Cut Your Costs 90% by Scaling Laterally Across Audiences

Say you have a campaign that’s been moderately successful for some time. Your creatives and targeting are relevant to your audience, and you’re getting a steady flow of conversions profitably. But, after some time, your campaign will invariably being to saturate the market. You’ll see conversion rates begin to drop and costs slowly rise, while traffic remains flat. This happens to everyone. Online marketing wouldn’t be much fun if we could just throw up a single successful campaign and sit on our asses collecting checks for the rest of our dats.

How do you scale the campaign up to more traffic and stop the regression to lower profits?

One way might be to expand your ad groups or targeting and find more keywords that will do as well as your existing ad groups.

Finding lateral keywords(keywords that describe the same term in a different way)  has been a well documented and successful strategy, especially in the early days of search marketing, when a marketer armed with a thesaurus and a good imagination could build massive campaigns.

But adding more keywords to a campaign is merely a band-aid, a short term fix that won’t solve the core problem of an audience that has gotten tired of what you’re selling.

I’d like to describe a new strategy that I have used very successfully to grow new campaigns as well as breathe new life into stagnant old ones. I call this strategy scaling to lateral audiences.

Just as lateral keywords are closely semantically related to the original keyword (i.e “meet single women” and “online dating sites”), lateral audiences are closely related based on their core attributes- their fundamental needs, desires, and problems.

In other words, if your product solves a problem for a specific group of people, thinking in terms of laterally related audiences would help you find more people like them, that also have a need for your product.

To illustrate the power of lateral audiences, we’re going to walk through an example using a free MixRank account.

Just to take a random example I made up, let’s pretend we’re a company selling gold coins- a growing industry in this economy. This is a very competitive space with lots of advertisers, so we’re going to have to get creative if we want to take some of their traffic for ourselves.

Google’s Traffic Estimator shows that the keyword “buy gold coins” has a very high average CPC for position 1 of $10.10. To get this data, I set the average CPC in the Traffic Estimator to an absurdly high number like $1000 to make sure I get the absolute highest bid for this keyword.

Let’s find a way to get a similar audience, one who’s interested in buying gold, less expensively. All we have to do to start is initiate a search from the MixRank home page and find a relevant advertiser whose strategies we can study. Let’s just search for our keyword and click on the first suggestion:


The search results will show us a few ads that are highly relevant to the keyword we searched for. I can already see one lateral audience here- several of the ads say “Buy silver coins”.

It’s important to be mindful of the distinction between keywords and audiences here. Lateral keywords for this theme would be phrases like “buy gold bullion” or “buy gold bars”. In other words, they describe the same product in a different way. “Buy silver coins” describes a different product but targets a lateral audience that has similar fundamental desires- in this case the desire to own precious metal. This desire could further be reduced to a core drive for security, financial stability, greed, etc.

The keyword “buy silver coins” has an average CPC of $4.92. Still high, but a significant improvement in targeting the same audience.

Let’s find out which core desire it is based on the current advertising strategies of the market’s leaders. From the search results page, I’m going to pick what looks to be the current leader in this space, “”. I can see more data about them by clicking on that domain in the “Advertiser” column on the far right.

The Advertiser Report for will show me their highest performing ads and traffic sources. You can also reach this report simply by typing their domain, “” into any search box.

Looking through their text ads, I’m noticing a common phrase that’s consistent across all of their split tests: “Free Investor Kit”:


They’ve probably tested many different positioning strategies and found that presenting gold as an investment is the strongest appeal. Here’s another lateral niche audience: people who are looking to buy gold as an investment (as opposed to collectors seeking gold for numismatic purposes, etc). Google Traffic Estimator shows that the keyword “invest in gold” has a max CPC of $8.87. This high number is encouraging, because it means that this is a valuable, high converting audience.

$8.87 is a bit rich for us- if we were running a campaign targeting this theme, I would probably see my margins plummet as I get squeezed out by competitors with bigger budgets.

But remember our other, related audience of silver buyers? “buy silver coins” was significantly cheaper than “buy gold coins”, so I would expect this pattern to hold across other, related keywords centered around buying silver.

