Thomas Edison vigorously tested everything.
The story goes he tried 10,000 different ways to creating artificial light before he invented the light bulb as we know it. There were a few electric lamps in the 19th century, but they all had flaws.
The challenge was making lamps that required low voltage, low manufacturing costs and most of all, could produce light for more than a few hours.
The man was intense.
In 1878, he began serious research. After two years of hard work, trial and error, he created a bulb that lasted for 1,500 hours, required far less electrical power and cost far less to produce than its rivals.
Edison’s success came after many episodes of failure – much more than what you’d call a fair share. The first bulb Edison ever made lasted a few minutes. But he never stopped.
In his attempt to creating a bulb that will last longer, Edison kept testing filaments made of various materials – filaments from all metals and even carbonized filaments of plant materials.
The man was insane with experimentation.
According to Edison himself, he tested no fewer than 6,000 different plant materials and ransacked the whole world for the most suitable filament material to use.
Edison neared success when he decided to try a carbonized cotton thread filament, which burned for just 15 hours. But it showed signs it could last longer, so he kept going. Further experimentation, one after the other, produced filaments that could burn longer and longer with each test.
Eventually, Edison produced a 16-watt bulb that could last for 62 days.
He was successful.
Following that same attitude of persistence and experimentation even after what seemed like failures, Edison came up with over 1,000 other inventions – including the phonograph, movie camera, microphone, radio, kinetoscope, cement, and others.
Edison’s experiments led to tremendous results.
When was the last time your tested something new in your AdWords account?
Your main challenge is reducing costs and increasing conversions.
If you’re not testing fast enough, you’re not evolving fast enough.
You don’t have to change the world, but you can change your bottom line. However fat it already is, make it fatter.
Thomas Edison didn’t invent the light bulb. He just drastically improved performance. Just like you can methodically and with persistence improve your AdWords account performance. So long as you never stop testing.
There’s no one master blueprint that works for all businesses, all markets or all Adwords accounts. The tweaks that worked like wild fire for Business A may bring nothing but losses to yours.
You can find your own magical tweaks. The subtleties that make your business success unique. And you can keep them to yourself for an unfair advantage over your competitors.
Some of your AdWords campaigns may have failed woefully in the past. This does not mean that they don’t work or that they will never work for your business. Just because something doesn’t do what you planned it to do doesn’t mean it’s useless.
A common example is the display network. A beast hard to tame for new advertisers. They quit before ripping the benefits that experimentation can yield. They say “I tried it and got lots of clicks with no conversions”. Well, you tried it wrong once and failed once. Make smaller test (so you lose less money) but more frequent test (for faster discoveries) until you succeed. While most unaware advertisers hate it, the display network is my favorite channel. That’s because I’ve tested so many different approaches that I’ve figured out what works for me and what doesn’t.
What differentiates experts from beginners is the ability to conduct meaningful experiments fast and gathering enough data to use on further experimentation. Always iterating. Tirelessly.
You won’t have to try 6,000 different solutions like Edison did. You may find your magic wand after some attempts – or even just one.
The 3 experiments you should always be running at any given time.
Campaigns determine when and how you ads show up. Ads are triggered by keywords and are the bridge between your website and Google. Landing pages lead people who’ve clicked your ad to perform an action that will (hopefully) make you a wealthier individual.
You should never stop testing those three: campaigns, ads, landing pages.
(I would add keywords and ad groups but those are implied in campaign tests — so are bid strategies, geographical locations and all other campaign settings.)
If there’s one of those elements not being tested right now in your account, you’re not pushing it hard enough. Excluding yourself from making discoveries that could change your world. If you don’t have time for it, hire someone to do it for you. Edison didn’t do it himself, he had a team supporting him.
I. Ad copy experiments.
The most basic experiment you can conduct. It’s so easy you have no excuse for not having at least (and “at most” for newbies) two ads running in every ad group at any given time. When I have a new ad group I’ll test a new ad every 100 impressions. On well-performing ad groups, I’ll test a new ad every day. And on highly performing campaigns I may test less often if an ad has been running successfully for long and it’s pretty hard to beat.
When you have a Tenscores account, you get 3 major recommendations for increasing quality score: a keyword clustering tool, add exact matches when necessary, to test new ads. Then it tells you to spend 90% of your time writing new ads. Because it’s important. Learn what kind of ads get the best CTR in Adwords.
II. Landing page experiments.
The most lucrative experiment you can make in my opinion. It’s not hard to double landing page conversion rates. When you do, your cost of acquisition will be cut in half.
