We have had a lot of experience with dealers changing tactics, adding, canceling, etc. There are a lot of factors that come into play when testing the effectiveness of one tactic or strategy. Every dealer has a unique situation (location, brand, history, population, socioeconomic factors, etc.), so unfortunately, the experience or advice from one person (“We dropped xyz and sales increased.”) isn’t going to benefit you that much. You have to test it for yourself.
The best way to do it is to take a scientific approach. Like
@Steve Stauning suggested, start with a good baseline of data that you wish to affect, but make sure you are measuring more than just that one metric. Trying to test one tactic and one KPI probably won’t give you the full picture.
Come up with your hypothesis, but only test one at a time. “Paid Search is a better investment than Cars.com” or “Car Gurus is better than AutoTrader.” Then test it.
What data should you measure?
Don’t look at sales or gross revenue. Those are lagging results, not leading indicators. Look at the activities at your dealership that contribute to the results you want.
Here are a few things to consider:
- Total website traffic
- Count and Percent of total traffic visiting inventory
- Count and Percent of inventory visits that engage (this can be any combination of actions you decide, but should be more than just VDP visits).
- Count and Percent of traffic converting into a lead (email, phone call or chat)
- Count and Percent of leads that convert to appointments
- Count and Percent of appointments to show
- Count and Percent of appointments to sold
- Total ups separate from appointments
- Count and Percent of ups to sold
I would suggest working with at least 6 months of data, and using daily averages. This way you can see trends much more quickly.
Understand and track the variables you do control
These variables are rarely tracked over time, but can have an incredible effect on total units. It’s actually kind of amazing that we will so easily compare YOY stats without this information.
- Product: Particularly with used cars, your inventory can change wildly from month to month, but that is rarely taken into account when understanding leads, sales and gross. You should be tracking mix and number of vehicles in stock.
- Price: The average price of vehicles you have on the ground are going to have a huge effect on leads and sales. Knowing this can help you diagnose problems before they happen.
- People: If the number of salespeople will directly affect the number of sales, then this is something else to track.
- Profit: We all know that gross and volume are directly relative, yet we don’t use this as a KPI. Tracking overtime will tell a story.
How long should you test?
I would argue that 60 days is a minimum and if you really want to learn something you should test for 90 to 180 days. Just based on the consumer’s average buying timeline of 10 – 12 weeks, you need at least two months.
We make the mistake of thinking that money spent this month will activate this month. It doesn’t work that way for buyers. It may take 6 months of ad spend to activate a buyer and create a conversion.
That said, consider the other variables you can’t control:
Interest rates, lease rates, weather, unemployment rates, stock market, home sales, brand trust (think VW 2015), competitive changes (more spending, lower prices, new point, new owners), political climate, etc.
For instance, if you decide to run this test in February and have a record setting year for snow…That is probably going to skew the results. The longer you run your test, the less that things like weather can affect the outcome.
I would also go as far as suggesting that tracking some of these variables over time could help you learn about what does and doesn’t “work”.
One last recommendation…Unless you are trying to settle into a comfort zone, don’t just turn something off and pocket the money. That isn’t a growth strategy. You should be seeking more efficient and effective ways to invest your dollars.
Good luck – it’s a big task, but it’s worth it.
? 
