Driving out the uncertainty

Editor’s note: Joe Richards is senior manager of research at Atlanta-based AutoTrader.com.

With 15 million+ unique monthly visitors and an aggregate of millions of new, used and certified pre-owned cars from thousands of auto dealers and private sellers, AutoTrader.com is in the business of connecting car buyers and car sellers. Along with private sellers buying ads to list their car(s), automotive dealers and associations, ad agencies and auto manufacturers also advertise on AutoTrader.com. Just like any other media buyer, these advertisers often raise the question of measurement. We often hear, “How can I measure and truly understand the impact of my advertising dollars?” or “If I only knew which parts of my advertising were working, then I could save a lot of money and increase my business.”

Unfortunately, one of the most common misperceptions today among the media-buying community is that online media is automatically more measurable than traditional media. This belief is, in part, our own fault. When most dot-coms started out in the ’90s, a primary message communicated across the board was that online advertising was more measurable than traditional advertising. And, in theory, that is true. Online media can actually offer advertisers more trackable activity via click-through rates and other analytics than traditional media does, but that’s only if online consumer behavior aligns with these trackable metrics. However, time and experience has painted a much different reality about how consumers actually engage and respond to online advertising.

Used primary research

As a result, AutoTrader.com has used primary research to dispel many advertisers’ limited focus on specific metrics and to examine the halo effect that online advertising has on traditional brand metrics. For example, when our customers advertise on AutoTrader.com or with any other media outlet (traditional or non-traditional), their ultimate goal is to drive people into their dealership. However, when it comes to their online advertising, dealers expect to be able to measure the advertising effectiveness by tracking phone calls and e-mail leads. The reality, though, is that among automotive shoppers, very few actually call or e-mail the dealer while shopping for a car. Given this information, how can you measure the immeasurable?

The answer, of course, is via survey research. The first question that our research team wanted to answer was: How do dealers believe car shoppers initially contact them? After conducting primary research, we learned that dealers think that 42 percent initially make contact via phone, 17 percent send an e-mail and 41 percent walk into the dealership. Our first impression - based on our site data and other research statistics - was that the phone contact rate seemed high.

We also wanted to understand the dollar value dealers place on phone calls, e-mail leads and walk-in traffic. Our hypothesis was that walk-in shoppers are more valuable because it is easier to sell to a car shopper in-person versus over the phone or via e-mail. The findings were, for the most part, what we expected. Franchise dealers would pay on average $65 for a walk-in shopper, $37 for a phone call and $30 for an e-mail.

Understand shoppers

Now that we had established the dealers’ perceptions, we needed to understand how shoppers truly made initial contact with the dealer. After many internal discussions, we defined the main objective for our consumer research: determine the impact of advertising on the volume of walk-in traffic at dealerships nationwide among AutoTrader.com customers and non-customers.

We concluded that the best way to do this was to intercept car shoppers as they exited dealerships and administer a short survey, asking them, “How did you initially contact this dealership?” and “What was the primary source that led you to visit this dealership?” To ensure we were interviewing car shoppers and car buyers only, we also would distinguish between actual car shoppers and other customers visiting the dealerships for parts, service, etc.

To guarantee the most objective and accurate representation of the current national market, car shoppers were interviewed at a variety of geographically-diverse dealerships, which included a mix of manufacturers, independent and franchise dealership types, luxury and non-luxury brands and new and used. We spent two days at each dealership and varied the time throughout the week to encompass both low- and high-traffic days. At the end of the research, we had interviewed car shoppers at 59 dealerships in 17 markets (stratified by population size).

Do not align

So what did we learn from the consumer intercept study? Dealers’ perceptions do not align with consumers’ actual behavior (Figure 1). Dealers perceive that 41 percent of shoppers who walked through their doors did not establish contact with them prior to their initial visit. However, the research revealed that the number was much higher. In fact, 80 percent of shoppers surveyed did not make contact with the dealer prior to walking into the dealership.

For the first time ever, we were also able to quantify the impact that advertising had on dealer walk-in traffic (Figure 2). According to the research, advertising accounted for roughly one-fourth of all dealer walk-in traffic - the No. 1 source for this type of traffic. This finding was closer to dealers’ perception than their view of how shoppers made initial contact with them; however, there was still significant disparity in these numbers as well. Dealers previously thought that 32 percent of their walk-in traffic could be attributed to the Internet when, according to the research, 54 percent of shoppers were primarily being led to the dealership by the Internet (Figure 3). Dealers also believed that newspaper advertising was driving more traffic than it actually was (27 percent versus 13 percent).

Another goal of the research was to quantify which specific Internet sites were responsible for driving dealership walk-in traffic. The study demonstrated that independent sites like AutoTrader.com were the primary source for walk-in traffic followed by dealer sites (Figure 4). Independent sites drove three times more walk-in traffic than search engines and seven times more than manufacturer sites. Dealership sites drove twice that of search engines and five times more than manufacturer sites.

The results of our study effectively “measured the immeasurable” and finally demonstrated to dealers the value of their online advertising. Or so we thought.

When we shared our findings, some dealers were still skeptical. They believed the research was not representative of what was actually happening in their dealerships because the study was controlled and the data was weighted (type of dealer, store traffic, geography, etc.). We had to take the research to the next level.

