I love data. It is strange to say that as someone who does not like complex math — I stopped acing that class once I hit pre-algebra and had to take calculus twice to get a passing grade in college, though the second time I did get an A. The longer I am in business, and the longer I work with other businesses, the more I realize that data is the fuel that powers every great business decision. Data is how you know that your marketing campaign is paying a 10 X return on investment, rather than costing you two dollars for every dollar you bring in. Data is how you optimize between different products and marketing channels, and how you make an educated decision on what to charge your clients rather than just taking a straw poll in your local bar.
More specifically though, how does data make or break your marketing – and by extension, your business? Here is this week’s example:
I wear a lot of hats: I work with lawyers, I am a lawyer, and I also work with nonlegal businesses to bring benefits to military veterans. One of my biggest projects is connecting veterans with a tool that compares VA loan mortgage rate quotes to ensure that they get the best deal possible. Our latest and most-enduring obstacle was getting our paid search ads back in the green. We get paid per lead that we bring into the banks — basically a commission on a potential mortgage. That lead value, just like a case that comes into a law firm, or a customer in a convenience store, varies based on the facts: credit score, geography, size of the planned home to be purchased, purchase or refinance, etc.
Now, you also need to understand the context of this project: I work for a large corporation. I am only in charge of the paper click advertising at this point — not the data team, not the creation of landing pages, nothing else. Basically, if I want to change anything outside of ad copy, I have to build a coalition across companies and departments. Everything is a massive undertaking.
When I arrived at this company, all of the paid ads were in disarray. Everything was losing lots of money. For products that had a predictable value, like a flat-fee legal service or a candy bar at 7-11, I was able to turn around very quickly: we rebuilt and adjusted the PPC campaigns, embraced new levels of automation when it came to bidding against a target cost per acquisition (cost per lead), and watched the money stacks accumulate. But, as I mentioned, mortgages have variable revenue per lead — we have to optimize to a moving target. In one month, the revenue per lead might be down in the $20 range, while the next month it might raise to the $40 range, with no warning in between – it actually typically varies greatly day-to-day.
Again, this is a lot like a law firm case. You don’t know if the next person to walk through the door is going to be a dead-end malingering slip-and-fall case with no real injury, or a man whose life was irreversibly ruined by negligent store owner, leaving him no longer be able to make millions of dollars playing professional football – you are constantly hitting a moving target.
What I did to solve this problem was to look for a system that could pass back these variable revenue per lead figures on a continual basis to Google’s smart bidding algorithm, which optimizes to hit that moving target. Instead of a human having to go in every single day and adjust for the previous day’s figures, hoping to catch the trend before it bankrupts you, the data will pass in real time to the algorithm which can move faster than any human. It can also pick up trends like leads from certain geographies being more monetizable than other geographies, leads from certain income or demographic levels being worth more than others, etc. Heck, something as simple as time of day that someone clicks on your ad can have a massive impact, and while that one is easy for us to spot and optimize on our own, I’d rather let the robots to the work.
For almost a year, I have been struggling to turn this around. For the last three months, I have been working to implement a cookie-less data tracking system that passes all of the data through backend servers, much like sending an instant message with revenue values from one server to another. We implemented the technology a few days ago to start testing it.
And what happened next blew my mind: it turned out that our paid search campaigns had, in all likelihood, been bringing in about 20 or 30% more revenue every single day since we started than we had previously thought.
How is that possible? We’ve always had a huge lump of revenue that we could not account for. This is the revenue that came in, but the old tracking system would get stripped off by privacy protections and browsers. The new system, because it does not rely on browser cookies, preserves the tracking and gives us much better insight on the data.
Big picture: using the old data tracking system over the last five days, our mortgage ads have been a little profitable — but barely so. Historically, they have lost us money and made me question my very existence. Under the new data tracking system, our paid ads are printing money. If this data trend holds, and if historic data (which we cannot get back, sadly) is the same as present and future data, I have actually made it rain on this company.
Now, if I were a small law firm rather than a large, faceless corporation, and I have looked at over a years worth of data under the previous system suggesting that paid ads were actually losing me money, I certainly would have made the logical decision: I would have stopped running those ads. In the process, I would have killed the goose that is now laying tens of thousands of dollars per week in golden eggs.
This is the power of data. Data has just turned an idiotic losing endeavor into a gold mine, and we haven’t even gotten to the best part: the optimization. The whole point of this entire endeavor was to get the data — and then pass it back to Google’s smart bidding algorithm so that Google can run amok and optimize at a level that no human could ever possibly do, balancing thousands of factors to optimize our bids to maximize our revenue potential. (Google smart bidding algorithm can target even a ROI figure — it is almost as simple as saying “Hey Google, make me 20% in profits.”)
Again, to repeat, we haven’t even gotten to the whole point of the data tracking system — the optimization. Even just cleaned up our initial data tracking has entirely reversed the narrative on our marketing strategy. Our campaigns have gone from idiotic and fruitless to downright brilliant. Just from fixing the data.
I’ve always known that data was important. Every career stop I have made, I have emphasized fixing the data first. From tracking page views back in my blog or days, to repairing the PPC tracking when I was directing the marketing of a multimillion dollar chain of family law firms, to doing all of the above for a dozen law firms at once at another stop, to this little mortgage project that I’m on now, I have pushed for better data and better tracking. Even with that background, I am still sitting here in awe at the difference that fixing the data tracking made. Last week, I would have considered firing myself if I didn’t know better. This week, I feel like I deserve a raise. Maybe a promotion too.
And while I have all of these happy, positive feelings, I’m going to make this promise: for any client we take on, we will have this level of data. We will require it. We will move heaven and earth to implement it. Because without the intelligence that it brings, advertising and marketing decisions are just guesswork. I’m not willing to risk your business on guesswork, just as I would not risk my own.