Calculating CPM and CPC for SEO Campaigns - Full Dissertation
Sean Mulholland, JWTSC SEO GroupINTRODUCTION
As Search Engine Marketers at JWT Specialized Communications we've noticed a few shortcomings in organic search engine optimization (SEO). For one, organic search is harder to measure compared to other search engine marketing methods such as Pay Per Click (PPC) advertising. Second, success is usually measured in your rank position and the number of Top 30 rankings on the search engines. But rankings don't really mean anything in and of themselves, and we've found that measurements of success sometimes fail to dig deeper than rankings. We feel that more depth is needed to really prove our value to our clients.
And speaking of clients...if you're on the client side keep reading! Not only will you learn a way to measure the value of your search provider, but you'll also be able to measure the value of search against other marketing and advertising methods. The results may surprise you.
So what is it that we've found? We've found a way to estimate the Cost per Mil (CPM, with 'mil' being Latin for 1000), Cost per Click (CPC), and ultimately Return on Investment (ROI) of organic search campaigns. At first glance this might appear to be nothing new. After all, can't a high powered analytics suite accomplish this? It can...for those who can afford such a service. Even entry-level in house analytics packages cost hundreds of dollars and usually require the additional costs of a dedicated server and setup expertise.
Another shortcoming is analytics packages can't predict the value of SEO beforehand, it can only report on results after the fact. Also, analytics do nothing to predict the visibility you have on search engines - a key element of your online brand strategy (and branding is an often overlooked value of search engine marketing).
HOW TO CALCULATE CPM AND CPC FOR SEO CAMPAIGNS
The first thing to do is choose a set of keywords. There are hundreds of articles and so called 'press releases' that cover this subject, but for SEO newbies the basic rundown is this: Keywords are the basis of search, what you type into a search box at Google is a keyword, and choosing and optimizing the proper keywords is essential. Now, whether this keyword list be a proposed list to optimize for, or an existing list you're currently optimizing with, the formula will still hold true.
Odds are good that in researching your keyword list you used WordTracker. If you don't know what that is, WordTracker is a tool that reports on the estimated searches per month for keywords. Import your list of keywords into WordTracker and do a competition search, it doesn't matter which search engine you use. See the 'Count' column? That's a major component of the formula. Take note of the count for each of your keywords (we copy and paste the table to Excel).
Next you need to know your web site's rankings for your keyword list. Web Position Gold 3 is the gold standard reporting software that most use, but if you prefer other methods feel free. For the newbies, Web Position Gold is a software program that will report on your positioning on search engines for a given keyword list.
Now, don't report on every search engine in the universe. It's unwieldy and it doesn't really help to know how you rank on a minor search engine in Romania. We stick to a core set of eight search engines which represent about 99% of searches done online.
AltaVista
AOL
Ask Jeeves
Google
Lycos
MSN
Netscape
Yahoo
You might choose a different set, especially if you're not in the USA, but for the purposes of this article those are the engines we'll be working with.
So now you need to know the search engine market share of each of these search engines. Not all engines are created equal, and as you can imagine search engines such as Google and Yahoo get many, many more searches than smaller engines like Lycos. Giving each engine the proper 'weight' is critical. Ultimately a #10 ranking on Google will outperform a #1 ranking on Lycos due to Google's traffic volume.
At the time of this writing, comScore qSearch reports the search engine market share for the above engines is:
| AltaVista | 00.1% |
| AOL | 08.9% |
| Ask Jeeves | 06.1% |
| 36.5% | |
| Lycos | 00.6% |
| MSN | 16.1% |
| Netscape | 00.1% |
| Yahoo | 30.6% |
| Total | 98.8% |
The other 1.2% of traffic comes from the thousands of niche search engines out there. As you can see, these eight engines represent the vast majority of searches, and having only eight engines makes life easy. And we like easy. Ultimately we can claim my estimate has 98.8% accuracy.
SO YOU'VE GATHERED THE DATA, NOW WHAT?
Now we get to the good stuff, turning all of the above into a search marketers' favorite metrics: CPM, CPC, and ROI! This is a little bit heavy, so don't say we didn't warn you.
So remember the 'Count' data you gathered for your keywords? Get it ready. The count data is WordTracker's count of searches performed on a keyword in the past 110 days. Here's what it means as stated on the WordTracker site, but if it's a bit confusing we don't blame you. Feel free to skip it over and we'll break it down for you:
|
How many searches are performed on the internet search engines each day? This used to be a much more difficult question to answer. However since we concluded a deal in June 2001 for all of Metacrawler/Dogpile data this task has been made somewhat easier. StatMarket informs us that the Metacrawler and Dogpile engines account for approximately 0.86% of all search engine queries. The total number of queries in our current data set is 316054297 for the last 110 days. If we extrapolate this information, then 100% of all search engine queries per day is ((100 / Metacrawler Share) x Total Queries In Dataset) / Days In Dataset, = ((100 / 0.86) x 316054297) / 110 which equates to 334,095,451 internet searches on all engines per day. Please also see our study in previous article which estimates 240,000,000 searches/day. So how do we calculate the predict? We have calculated that there are roughly 334,095,451 internet searches on all engines per day. E.g. if 'test' gets a count of 9,056 in our dataset over last days, then it's daily internet prediction is: (Total Searches Per Day For Internet As Whole / Total Queries In Dataset) x No Of Occurrences Of Keyword In Current Dataset = (334,095,451 / 316,054,297) x 9056 = 9,573 searches. |
Does that make sense? Don't worry, it took a couple of reads too. Basically it all boils down to is this:
Daily Searches = Count x 1.057
Now that is MUCH easier to digest. But note that those are daily searches. Since SEO programs are typically measured monthly, multiply the result by 30 to get the monthly search info. With this, the formula becomes:
Monthly Searches = Count x 31.712
This formula gives you the keyword searches per month. Because WordTracker constantly updates its data, this figure will change slightly from time to time. Go ahead and apply this to all your keywords.