Had to be modified

Following the national dealer walk-in study, we rolled out a pilot research program called the AutoTrader.com Sourcing Program and began conducting research at individual stores among those dealer customers who doubted our findings. However, this type of individualized research was expensive, so the methodology of the original study had to be modified. With the AutoTrader.com Sourcing Program, we focused only on car buyers at the individual store instead of all car shoppers visiting their dealership. We collaborated with AutoTrader.com salespeople to obtain buyer lists from their dealer clients and we contacted their customers over the phone post-purchase and administered a short, three-to-five-minute survey, which consisted of the following key questions:

  1. Which of the following source(s), if any, did you use or see while vehicle shopping? 
  2. Please tell me the names of the specific Web site(s) that you used?
  3. Which of the following source(s), if any, primarily led you to purchase from [DEALER NAME]? 
  4. While shopping for your [MAKE AND MODEL BOUGHT], how did you initially establish contact with [DEALER NAME]?

To coordinate and manage the logistics and reporting of a program of this magnitude, which would consist of hundreds of “mini” research studies, we chose to collaborate with an independent third-party research partner to assist us with administering the program. Choosing an independent administrator also contributed to the objectivity of the program.

To date, we have completed 565 individual sourcing studies from all types of dealers throughout the country and have interviewed over 75,000 car buyers. We have created a turnkey program that allows AutoTrader.com salespeople and their dealer customers to source car buyers. Our senior vice president has said the program is “one of the most effective consultative tools we have in our toolkit to assist our dealers with their success.”

Here are some specifics of the program:

  • With the permission of the dealership, the dealership phone number is displayed on their customers’ Caller ID. 
  • We interview roughly 40-50 percent of all buyers at each store.
  • It takes approximately two to three nights of calling depending upon the number of buyers at the store.
  • To date in 2011, we’ve sourced approximately 15 dealerships per week. 
  • The only days we do not call dealership customers are Fridays and Sundays. 
  • We record all calls and post them on a portal for the dealership to listen to. This creates transparency.
  • We ask a question at the end of the survey for feedback on the dealer. We send the verbatims to our salespeople to share with their dealer. The dealers love it because it provides them an opportunity to turn unsatisfied customers into satisfied ones.
  • Reports are issued to our sales consultants seven days after the last interview. 
  • Dealers can insert a custom question if they so desire.
  • We can conduct interviews in Spanish for dealers whose client base consists predominantly of Hispanic buyers.
  • We database the survey results and site metrics for each dealership.
  • We conduct a follow-up survey to understand the impact of sourcing as well as to perform advanced analytics, which allows us to better consult future clients and offer case studies.

Easily understand them

The reports we provide to our dealer customers are short and concise so that our salespeople and the dealership can easily understand them. They consist of three main metrics: usage multiple; one source that led to the dealer; and how they initially made contact with the store.

Each report contains around nine data slides, with one slide for each metric that breaks out the total number of buyers, used car buyers (Figure 5) and new buyers. This information helps to affirm the credibility of the national study since the results of the Sourcing Studies align with the results of the dealer walk-in study. We discovered that there were geographic differences in the results, so we release normative data in collateral by state.

We also collect the reasons why we are conducting sourcing studies and we also review the outcome of a sourcing study by surveying our salespeople three months after the study is complete. As you can see from these numbers (Figure 6), the program has been very successful and the ROI has been three to four times the investment.

Understand the impact

In addition to the sourcing studies, we are using primary research in other ways to help our salespeople consult our customers. For example, for our national customers who buy display advertising on our site, we conduct exposed/unexposed ad effectiveness research for them to help understand the impact of their campaigns. We have normative data for our site to help put the results in perspective. We also are able to link it with site behavior.

Further, we just finished conducting a max-diff messaging test to understand what messages appeal to car shoppers when selecting the dealer. We tested 39 different messages and are creating a simulator for our salespeople to use when discussing marketing messages with customers. For example, if I am a Ford dealer targeting new-car shoppers and my differentiators are “ease of financing,” “free vehicle history reports,” “family-owned business” and “free maintenance service,” then the salesperson would select “New Car” along with these differentiators to filter the data. The simulator, lovingly called the AutoTrader.com Messaging Wizard, ranks from highest to lowest which of these messages is the most motivating to the shoppers they are targeting. It automatically removes the utilities (zeroing them out) for those features not selected and takes the average of the remaining rescaled messages. It is a great tool to use while consulting clients about their marketing plans and messaging on AutoTrader.com.

Of course, none of the endeavors explained in this article would be possible without our research vendor partners so I would like to thank KS&R, Morpace Inc. and Harris Interactive for all of the “stuff” they do for us.

We can’t measure everything

While I have to keep reminding myself that we can’t measure everything - there will always be shoppers who show up at the dealership and don’t buy for various reasons (didn’t have the car, couldn’t agree upon price, etc.) - using research to assist our salespeople and improve our clients business is a win not only for AutoTrader.com and our dealer customers but also for client-side researchers like me.

Among other metrics, I track the return-on-research of how much we saved from downgrades/cancellations and how much we upsold. That type of information has a lot of internal impact. When I attend sales and executive meetings and show them how we are helping them help our dealer customers and, in turn, help AutoTrader.com, I’m pretty certain that they walk away with a new appreciation for the research function and what we do.