Why Digital Advertising Agencies Suck at Acquisition and are in Dire Need of an AI Assisted Upgrade


Are you happy with digital marketing agency?

I mean sure, they are great for creating “awareness” but how often do they result in conversions or even a solid lead? Unless you are marketing FMCG goods like cherry flavoured cola, traditional digital marketing agencies can be sinkhole for cash without any significant ROI. This is especially true if you are selling high-involvement product like loans and insurance.

I work in the Malaysian fintech industry, for an enterprise solutions provider that builds loan origination and debt collection software for most major banks in Malaysia. We build really great products, and our customers can vouch for us. However, the problem our clients face is not getting enough leads or conversions from their digital marketing activities.

A simple example, we recently created a few marketplaces for various types of loans for some of the largest financial institutions in Malaysia. In one of these marketplaces, a customer can play with a loan calculator where they key in how much they can afford to pay per month, the preferred loan amount, duration, and voila; they get matched with banks offering loans on their terms. On top of that, they can apply for those loans by uploading relevant documents on the same web-based platform. A lot of fancy technology went into making that happen, but sadly our client reported they are not getting enough web conversions and more importantly they are not getting traffic from the RIGHT people.

Here’s the problem. You built a great store to sell your shit. Many people who have thought about buying your shit end up visiting your shit-store. But the problem is, just because someone has the intention of purchasing your shit, doesn’t mean they have the means to buy it anytime soon. Therefore, you end up with a lot of irrelevant people muddying up your store without actually buying some shit.

That’s how the digital marketing agency scene is right now. Their targeting is simply too broad. I have banks complaining to us all the time about how they are spending hundreds of thousands of dollars in search engine and social media marketing with pitiful results. And this is from one of the biggest banks in Malaysia that employs one of those top tier award winning digital advertising agencies in the region.

As someone who has spent half a decade in advertising, I am well aware of the sloppy targeting mechanisms used by these fancy pants agencies who claim to be “famously effective”. A typical targeting procedure orchestrated by a typical digital agency (although they all uniformly claim to be “unique”, “innovative” and “data-driven”) looks like this:

Targeting people in Kuala Lumpur who “might” be looking for home loans:

Age: 25-55 (Cause they found out after conducting a focus group that people in this ridiculously broad age range are most likely to apply for loan)

Interests: Home loans, housing loans, mortgage, and a whole bunch of relevant keywords based on search volume (higher the search volume, the better)

Geo-targeting: People residing in Kuala Lumpur

And tweak other settings like campaign objectives (brand awareness, lead generation, conversion) and target people who are looking into property purchase sites. Some wily agencies will even run two concurrent campaigns simultaneously: an edgy campaign that appeals to 25-35 year olds, and a conservative campaign aimed at 36-55 year olds.

Sounds like a good strategy on paper. Except it fails miserably in real life. Because these targeting criteria fail to separate people who are casually browsing houses or window shopping for their dream house — that they may or may not be able to afford ten years from now — from bona fide high potential shoppers who already have their hearts set on a specific piece of property and want their home loans yesterday.

What most self-proclaimed “cutting edge” and “data driven” digital marketing agencies fail to identify is where the customer actually is in their purchasing cycle. Not to mention they have no quantifiable measurement for an individual customer’s potential for conversion. This is where data-mining and big data comes in handy. This is where digital advertising agencies fail.

A study done in 2014 by Massachusetts Institute of Technology shows that ads using machine learning and big-data driven marketing result in 13 times more conversions than traditional marketing best practices in multi-national organizations.

Imagine if you are the world’s largest designer luxury lingerie store and each of your items cost between $2500 – $25,000. Let’s say you have about 60,000 people coming into your site every day. Out of those 60,000, less than 1% have actual purchase intentions and the rest 99% are just casually browsing or window shopping and maybe adding items to their wish list for the distant future. A typical digital marketing agency ends up targeting all 60,000 people who visit your website, and more if they are trying to acquire new potential customers who have never visited your online lingerie store. Because with the little data they have from their “focus groups” and from their “test campaigns”, the best they can hope to do is create a campaign which gets people to click on the “learn more” or “shop now” button. However, the problem is not getting people to click on the “shop now” button. The problem is getting people who are MOST LIKELY TO BUY to click on the “shop now” button.

Now in countries like US you can target people by their income groups and make sure that people who cannot afford your luxury lingerie never get to see your advert. But again, we fall back into the same trap of overtly broad targeting. Just because they are rich, doesn’t mean they are looking to buy pricey lingerie. In fact, one of your biggest segments could be sugar babies and mistresses who’ll nag their men into purchasing your over-priced gem-studded lingerie. This is precisely why you need data-mining and artificial intelligence. An AI can mine data about every single customer who has ever bought a piece of lingerie from your store and look for patterns that identify consumers who are most likely to make an actual purchase as opposed to consumers who are just window shopping. The AI can generate a very lucid “potential for conversion” score, like letting you know whether a customer has a 9% or 90% chance of buying.

Once you have this score you can do all sorts of things with it. Once you know (with a 90% certainty) that someone is going to spend at least $2500 on a piece of lingerie, it makes total sense to send them a highly personalized $80 goodie bag with a catalogue of your new summer collection.

I’m not saying that advertising agencies are worthless. They are still good at coming up with compelling visuals and catchy payoff lines. But when it comes to targeting, the paradigm is shifting from old-school advertising companies to new age tech like deep learning, AI, large scale data mining and prescriptive analytics.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s