I've just taken over management of an AdWords account where all campaigns have been set for "any" position, but by changing Position Preference to 2-10+ (meaning that I prefer not to have my ad shown in first position), I've reduced the cost and increased ROI in just a couple of weeks.
I first wrote about this in 2001 (below is a copy of my 2001 post on this topic to the I-Sales discussion list).
TOPIC: CONVERSION RATES FROM PPC SEARCH ENGINES
From: Mark J. Welch
Gary Salzman wrote:
> "Are there any studies, opinions or predictions on the cost difference in relationship to number of sales vs the cost of being ranked 1, 2, 3, etc on Overture? Such as Rank 1 costs 2.00/click, Rank 2 $1.99, Rank 3 $1.55, Rank 4 $1.54." <
I'm glad you asked this question.
The Overture bidder in position #1 for a specific term, will almost always draw more clicks than any other listing. But the ROI from those #1 clicks is lower than the ROI if the same listing appeared, at the same price, as listing #2.
I call this the "idiot click" problem, although a fairer term might be the "lazy searcher" problem.
The most extreme example I can offer is someone who searches for "Lolita" because they are looking for kiddie porn. The first listing (at one time) was a one-cent-per-click bid by MovieGoods.com, which sells movie posters and stocked the movie poster for the film "Lolita." The listing title was "Buy 'Lolita' Movie Poster from MovieGoods' and the description was "MovieGoods stocks 77,000 movie posters, celebrity photos, and movie memorabilia, including movie posters from the film 'Lolita.' "
None of the kiddie-porn seekers read the title: they just went ahead and clicked on the first link, discovered it was a page selling movie posters, and abandoned the site. We drew about 100 clicks per day, at a cost of $1 per day, from this one search term, and of course nobody bought anything. We deleted that listing.
Now, move on to a more practical example: MovieGoods sells movie posters, and thus it should not surprise anyone that it bids on the search term "movie posters." When I first launched the MovieGoods campaign at Overture (in early 2000), I bid on that term at various rates from $0.05 to $0.85, and the bid needed to hold the #1 position varied from $0.29 to $0.86.
One issue is that during the course of a day, our $0.33 bid might be in the #1 position for a while, and then someone outbid us and we found ourselves in #2 and as new bids come in we might get pushed down to #2 or #3. Thus, it's hard to know what position the term was in, when it was clicked.
But over time, I was able to analyze the difference from being in position #1 vs. #2 or #3 or "lower." And because of the "idiot click" problem, I found that we could actually bid HIGHER for position #2 than for position #1.
I can't share the "actual" figures, but here are some "ballpark" numbers that show the analysis.
First and foremost, my goal was to acquire customers through Overture while spending no more than 20% of gross sales on our bids. Thus, if I bid $0.33 per click, I had to be confident that on average, each clickthrough would generate $1.66 in sales -- or as a practical matter, if we paid $0.33 per click and our average Overture-driven transaction was $38.00, we had to see one sale from every 23 clicks).
If the MovieGoods listing appeared in position #1, we drew a lot more clicks, but the conversion rate was below 4%, or about 1 in 23 clicks. At that price, assuming an average transaction size of $38, we could only afford to bid $0.33 per click to hold position #1 at Overture. At any higher bid price, we did not generate enough sales to justify the spending. But if the MovieGoods listing appeared in position #2, our conversion rates were higher. We drew fewer clicks (and spent less money at Overture as a result), but instead of converting 1 out of 23 clickthroughs into sales, we converted about 1 in 18 clickthroughs to sales. As a result, we could profitably spend $0.43 cents per clickthrough to be in position #2, but if the #1 bidder dropped out, we had to lower our bid so we were either in position #2 again, or in #1 at $0.33 or less.
So to sum up: we could profitably pay up to 43 cents per click for ANY position other than #1, but we could only pay up to 33 cents per click for the #1 position at Overture. (The math has changed since early 2000, and since I quit as Chief Strategist at MovieGoods, I don't have access to the current numbers, but it appears that the company is bidding a little bit higher.)
Naturally, every type of search term will work differently. For example, MovieGoods sells movie posters, and sometimes bids at Overture on "movie titles" or "actor/actress names" (e.g. "Star Wars" or "Sandra Bullock"). For those terms, we almost never wanted to be in position #1 because most people who search for those terms are not looking for movie posters, and we absolutely did not want any "idiot clicks."
However, when bidding on a more specific term like "Star Wars movie poster" or "Fred McMurray pictures," MovieGoods is comfortable as the #1 bidder because the search term indicates a strong interest in the stuff MovieGoods actually sells.
At the opposite extreme, we found that if we bid for any position, even at one cent per click, on terms like "movie review" or "movie trailer," the bid was unprofitable.
Another way of looking at this is to admit that MovieGoods' products are "bought, not sold." If someone is looking for information about the Spider-Man movie, it's quite unlikely that we can convince them to buy the Spider-Man movie poster -- we just can't "sell" well enough to justify bidding aggressively on the movie title alone.
However, is someone searches for "Spider-Man movie poster," then they are showing a strong "buying" interest and so we want to appear highly-ranked in response to that search phrase (currently I see MovieGoods is bidding $0.16 and is in position #2 for that search term at Overture).
Some people are confused because they look at traffic levels. For the same search term, a 43-cent bid for position #1 might draw 500 clicks per day, while a 42-cent bid for position #2 might draw only 220 clicks. But the real measure is the "cost per dollar of profit," which should be less than a dollar. Most of us use a surrogate, such as "cost per dollar of sales" (and at MovieGoods, my benchmark was that I could spend only 20 cents per dollar of sales generated, assuming an average 20% profit margin, which of course was unfair since some products had higher margins and some loss leaders had less margin).
Finally, when "managing" your Overture campaigns, it's important to use your titles and descriptions not only to "sell" or "encourage" clicks, but also to "pre-qualify" clicks. Including price information, or the word "buy," will discourage clicks from people looking for cheap or free stuff (but in position #1 you'll still get those "idiot clicks").
And since not all of Overture's partners display the complete title or the complete description, it's important to put that pre-qualifying language at the beginning, not the end, of your title or description.
Mark J. Welch (2001)