What if you ignored an opportunity that was already making you money, just because you didn't have a way to attribute it?
There's a good chance that most businesses are doing exactly that when it comes to local search.
According to a recent study conducted by Google, 50 percent of smartphone users who performed a local search visited a business within a day. Computer and smartphone users did that 34 percent of the time.
On top of that, 4 out of 5 people use search engines to find local information, and 51 percent of them use their smartphone to search for that information when they're already on the go.
Assuming people can find you in the search engines, that means a good portion of the people who walk through your front door are doing it because they found you in a Google search.
And if you don't know about it, you're probably not investing as much in local search as you should be.
So how can you attribute walk-ins to local search? Does it take a crystal ball?
The Measurement Problem
Online attribution is easy. You can see how many people clicked, and how many of them bought something. It's straightforward, it's concrete, and there's nothing to second guess.
When you use SEO or search ads to spread awareness about your business, on the other hand, you can second guess almost everything.
That's why you need a way to attribute those walk-ins to the search engines. Fortunately, there are a few ways to do this.
- Coupons. While it's obvious that not everybody who finds you online is going to use a coupon, you can at least track the ones who do by using a coupon code that can only be found on your website.
- Phone numbers. In some cases, it may make sense to use a specific phone number or extension for online correspondence. If you list one extension online and another one in the yellow pages, you can estimate the proportion of your callers that are coming from search.
- Surveys. One of the most trusted ways to find information like this is with a survey.
While all of these are useful, each of them have their own issues:
- Coupons won't catch everybody. No matter what, they will underestimate the number of visitors that are coming from search. And, of course, there's also the fact that those coupons make the visits less valuable in the first place.
- You can't expect everybody to type in an extension, so phone extensions suffer from the same problem as coupons. Using a completely separate phone number, on the other hand, can actually confuse the search engines, and end up harming the exact thing you're trying to measure.
- There's no such thing as a mandatory survey, so you're always going to end up surveying the portion of your audience that's willing to take a survey in exchange for the chance to win some kind of prize. That's not necessarily a representative sample of your customers. People can also lie in order to get the survey over with, or their memory of exactly how they found you might not be too great.
Thankfully, there's one more way to attribute your visitors to local search, although it's a bit techy.
I'm sure you're just aching to learn more after hearing a phrase like that, but hear me out.
Regression analysis is a way to model how things work based on the data you have to work with. It's in the same boat as correlation, and yes, it does suffer from the whole "correlation isn't causation" issue.
The good news is that regression analysis allows you to rule out certain things.
For example, I can use regression analysis to model how many visitors are coming to your store based on how many visitors come to your website that day.
You might argue that you get a lot of business on Saturdays, so of course more people are going to visit your site on Saturdays as well. That doesn't mean that the extra people who visited your site on Saturday actually caused your number of walk-ins to go up, right?
But here's the thing. With regression analysis, I can actually compensate for that. I can set up some dummy variables to account for the day of the week, or the month of the year, and I can model those too. I could even throw in some variables to account for whether or not a TV ad was running that day.
While it's true that there could always be some hidden variable we don't know about, we can at least rule out the influences we know about.
Here's why I like regression analysis better than the other options:
- Unlike a survey, it deals with people's actual behavior. I'm not saying that people lie on surveys, but I am saying that what people say and what they actually do are often very different.
- Unlike a phone extension or a coupon, it's not going to leave anybody behind. It measures the influence of the total number of site visits on the total number of walk-ins.
- It's predictive. While there are no crystal balls, regression analysis can actually help you make estimates about future behavior. The other approaches can't do that.
I believe that all four approaches have their uses, benefits, and weaknesses, but regression analysis is my personal favorite, even if it's a bit nerdy.
Image credit: aussiegall