Indeed, “invest in silver” has a maximum CPC of $3.71, which is a huge 58% discount from the gold keyword, yet targeting an audience that’s closely related to the original keyword, and one we can be reasonably sure will convert just as well, because they’re interested in buying precious metals as an investment.

But let’s keep going and see if we can cut our traffic costs even further using MixRank’s database of millions of ads.

MixRank’s ad search uses sophisticated matching algorithms that go beyond simply looking for the appearance of keywords in ad copy or landing pages and identifies campaigns that are thematically relevant to the query. For a great example of this, let’s search for our new keyword that we derived from the ad copy we saw running- “investing in gold”.

Our goal with these searches is to identify keywords and audiences that don’t match our search query exactly, but are somehow related.

There’s one result that jumps out at me immediately.


Of course! Some people that are interested in investing in gold are part of the small, but lucrative niche audience of people stocking up for an impending economic collapse. The Ron Paul audience, if you will. Analyzing the ads of advertisers like this one will give us great insight into this market. Let’s make the assumption that people anticipating an economic collapse are highly motivated to turn their paper dollars into gold, which we will be happy to sell to them.

The keyword “economic collapse”, which features prominently in these ads, has a suggested max CPC of $1.02, a 90% discount on our original keyword of “buy gold coins”, which cost over $10 a click.

An inexperienced marketer will suggest that “economic collapse” is a bad keyword to target, because it doesn’t show intent and is not a “buying keyword”, so it will not convert as well. But remember, we’re getting this traffic 10 times cheaper!

Let’s say you have a stagnant campaign based around the “buy gold coins” theme that’s barely breaking even. If you target the “economic collapse” keyword, all you have to do is achieve 1/10th your current conversion rate from this audience to get a huge bump in traffic and profits.

“Prepare for impending economic collapse- buy gold!”. The ads practically write themselves.

No thesaurus or keyword tool will tell us that new main keyword, “economic collapse”, is strongly correlated with the keyword we started with, “buy gold coins”. But by using MixRank to pull relevant keywords out of ad copy and searching for thematically relevant ads matching those new keywords, we can quickly identify pockets of opportunity that advertisers without the benefit of this data will miss.

In a later post, I’ll show you how to delve even deeper to identify even less expensive, high converting audiences and leveraging them for a flood of massive traffic. But following the strategy outlined above should be enough to get you started scaling your campaigns across lateral audiences very quickly.

Google Hack Lets You Include Images in Search Ads

One of the most overlooked features of Apple’s recent OS X 10.7 Lion release is support for emoji characters. Emoji is an obscure set of emoticons that’s quite popular in Japan. In Lion, you can insert Emoji characters like smiley faces into any text field, much like the Wingdings font on Windows.

My experience with these special characters has been a little different. The tactic of inserting special characters in an ad to draw attention to is and make it stand out from the rest has been used very effectively by advertisers in the early days of Google, Facebook Ads, and pretty much every advertising platform out there.

I’ve seen firsthand what a difference including a special character in an ad headline makes. A single arrow pictogram used in a small text ad can double, or even triple CTR.

Google has of course long since wised up to this practice. Any ad that contains a character from a predefined blacklist of special characters is automatically flagged for manual review and promptly denied by Google’s approvals department.

But what about the newfangled emoji? Are those blacklisted as well? To test this, I submitted two ads- one using an older special characters font and one using emoji.

See the difference:


The ad with plain old special characters is promptly flagged for manual review on it way to be approved for denial.

But the ad with new emoji characters sails right through automated review, and is instantly eligible for release on an unsuspecting public.

Imagine how much an ad containing graphic icons will stand out on a search results page containing only drab text ads… I bet any such ad would completely obliterate and outrank any text ads it’s competing against.

Of course these images will only render on operating systems that natively support emoji. Right now, that’s limited to Mac OS X Lion and iOS on iPhone and iPad.

So, if you launch a standard search campaign, ~5-10% of the traffic that has OS X Lion will see the Emoji icons in your ad, which may or may not be enough to give a significant boost to your CTR.

