Making A.I. understand your business goals

Quickly, cheaply, or well done. You can have only two of these at once.

When you’re running the business, you have to face dilemmas and make decisions all the time. A natural and wise way to mitigate a dilemma is to make one’s decisions based on measurements, but sadly, to measure can sometimes be even more difficult than making decisions based on intuition.

To be precise, to measure it correctly, since, in general, the available metrics can all be tempting yet not entirely accurately represent what you truly aim for.

Wasted half

One of the most famous quotes regarding measurements belongs to John Wanamaker and directly relates to marketing:

Half the money I spend on advertising is wasted; the trouble is I don’t know which half.

The issue seemingly cleared up when we started to explore and harvest the digital realm in our marketing efforts – the BTL approach in general. On a daily basis, we use hundreds of metrics and tools that provide those metrics and reports, estimations, and other supportive materials. So this dilemma seems to have been solved? Not necessarily.

Digital Blessing

Digital tools are a blessing and definitely are helping save money and time. The catch is that measuring is something more than simply checking the temperature and reasoning if you wear the coat or not.

The rule of thumb is to clearly define your goals which we can call now a strategy, and pick the right actions – the measures to achieve them.

It becomes (sometimes very painfully) clear when we start using A.I. Smart tools aren’t genuinely smart; they just mimic intelligence. Think of them as the child who trusts you. If you say that black is white, don’t be shocked if it’s picking the white shirt if you ask it for the black one.

And as artificial intelligence-fueled tools are a must-have in modern marketing. It’s essential to make sure that everyone involved in designing and developing this strategy fully understands it, and all targets are well defined and precise upon mutual agreement.

So what?

If you are asking yourself if you should bother if A.I. understands your goals, the answer is yes, you should. And as artificial intelligence-fueled tools are a must-have in modern marketing. It’s essential to make sure that everyone involved in designing and developing this strategy fully understands it, and all targets are well defined and precise upon mutual agreement. And there is more than one reason for that.

1. You already augment your business with A.I. without noticing it.
2. You will use A.I. over time more and more, and it’s good for you to incorporate a data-driven approach deeper and deeper in your company DNA. You want to grow, don’t you? Then you want to be prepared!
3. Last but not least, if you made A.I. understand your business correctly, that means you understand your business goals as well. It’s a sort of Rubber Duck method, often used in programming. In this method, you are explaining your code, its mode of work to rubber duck (or another inanimate object) line after line. By doing so, you often find several small, detailed issues, you missed. Making A.I. understand your business goals helps to overcome the curse of knowledge.

First things first.

Let’s start with smart ad targeting, as it’s probably the easiest and first way you face A.I. applied to marketing. We will use Facebook as an example, but you could apply the way of thinking about it to every performance marketing ecosystem.

Let’s present it on the bit overly exaggerated example.

Let’s imagine a webstore. The naive analyst checks which digital campaigns sell the most, picking revenue from the campaign as the dominant KPI. There are two main – brand awareness and conversion with budgets divided 50/50. It turns out that 95% of the overall purchases are attributed to the conversion campaign. No shock here; this is its primary target.

In theory, we can find out which part of the expenses is wasted. Still, usually, when we assign the whole budget onto the “efficient half”, it turns out that overall activities are not as effective as they really should be.

Upon reading metrics analyst decides to move brand awareness funds on the sales campaign: it’s business, after all, and it’s about making money.

The new tactic seems to work for a short period. Sales jump to pleasant 180% of the former value, however not the 190% as the analyst initially assumed. Later quickly drops to 150%, and overtime falls even above referral value – 70% and lower.

In a given example, the analyst tried to optimize sales while dismounting the sales funnel.

A slightly different approach is essential at each stage, and the results need to be interpreted differently. Thus each stage needs to be tuned differently. The best explanation I’ve heard is to compare marketing efforts to a soccer team. The striker scores goals and usually becomes the heroes of the matches, but it doesn’t mean that the rest of the team is less important. The main target of the team is to score points, but not every player is striking.

Events

Ad set optimized for conversion (sales) will deliver less traffic to the page, while in terms of the goal itself, it will theoretically bring greater, and certainly more measurable, results. After all, you want to sell, and that’s correct, but you need to push an already convinced audience.

Let’s climb up the sales funnel. Add to cart event. Is such a campaign less profitable, even though it will deliver, in a direct way, weaker sales results? The answer here is no; it’s not. Once again, you aim to push the client one step closer to the sale. The actual name of your target here is to make your client hotter. Not sell, and you have to remember that, no matter how tempting it was to move some budgets, like in the example we used before.

What with acquiring new potential clients to heat them lately?

Excellent results are achieved by campaigns that work on so-called Lookalikes. These are large, diversified target groups generated by Facebook’s algorithms based on other model groups – fans of the Fanpage or, better, people who have made a purchase. They are the best choice for acquiring new, valuable traffic.

Once again, you are leveraging the A.I.. Still, this time, what you are really doing is giving your A.I. save space to training and learning how to pick your potential clients – that’s why Lookalikes are usually used in the early steps of the sales funnel.

Their strength lies in their diversity in terms of age, gender, and general interests. Diversity is the key here, as our presumptions often bias us. You are not forcing A.I. that middle-aged men are your clients, and you are just showing your clients to A.I. and allowing it to indicate the actual target by itself.

Input Quality

Every advertiser ultimately cares about sales. Some users are more likely to buy, others less so. So why do most advertisers solicit the less valuable ones?

The main steering wheel of the ads delivery is their budgets. At the same time, what is not always clear to the advertisers, it acts as a filter – critical for overall system efficiency. By shrinking to low budgets, you are giving up valuable audiences and delivering lower quality traffic at the beginning of the sales path, giving your A.I. another bad example.

In most cases, when the rate per click increases to enormous proportions, it is a signal that something wrong is happening in the campaign, and that is not debatable.

However, you also have to deal with higher rates.

So more expensive advertising is better advertising? Sometimes – yes.

For example, at the last levels – abandoned shopping carts, order finalization – you probably use retargeting, i.e. returning with different ad content, to customers who already know your brand. When we talk about Facebook Ads, in 90% of cases retargeting is people who have been to your store page or visited your Fanpage and performed any action there.

Your hot prospects are usually, at some point, handpicked out of the audience crowd, so it’s natural their price is higher.

The second thing is that even if the prospect isn’t white-hot, opening the doors wider allows you to fight for more interesting ones – their price is higher because they are more desired.

To use a comparison, imagine the campaign as the open gate. The wider you open it, the more traffic you provide, but at the same time, you allow larger objects to pass through it.

It is a precisely defined, narrow and precious group. Its devotion comes at a price – in the form of high rates for reach (CPM – cost per mile), which are high or at least higher than you would expect from reach campaigns where you cast your net a little wider. Still, it’s easy to rationalize this expense by simply converting the cost per order (CPO) and confronting it with the return on investment.

Alright, but what’s after acquiring the client?

You will learn in the second part!