In the spring of 2017 I was asked to help experiment with what paid traffic could bring to a cosmetic treatment review and connection website. There was an initial budget that wasn’t expected to make returns, so the stakes weren’t very high—but it was uncharted territory for the company and we knew that it was something we should get good at. Google had been steadily reducing our organic traffic and showing no signs of slowing down.
My role was initially just to define the experience of potential landing pages but it grew as strategy became increasingly important. At key points I was responsible for enabling this experiment to become a new revenue stream: delivering an average $180k - $250k per month depending on the topic.
Finding the right experience
The project started out very open-ended and no one on the team had direct experience advertising with Google, Facebook, and others for the express purpose of driving general traffic to our web pages. As such, my initial research and designs were varied and elementary.
I started with gathering all the search data I could find in Google Analytics, Webmaster Tools, and my company’s big data storage solution. I found that search patterns to arrive at the site and performed on the site were far simpler and more ambiguous than what most employees had been thinking. This would effect the keywords we’d be buying and how specialized we could make the landing pages. In the fewer instances where searches were more complex, like “[treatment name] near me” or “[treatment name] risks” I was able to group terms into categories for Synonyms, Researching, Shopping, and Discovery.
Next I did competitive analysis of high conversion pages and paid traffic funnels in the same and different industries.
I found a few patterns that I favored (optimized info and lists) while dismissing others (gating, forms, and wizards/quizes) as being too untrustworthy.
Later on I’d find out that throwing out those other patterns was a mistake…
Armed with this information, I suggested which keywords we should buy and designed a few pages to try out for different purposes: treatment info, doctors/providers for a treatment, special offers, cost info, reviews, etc.
The results were compelling and converted higher than existing pages but actions we were hoping visitors to take weren’t high enough. These pages were not enough of an improvement to be a financially sustainable alternative to organic traffic. Hmm.
With these initial test results, company executives asked us to focus exclusively on “shopping”: contacting a provider. Given how much ad traffic costs, this seemed like a good choice to me and I focused my designs on that.
Listening to advice
Someone also had the bright idea of asking for help, in the form of hiring a consultant who’d spent the last ten years using paid traffic to sell sales leads to all kinds of companies.
The consultant looked at what we were doing and offered suggestions pretty far from what we’d been trying. He favored patterns that I’d dismissed and everyone on my teams (sprint, design, product) wanted to take all this with heavy grains of salt. Though everyone was skeptical, I could tell I was getting good advice for our new focused goals.
As the consultant explained things that had worked for him, I drew connections to cognative biases that I’d been overlooking. His findings were sound and if we followed his advice we should expect around 3% conversion, which is a good standard for most industries.
A new test
I took our goal, his advice, and my own secret sauce into a new design: a guided quiz experience that ended with contacting a treatment provider. Internally we called them wizards.
On one hand I threw in the kitchen sink for conversion tactics and on the other hand, I added steps in a way that would record data we could act on and make smarter in the future. We got the first version built and bought targeted traffic for it.
To my surprise we didn’t see a 3% rate—we saw a 27% conversion rate!
The consultant was flabbergasted and I was pleased because this could give me a lot of room to add friction and features—things to make the experience more helpful for both the paid traffic audience and the providers we were sending contact messages to. This would be critical if we hoped to make it a sustainable tool.
Making it a product
With this record setting conversion design, we had a good tool to add to our website but I was still troubled by the uncertainty of paid traffic. How would we know how much to spend? How could we calculate ROI? I wasn’t getting any answers from the finance and marketing folks, who at that time were trying to scale our test up to become part of what we offer to our provider customers.
A bit of good luck
It was around this time that we were visited by another consultant. This one was all about testing and being scrappy about it. For an engineering focused person he was also surprisingly all about making money with it. In a small side meeting he was introduced to our paid traffic quiz. He was impressed with what I’d achieved but was more troubled than I was about the business model. I would say from his expression that he looked horrified.
He said, “I would never start throwing money at ads without knowing how much I’m going to make in return. Everything we tested was to determine how it would scale, because that is the model. We’d have meetings with Google to ask them how we could give them more money. The more we spend is the more we earn.”
Paid traffic needed to have this business model. Doing otherwise would be a wreckless gamble. I took this as the most important feedback of his visit.
At the final recap meeting I attended, I was surprised to hear that this wasn’t being brought up again—so I facilitated that subject change. I re-introduced the idea, made a short case for it, and asked the consultant to go over what he’d discussed previously.
By the end of the meeting, everyone was on board.
A reliable product
Soon after, we introduced it as a separate offering to the providers and only then did we scale up the spend on search and social ads. As a product, it was working and doing so dependably. As mentioned at the beginning:
It started generating an average of $180,000 to $250,000 per month, depending on the treatment type.
Like most projects, there were still plenty of things to be improved for the guided quiz experience. I outlined a large number of these and hosted workshops to gather and maintain strategies for yet more possible improvements and adaptations.
Having gotten this off the ground, I was moved to other efforts and teams. I always kept tabs on it though and stayed on hand for questions about early decisions and keeping track of roadmap plans.