partner Network team leader and one time eBay UK SEM manager