Often small to medium size businesses think that big advancements in marketing technology will have little to no impact on them. That they can just carry on doing business as they have always been doing it.
Last year 75,503 businesses closed in Australia according to IBISworld. If the introduction of the internet, Google and Facebook have had no impact on your business in the last decade, you are one of the lucky ones and please skip reading this article.
What is Marketing Machine learning?
Machine learning is the ability for a program to automatically learn and improve from experience without specifically being programmed, which means a marketing system could test and learn what is working without human intervention. An example is the Google Display Network. It has machine learning built into the platform that runs multiple creative ad tests to figure out what is the best website and best time of day, to put your ad in front of the target audience you have selected.
The system can have some parameters set in place that allows you the advertiser to say that you are willing to spend say, $20 to acquire your customer and the system takes that into account when it is bidding for position against other advertisers looking at placing an ad in the same spot at the same time to the same audience. The bidding is happening in microseconds and it has become very difficult to outperform the system as a human sitting and optimising ad placement.
Why do you have to be aware of this?
Machine learning operates in various forms on Google’s search engine, Facebook ads, CRM systems, and various Demand Side Platforms. If you’re advertising online, machine learning is impacting you and your business right now. The only reason you’re still getting traffic to your site is that generating a creative hasn’t been automated yet. Currently, as a business owner, you either ask your inhouse design team to create your ads, or you outsource. Once the ads are approved, they are loaded into the platform with some parameters and you sit back to see how the creatives perform.
Imagine if the creative part of the process was automated and the system could create and test its’ own designs based on the response it got. The system would be able to create millions of very individualized ads that could target an audience of one. As marketers, we know that targeted ads outperform non-targeted messages to large audiences but at the same time, we know that it isn’t feasible to target small groups of people because the creative cost is too high.
Here are some numbers to help you understand the impact. Say you spend $1,000 per month on Facebook advertising, you selected very specific demographic information to target your ad and you’re paying roughly $20 CPM (cost per thousand). For your $1000 investment, you get about 500 clicks (at a 1% Click-through rate) and you manage to convert 10 people into buying your product. That means you’re paying $100 per acquisition. With machine learning in place, the 1% click-through dramatically increases. Remember the system knows everything about you, your favourite colour, your mother’s birthday, where you went on your last holiday, the hobbies you have, where you work, your favourite movie and the kind of language you use when you talk to your friends. With all this information the system could create the perfect ad just for you at the right time because it knows what time you clicked on the banner the last time and bought something.
So, at a very conservative estimate, I believe the system could easily increase the click-through rate to 20%. This means for your $1,000 monthly spend, you could now get 200 customers assuming you could maintain the same conversion rate of converting 1 customer to every 50 clicks. This however also assumes that the cost per thousand would stay the same but because I’m just targeting 1 person there is less competition for that person and the $20 CPM could drop down to $10 CPM, which essentially means I can get twice as much traffic which means twice as many customers.
As a company, you could never compete with a competitor using machine learning, as it can enable them to drop their product price by $50 per sale and still save $45 per customer on marketing costs. A company with this power will put you out of business in a couple of months. How long would your loyal customers support you if they could get the same product for $50 cheaper?
What’s the answer? You can sit back and wait for all the magic to happen or you can start playing with it and see if you can make it work for your business. There is a case to be made either way. I’m in favour of being part of the new technology as I would rather see some opportunity in the space than be told about it later… Like the two Ph.D. students from Stanford who back in 1998 needed $100,000 to help build a better search engine and when the company named Google listed 5 years later it was valued at $23 billion!