So now you have the keyword searches per month for your keywords! This is useful. To make it even more useful, we have to break it out by search engine market share. Remember the market share table above? What that means is that of the total monthly searches for a given keyword, 0.1% of those searches were performed on AltaVista, 8.9% performed on AOL, etc... For example, if a given keyword gets 100 searches per month, 0.1 will be performed on Altavista, 8.9 will be performed on AOL, 5.5 on Ask Jeeves, etc... By applying the search engine market share percentages to your keyword searches per month data you'll have your keyword searches per month per search engine. Apply this to all your keywords.
So now you have the monthly searches for a given keyword on a given engine. Now you're going to want to see how rank affects the value of your keywords. Before we do this, we've compiled some data from studies by Enquiro. The two studies are 'Inside the Searcher Mind', which gives some interesting figures relating to how many users click to the second and third page of search results. The second study, 'The Golden Triangle', used eye tracking technology to actually track where users looked on a search page.
The 'Inside the Searcher Mind' study researched user search engine usage during the buying process. What Enquiro found was that only 20% of users will click to the second page of search results (Ranks 11-20), and only 1% visit the third page of results (Ranks 21-30).
In 'The Golden Triangle' study Enquiro employed EyeTools' eye tracking technology to actually see where users were looking when searching on major search engines. The results showed a 'golden triangle' in which the top search results were looked at by nearly every searcher, with fewer and fewer going lower in the search results. We averaged the exact percentages for simplicity and will assume that 96.25% of users will view the top four rankings on any given page (Ranks 1-4, 11-14, 21-24, typically referred to as 'above the fold'), and 40% will view the next six rankings (Ranks 5-10, 15-20, 25-30, typically referred to as 'below the fold').
What that means is that 96.25% of searchers will likely see the top four results on Page 1, and 40% will see the next six results on Page 1. What about pages 2 and 3? Well, since only about 20% of users will bother clicking to Page 2 and only about 1% will visit Page 3, we have to take that into consideration. Sparing you the math, please reference the table below to see what percentage of users will see your results based on your ranking:
| Rank | Percentage of users who view the result. |
| 1-4 | 96.25% |
| 5-10 | 40.00% |
| 11-14 | 19.25% |
| 15-20 | 08.00% |
| 21-24 | 00.9625% |
| 25-30 | 00.40% |
As you can see, higher ranks are better, but now they have been quantified to show why they are better.
So now you have the searches per month by search engine for each of your keywords, and a table showing you how many searchers are likely to actually see your results based on your rank. Remember your Web Position Gold 3 rank report we told you to run? Get it ready.
What you need to do now is apply the above percentages to your keyword searches per month per search engine. Go through your Web Position Gold reports and based on your rank, reduce the keyword searches per month per search engine data appropriately. So, if you rank somewhere between 1-4 on AltaVista, you would reduce the keyword searches per month per search engine data for AltaVista to 96.25%. If you ranked between 5-10 on Google you would reduce the keyword searches per month per search engine data for Google to 40%. Do this for all keywords and all search engines (helpful hint: use Excel to make this much easier).
By applying those percentages to your keyword searches per month per search engine you'll have a pretty good estimate of how many people viewed your listing on a search engine. By summing the result for a keyword across all search engines you'll have a good idea of how valuable that keyword is. By doing this for all your keywords and aggregating the results you can get some pretty nifty numbers that allow you to look at your SEO campaign from a completely different angle.
But it doesn't stop there. If you determine the average monthly cost of your SEO campaign you can then divide out to find CPM metrics. If you further apply assumed clickthrough rates to the CPM data you can predict costs on a CPC basis. Doing so for one of our clients proved that SEO was very cost effective and able to outperform PPC advertising in terms of cost. But then consider that SEO campaigns persist even after we stop optimizing - with PPC, once you stop paying all your stop as well.
We bet you're wondering how such a long and involved process can be automated. Well, we've developed a series of scripts that integrate with Microsft Excel that factor in changes in search engine market share, searches per month, rank, etc. Contact me if you're interested in learning more about it.
We welcome any questions or comments, and would be glad to help you understand the formula more completely. Feel free to comment on our blog or email me.
ABOUT THE AUTHOR
Sean Mulholland (click here to contact him) is a Cybernaut, Innovator, and Problem Solver at JWT Specialized Communications, where he has been developing strategies for marketing and recruiting through search engines since 2002. Please email him or visit http://www.jwtsc-seo.com/ for more information regarding SEO and SEM related topics.