Setting up brand new landing pages can take some time and effort, but while you prepare a big new test, nothing stops you from testing small things like button colours, calls to action, new headline, etc.
Anytime I don’t have a landing page being split-tested I feel very uncomfortable. So I always have a landing page test running at any given time. Here’s an example page I tested recently as our home page (don’t get offended, it’s just humor). There’s a test running right now on our own home page and it’s already promising to double our conversion rates. Soon as it’s done, I’ll run another… then another… then another. Do whatever you have to do, hire designers, or prioritize your own work, whatever it is… always make sure you’re testing your landing pages.
III. Campaign experiments.
The most under-used feature in Adwords I beleive. I’ve only seen seen seasoned advertisers use it and I’m guilty of not using it enough myself. For that reason, I won’t cruxify you today if you have none running right now. However, I will tomorrow if you still have none.
You can influence the amount of traffic you get in an ad group by adding keywords or increasing bids. It could have a positive impact. It could have a negative one. A quick campaign experiment will tell you which it is. And you can define how much traffic to allocate to the test — say 10% for example — lowering the risk of a negative impact.
Campaign Experiments (ACE).
One of the most effective and under-appreciated AdWords feature .
“We improved our cost-per-conversion by 50% using ACE to experiment a 33% lower bid.”
I had a particular client who wasn’t really happy with their cost-per-conversion.
They had a long standing account that had been optimized pretty well but cost-per-conversion were falling off. I suspected that maybe they were over-bidding a little bit. But I didn’t want to loose them sales because they were getting a decent amount of sales, so we used a campaign experiments.
I lowered the the bids by 33% and we found that their cost-per-conversion improved by about 50% and they really didn’t loose any sales at all. It’s really very powerful for bidding and we were able to get statistical significance in about a week. A very quick turn around.
Search engine marketing professional
Listen to the full audio:
You can test virtually everything with campaign experiments: ads, landing pages, bids, match types, negative keywords, placements, site exclusions and the list goes on. But that DOES NOT mean you should test everything with ACE.
What to test first with Campaign Experiments:
I. Higher or lower bids.
Experiment with higher bids to see if higher Ad Rank will lead to more conversions without impacting their cost. Experiment with lower bids to see if you can reduce cost-per-conversion while maintaining the same level of sales & lead flow.
II. Risky keywords.
You don’t have to test each and every new keyword you add to your campaigns. But those risky ones, like that 1-word broad match you’re not sure about, ACE will help you gauge it before deciding to keep it. You’re not sure about a negative keyword? Test it. You’re not sure about a match type? Test it. And of course, test intelligently. Some changes can be applied without launching a full experiment.
III. A new domain name.
Big brands can’t do this and they probably shouldn’t. But smaller advertisers can and that’s a significant advantage. At this level, you’re not just testing new landing pages, you’re testing mini-sites living on their own domain.
It’s a forgotten truth about PPC ads: domain names have a huge impact on your CTR and conversions. And I’m not referring to the display url, I mean the domain name itself. Sometimes the domain name is more powerful than the headline.
There was a time when it wasn’t forbidden to write 2 ads with different domain names in the same ad group. We would test the hell out of different domain names that we didn’t own. For those of us who know, the domain name part of the display URL can make or break a campaign. Well, with ACE you can test a different domain name, but you have to own it first so choose wisely (hint: try something that includes the keyword in the name).
Here’s how to set up an ACE for this kind of test:
- Duplicate the ad group you want to test a new domain name with. If you want to completely isolate the test, create a new campaign with two identical ad groups.
- In campaign settings, define a new campaign experiemnts and change the ads of one of the ad groups to display your new domain name.
- Make sure one ad group is the “control” and the other is “the experiment”.
Here’s a more detailed post about how to set it up, the only difference being that you’d use it for a new domain test not a landing page test like the author suggests. I wouldn’t test a landing page with ACE.
The above is to give you something to test right away. Once you’re done with the above, analyse your own campaigns and the levers that influence results. Then test the hell out of them.
To be completely honest with you, although landing pages and ad testing is now second nature to me, I really don’t use campaign experiments as much as I should. It’s a shame, I know. I intend to remedy that and sometime write case studies like I have done for landing pages and ad writing (no promises though).
Also, I haven’t gone into the specifics about how to build all these types of experiments, and that’s not the goal of this article (if you want specifics, leave a note in the comments) but the idea is to build a culture of experimentation within your organisation or for yourself if you manage your own own AdWords account.
Bran Eisenberg wrote a great article about culture of experimentation. Adopt it.
And if you fail, keep failing until you succeed.
Don’t listen to me, listen to what a 19th century mogul would tell you: