Yes, programmatic ads work - real world example

Hope y’all are doing well. Thanks to all who provide their expertise and opinions in this community.

Earlier this week, I noticed a comment from Broadway Dan that I’ve wanted to respond to. The post was here: https://discussion.fool.com/hey-all-sorry-i-missed-that-others-p…, and the comment was:

“…the fact is programmatic ads are a theory that’s yet to be proven. We don’t know that some ads will be targeted with great precision and deliver incredible results.”

I’ve been in digital marketing for over a decade, including my current role as the Director of Digital Marketing for a financial institution. We have a seven-figure paid media budget that is tied directly to expectations for growth in loans and deposits. Most of those dollars go into programmatic advertising, and I can definitively say that it’s effectiveness is a reality. A little bit more detail…

When I first stepped into this role a few years ago, I learned that the paid media budget was driving hardly any loan or deposit applications through our website. We were spending a lot of moeny with little to show for it. For example, display ads were only tied to geographic targeting and very little other criteria. Our vendor and I quickly switched strategies.

We started an account on a programmatic ad-buying platform (a competitor to The Trade Desk) and added their analytics tracking to the end of the online application for each loan and deposit product. Let’s take mortgages as an example. Every time someone completed the online mortgage application, the platform added that user to an audience modeling pool. So, after five hundred people completed the online mortgage application, we had an “ideal prospect profile” based on those five hundred people. The profile could include age, sex, interests, propensity to be in the market for a house, and much more. And all of those factors of the profile were influenced by the actual people who were submitting a mortgage application.

How does the platform know those things about users? You know those websites you browse that serve up ads? There’s a high likelihood that this platform (or The Trade Desk or another platform) is paying those websites to serve ads to their users (thousands of sites), which means their code is on those sites. So, if you visit a dozen of those sites, the platform has placed a cookie in your web browser and knows your activity on those sites, thus creating a profile for you.

While we may have had an ideal prospect profile based on 500 applicants a while ago, that profile is dynamically updated every time someone else submits an online mortgage application. Good thing is that it’s not only applicants from paid media campaigns that influence the prospect profile.

Let’s say we run an email marketing campaign, targeting a subset of our existing customers that have been filtered for numerous criteria based on our internal data. Every one of those customers who fills out that mortgage application adds to the audience modeling (even though they didn’t come from a paid campaign) because the analytics tracking is at the end of the application for all users who get that far. In that scenario, we presumably made our audience modeling even stronger.

Here’s what a campaign looks like: We set a budget and a geographic boundary (a handful of counties, an entire state, etc.), and assign the audience model (mortgage applicants) to that geography. Every time someone in that geographic area hits any of those thousands of websites, the platform determines if the user fits the mortgage applicant profile. This determination happens in real time on every page load of every one of those websites. And all these platforms are competing for the same ad space in a matter of milliseconds. You and I may live across the street from each other and both browse ESPN.com on the same day, but we have unique profiles. So, you may see our mortgage ad, and I may see an ad for a sofa from Wayfair.

How do we know it works? We track cost-per-application (ad spend divided by total submitted applications), loan approval rates from underwriting, and loan booking ratios from credit administration. Over time, we have lowered cost-per-application, set records for application volume, and set records for digital loan & deposit growth. We’ve increased our paid media budget each year because of this success, winning over the support of our executives.

Last month, our vendor notified us that The Trade Desk is being added to their series of networks. They were really excited about this because, as they explained, it opens the door to more targeting techniques that our current network does not provide.

Hope this helps.

-ElonFeeNix
9.2% position in TTD

161 Likes

Hi ElonFeeNix, I was curious about some of your results with programmatic. When you say you guys have lowered cost per application, are you only looking at total paid submitted applications or total applications? Also, was curious about your general marketing mix, are you also doing SEO/SEM, paid social, etc. Generally with programmatic, from my understanding, most of the conversions are going to be view through conversions vs. click conversions, so what are you guys using to determine attribution for programmatic? Finally, interest rates are rock bottom right now, could that be a contributing factor to the volume you’re seeing now? Sorry for all the questions, just curious about how you guys measure programmatic performance.

3 Likes

RunnerGuy,

When you say you guys have lowered cost per application, are you only looking at total paid submitted applications or total applications?

We’re only looking at total paid submitted applications when determining cost-per-application. I realize my original phrasing may have confused this point.

Are you also doing SEO/SEM, paid social, etc.

Yes. Our paid media is comprehensive (including display, search, and remarketing). We do SEO. We do social media and mix in paid social campaigns from time to time. There are also various customer-specific campaigns via email marketing and on our digital banking platforms.

Generally with programmatic, from my understanding, most of the conversions are going to be view through conversions vs. click conversions, so what are you guys using to determine attribution for programmatic?

We do see more view-thru than click-thru conversions with programmatic ads. Of course, this indicates that those ads can often serve as early touchpoints in a buying journey and be considered an assisting source when we analyze a last-click attribution model. For us, programmatic ads play varying roles as part of comprehensive marketing, whether it’s an assist for some conversions or direct attribution for others.

Finally, interest rates are rock bottom right now, could that be a contributing factor to the volume you’re seeing now?

Certainly. However, we had notable growth (relative to our past performance) long before the pandemic impacted the market. This has been a multi-year climb for us. These past few months have taken it to another level because of increased digital usage by the market and the low interest rate environment that you mentioned.

I hope these responses are of value. Thanks for the discussion.

46 Likes

Don’t want to pull us off topic, into psychology of investing, but I had just sold a lot of $TTD (want to buy house) and suspect I added that almost off the cuff to comfort myself about the sale. I bring this up solely to remind all that every poster has biases and personal factors that don’t always appear in the text. And that’s why Saul’s frequent admonition to make your own decisions is crucial. Thank you ElonFeeNix. Important post. Appreciate your insight.

13 Likes

Good thread on $TTD. TTD is my largest position and grew to the #1 position which is above 20% of portfolio recently. I am also a newer investor learning from this board.

To contribute on the discussion around TTD, I think I also remember Saul had provided context around exiting TTD with mentioning his potential bias of not trusting the advertising space - i think it was from having been burned before.

Also, if i recall correctly. There was also a large thread and discussion around why TTD did not reveal a certain breakout on their ER about what channel the revenue% and growth was coming from. There was a lot of bandwagon-like discussion around this.

I made my own opinion after thinking about it for a few weeks.

My thought was that not all companies are going to give us all the information we would want. I felt like it was similar to how Amazon does not reveal prime subscriber numbers and related KPIs in great detail when many investors would want that. I think this has more to do with the individual’s comfort level with the amount of information they can have with any given stock and this isn’t a reason for me to sell as long as the thesis and company is executing.

One method I use to try and make my own opinion is to apply the similar context for buying/selling to a number of companies and see if I feel like i would make the same decision with those companies.

5 Likes

“So, after five hundred people completed the online mortgage application, we had an “ideal prospect profile” based on those five hundred people. The profile could include age, sex, interests, propensity to be in the market for a house, and much more. And all of those factors of the profile were influenced by the actual people who were submitting a mortgage application.”

Now hook up with a company expert in datamining for further analysis to develop clusters of various consumer types and determine which specific factors - and to what degree - are influencers for each cluster type. As a side note, I’d go with an Oracle product rather than one by IBM (like it’s IntelligentMiner).

Hello Elon. Recalling your great post on 18 Jul. Have a question for you.

Just finished reading the earnings transcript. I can’t reconcile the CTV opportunity that Jeff keeps speaking to. Seems like every CTV opportunity he mentions is one that ROKU has ---- and only is a bigger opportunity for ROKU. The rise of CTV streaming ---- AVOD, SVOD, TVOD. All Roku opps ---- given that is their single focus. Jeff spoke to Peacock, Pluto, Vudu, etc., Again, all ROKU opps. Eventually, HBOMax, etc., will all be on ROKU too ---- once the negotiations are complete. 1/3 TVs in the US Roku OS + the millions of Roku sticks, Roku devices connected to TVs.

Note, their their revenue actually picked up significantly starting in June and into July ---- above last year’s pace. That + “the opportunity” is what the Street is hung on ---- and why it didn’t fall off a cliff on Friday. Oh, and Jeff’s salesmanship.

I’m a 14.5% TTD holder and 5.5% ROKU holder. But after reading the report, I think I’ve got that wrong ---- clearly not wrong to this point but I believe wrong on a go forward basis.

Your thoughts?