How the Finance Industry Can Tackle the Incremental Lift Problem

By Jeremy Fain

Many major financial institutions, such as TD Bank, Citibank, and Chase face a fundamental problem: they already have big brand recognition. But wait, you ask: isn’t that a good thing? Not necessarily. Not if your goal is incremental lift.

These financial institutions are already engaged in copious amounts of advertising, especially on television. The question these marketers constantly face is this: if I spend my ad dollars in a particular way, am I getting access to new consumers? Or am I simply advertising to people who are already going to purchase my products anyway?

Let’s take the example of a bank, any big bank.  They are always looking to add new accounts, but it is not as easy as it sounds.  People choose banks based on their parents, or what their first job recommended, or because they are upset with their old bank.  There are many products that banks have that they can advertise to their current customers or try to find new customers. But how do they find these new customers without wasting advertising on people who are already going to be customers?  If someone is looking for an online brokerage account, and they are already ready to give that business to TD Ameritrade because they have seen so many TV ads or their friend told them it is the best, then TD Ameritrade wants to advertise to anyone BUT them. All of these businesses are interested in figuring out where they can advertise to bring new customers into the fold (incremental lift)  – without wasting valuable ad dollars on consumers who have no likelihood of converting to a sale.

This pain point – achieving this incremental lift – is what inspired me to start Cognitiv in the first place. Not just because solving for it was an exciting problem, but because after surveying the market and hearing about all the poor reviews of the current offers, I became convinced that nobody was cracking the code of consistent incremental lift. Now, by using the newest AI technology called Deep Learning, I’m happy to say that Cognitiv is finally solving this big marketing problem.

The methodology to show incremental lift  has been standardized over the years. I’ll be the first to admit that our measurement methodology is not groundbreaking. The groundbreaking part is how Cognitiv uses Deep Learning to solve for who, when, and where to target, that is delivering never-before-seen results to these types of big brands.

Without revealing too much of our secret sauce, we are able to train neural networks to recognize who has a high likelihood of already converting without more ads (non-incremental customers) and who has a high likelihood of converting if exposed to an ad (incremental customers). To go even further, Cognitiv can predict how the placement and time of the ad will improve the effectiveness of that ad for the potential new customer. This solution not only delivers higher incremental lift than more traditional solutions, but it improves the conversion rate and return-on-ad-spend of the incremental customers.

Plenty of people ask me how Deep Learning achieves this result. The answer? Deep Learning can do this since it uses large amounts of data from the brands themselves to train algorithms that are unique to their needs. Cognitiv is the only company right now able to build these automated, self-learning miracles at scale. Our neural networks are sophisticated enough to find the patterns in user data and discern the differences between incremental and non-incremental customers.

That’s why we’re so happy that we have been able to deliver dependable results in the financial services sector. That’s a major boon to any financial services firm, as prospective new customers are constantly entering the market. The financial services providers that find those new entrants and convert them are the ones that thrive.

Cognitiv’s algorithms are consistently updated, which is an important thing to note, because there will always be new consumers to advertise to. And with deep learning, the algorithm will be able to target them. Best of all? When companies stop wasting money on prospects who will not convert, they can reallocate those funds towards advertising to consumers who will.

Solving for incremental lift is extremely important. The reason why more agencies and platforms aren’t tackling it, however, is that they don’t have a good solution. Incremental lift solutions have not been shown to produce any lift in the past, because it was too difficult to see who was going to convert, and who wasn’t. Deep learning will power a future where achieving incremental lift is a distinct possibility, and Cognitiv is proud to be leading that revolution.

Data Possession – Not Just Ownership – Is Key To Improving Advertisers’ Results

This post originally appeared in AdExchanger on September 15, 2016.

In a time where big holding companies are expecting all buying to go programmatic (Dentsu is the latest), it is more important than ever that advertisers not only own their data, but actually be in possession of it.  Many advertisers may not realize that ownership is not enough (or what the difference is).  The benefits of data possession in today’s world of Big Data cannot be discounted. Without data possession, the advertiser cannot take full advantage of its most valuable asset and, therefore, cannot compete against those that have already learned this lesson.

Owning and possessing data are two very different things. Owning data means you have control over how the data is used. Possessing data means you can use all of the data to get a full view of everything you own across disparate partners.  This holistic data set, including advertising campaign results, site visits, and customer data, is already helping the most advanced advertisers drive big improvements in ROI.  For those not yet centralizing this detailed data, it is lying around in pieces and produces limited understanding and suboptimal ROI.

For years, advertisers have given permission to ad tech partners, such as DSPs and DMPs, to safely collect and use data on their behalf.  Programmatic advertising’s healthy ad tech ecosystem has created incredibly detailed and diverse data on every ad impression and user. This data, however, tends to be siloed.  Where the ad ran, how much it cost, information about the user it was shown to, viewability scores – all these pieces of data are generally possessed by different ad tech partners.  New best-of-breed solutions are popping up every day, so this situation is likely to continue for a while.  Since all of this data is contractually owned by the advertiser, the advertiser is the only one that has the right to take possession of it all and join it all together into an incredibly rich, deep data asset.

But what is an advertiser supposed to do with all this data?  Today, the beauty is that an advertiser does not have to do anything with it.  The marketing and ad tech ecosystem has many uses for this data already and teams of data scientists champing at the bit to get access to it.  New companies are sprouting up every day with great, new Big Data ideas.  An advertiser’s central warehouse can cost minimal amounts of money relative to overall marketing budgets and the advertiser’s partners can be given access to it so they can add their unique value to the core data set.  Even if the “warehouse” is only a collection of raw log files and downloads from vendors, it can serve as the source for improved algorithms, insights, and strategies.  It can also serve as a sandbox for those at the agency or advertiser as well as valuable third-party partners to experiment with new analysis, research, and technology.

It seems like digital advertising evolves every six months.  Bring Your Own Data (BYOD) has evolved into Bring Your Own Algorithm (BYOA).  Cross device targeting and attribution is no longer science fiction. Native advertising and dynamic creative are delivering improved, personalized calls to action.  All of these new pieces of the advertising and marketing ecosystem are being driven by data – and they will drive even better results for the advertiser if they have a more comprehensive data set.  As a first step, advertisers should begin saving the last six months of historical data from their partners at the most granular level possible – hopefully at the impression or user level. Then they should let their partners loose on it all.  Advertisers will be richly rewarded.