Ask your customers where they came from
For years we've asked one question right after checkout: how did you first hear about us? Roughly one in four buyers answers it, which at our volume is tens of thousands of answers a month, and it costs almost nothing to run.
What we use it for
The main thing we use it for is the relative effectiveness of each channel at the top of the funnel. Every channel takes some share of the responses, and putting that share next to what you spend on the channel tells you a lot. When a channel comes up more than its share of the budget, you're getting more attention per dollar. That's the arbitrage.
That picture has changed a lot over the years. Early on, Facebook and Instagram brought in most of our new customers. Over time, the video channels at the top of the funnel, YouTube and TV, started taking more of the share, and word of mouth kept growing alongside them.
TV was the clearest example. Within a few months of the first spots, it was the most common answer, and it stayed there long after they aired. That kind of lag never shows up in click reporting, but the customers told us. The part I keep coming back to is word of mouth holding near the top through all of it, because word of mouth is what carries long-term growth.
The usual options either cost a lot, the attribution platforms, or take a lot of time, the brand lift studies and longer-form measurement. The survey is close to real time and cheap. When you're spending millions of dollars a year on ads, a directional answer you can check this month counts for a lot. It gave us the confidence to keep investing before the results showed up in the P&L.
Not every channel sends the same customer
Every response comes attached to an order, so you also see what the people naming each channel spent. In every version of the survey we've run, the buyers who said TV or YouTube had some of the highest order values of any channel, usually 20 to 30% above our average.
Some customers are worth more than others, and they tend to come from different sources. That's quality of revenue at the customer level, and it's a big part of why we got comfortable pushing more into TV and YouTube.
Rotate the question
How did you first hear about us is the question we always come back to, but the survey can ask anything. The completion rate stays high no matter what you put in it, so you can keep asking people things.
One we've rotated in is gifting, asking who the purchase is for. Knowing how much of your holiday traffic is shopping for someone else changes how you message the whole site in Q4.
Keep the question wording and the option names consistent, though. Change them too often and this year's answers stop comparing to last year's.
We also always leave Other as an option. When people don't have a direct answer, they'll type one, and what they type is often the most useful part.
How we read it
We look at it about once a month. I don't think you get anything from checking more often, because it's a directional read, and it doesn't replace anything else we use.
I've written before that people lie in surveys, and that's still true here. There's a bias toward certain channels because of what people think they remember. Someone who saw you in five places will name the one that stuck, not necessarily the one that came first. So we don't treat any single number as the truth. We just watch the trend.
Even the bias is information. If people over-index on one channel, it means that channel landed, and that's worth knowing.
What I look at
You don't need expensive measurement tools to know if your marketing is working. Google Trends, branded search, and a post-purchase survey are all cheap, and each one is a decent proxy. Together they get you most of the way there.
The other thing I like about the survey is that it sits outside platform reporting, so there's no pixel loss and no modeling in the middle.
If you're not running one, start this week. People who just bought from you want to talk to you, and a month of answers will give you something to put next to what your dashboards say.
If you already ask something different at checkout, reply and tell me what it's taught you.