But there’s another popular operating system made by Apple that supports rendering emoji. I’m talking, of course, about iOS, and the millions of people searching on their iPhones and iPads. And, conveniently enough, Google lets us target iOS users only when setting up a new AdWords campaign:


The homogeneity of that platform means that you’re virtually guaranteed that 100% of searchers will see your ads and will click, if only out of curiousity at this new ad format. The intrinsically low cost of traffic and dearth of advertisers on mobile mean that, using the emoji trick with broad targeting, you can get massive volume at pennies a click.

The amount of extremely cheap traffic you could get with this method is staggering.

Now…say for some reason your ad is subjected to manual review(for example, if you use this trick on the Google Content Network). No need to be concerned. What browser do you think the Google reviewers are using? It’s a lot more likely that they’re using Chrome than Safari, isn’t it?

To Chrome users, even on OS X 10.7, an emoji emoticon appears as an inconspicious blank space. Move along Google reviewer, nothing to see here…

Inserting emoji into any text field on OS X Lion is easy- just click Edit ->Special Characters and select Emoji in the left sidebar.

My guess is that this hole won’t remain open for long- so make the best of it while you can.

Props to Panic for taking this to the next level with an emoji domain. I wonder if this domain can be used as an AdWords display URL…

How to Significantly Decrease PPC Click Costs

Most paid traffic campaigns you launch will start off losing money; that’s just a fact of marketing you will have to accept. The long term profits will come later, after you’ve optimized the campaign by removing poorly performing keywords and traffic sources, and increasing bids on high converting ones.But of course it’s always painful to keep running a campaign that is losing money every day, even if you know it will become profitable eventually. So let’s talk about ways to decrease costs as soon as possible and get closer to profitability without blowing through your budget.

Optimize for an Easier Conversion Target

As you know, you need to wait until you get to ~30 actions before you can get enough statistically significant data to make a decision. If you’re selling physical products, you might have to wait a pretty long time before getting to 30 sales for every single ad. This goes directly against our prime objective of building many campaigns and learning fast. To get meaningful data faster, it might make sense to make your initial conversion goal something with a higher conversion rate, like an email opt-in on a squeeze page or even a click through to your shopping cart or sales page.It’s a lot easier and cheaper to get 30 emails than 30 sales (or even long form leads). Optimize for those, and you’ll be able to see what works and what doesn’t much earlier.With me so far? How about optimizing for time on site or bounce rate? Very few advertisers do this, but the ones that can do this successfully are reaping the rewards.This is especially important for CPV/popup traffic, where even small differences in time on site may be a good leading indicator of conversions.If you’re direct linking to an affiliate offer, it may be worth it to switch to a landing page, just so you can collect this type of data and optimize for clickthroughs.This kind of optimization becomes a lot easier if you have good numbers about your entire sales funnel. If you know that 10% of email opt-ins consistently convert into sales or that your EPC on clicks from your landing page is always around $0.50, you’re not risking much by focusing on clicks instead of conversions.

Cut off Losers Fast

Earlier I said that you should always wait until you get to 30 actions before making a decision about a traffic source. This is true if you have an infinite budget in a market with perfect information. But since we don’t, it may be worthwhile to cut a few corners.If a campaign you just launched isn’t converting at all, it might make sense to kill it early, before you have spent lots of money testing, and move on to greener pastures.Sure, you’ll miss out on a few good keywords due to random chance. But accepting a higher incidence of Type II errors may enable a good chunk of your budget to live to fight another day.This is especially important for traffic sources like Facebook Ads or RON campaigns anywhere, which can devour budgets very quickly if you let them.

Optimize Your Landing Pages with Cheaper Traffic Sources

It’s certainly true that different landing pages convert differently with different traffic.But some factors, especially your headlines and calls to action should affect conversions in a somewhat consistent way, regardless of traffic source.When you’re in the early stages of testing landing pages and collecting data for a campaign, try to do it as cheaply as possible. That means getting traffic from Bing/Yahoo through AdCenter, going to PPV networks with lower minimum bids, targeting a country with cheaper traffic like Canada, and so on.Then, when your landing pages have been split tested and optimized as much as possible, you can move over to a more expensive but higher volume traffic source like AdWords.See also this excellent post by Finch on fighting back against rising click costs.