A couple weeks ago I wrote an article about desired outcomes in PPC. A big idea behind that article is to know what you are doing and why you are doing. This is a follow-up to that and some advice on how to retrace your steps in an effort to duplicate great account performance of past. I’ll also give you some insight into how I go about diagnosing account issues. In my current role as Account Supervisor, I have my hands in every Hanapin Marketing client account from an analysis and strategy perspective. Sometimes you simply need another set of eyes on an account to realize opportunity and risk in PPC and my role is to be that set of eyes, giving us the best chance to capitalize on our client’s opportunities and minimize their risk.
Because we have a good amount of clients, the amount of touch time I have with each is extremely valuable, putting an emphasis on efficiency. With that said, I typically need to dig a little deeper because our staff is extremely talented, which makes it difficult for me to poke holes in their existing strategies. I’ve found that the most loyal tool in my analysis arsenal is the AdWords graph feature (And AdCenter’s now that they’ve added this). It’s always the start of any diagnosis I’m conducting. It allows me to quickly sort through and visualize large data sets at many levels of granularity and at several different periods of time. What I’m looking for with this information are trends. I want to identify changes in the account and draw attention to significant or subtle changes in the account that can be acted upon. Here are some examples of graph settings I look at and what I’m looking for in each:
Year Over Year CPL Performance:
I take a look at year over year performance to have a better understanding of seasonality and how far an account has come in the past 12 months. The above chart is a branded AdGroup. You can see that 2012 January performance on cost per conversion is considerably better than 2011. Not only is it better, it’s also more consistent as the line is more flat than previous years. These are the types of results I want to see and are something that is valuable to show a client, especially on accounts with a high level of seasonality where month over month stats may vary. This chart only shows part of the performance though. To get a full understanding of the performance you need to start swapping the first metric:
Year-Over-Year Cost Performance:
This is a shot of the same AdGroup but looking at cost. What this allows me to see is that, while CPL did improve dramatically on the previous chart, spend actually decreased. Sometimes true performance can be clouded by the performance of metrics and drawing out additional information could allow me to better understand what is happening on an account. An assumption I can make here, and one that took only a few seconds, is that CPL performance improved because this account manager “cut out the fat.” The manager probably reduced spend on keywords that weren’t working and probably invested more heavily in the keywords that were working, which allowed her to spend more efficiently and produce a lower CPL. To prove this, I’ll need to look at one more metric for year-over-year.
Year-Over-Year Conversion Performance:
Here’s that same AdGroup with the same date range on conversions. As you can see, 2012 is actually driving more conversions than 2011. This is what ties it all together and proves to me that optimizations over the past year have resulted in significant improvements in account performance for this branded AdGroup. This account manager is driving a significantly higher volume of leads for a significantly lower cost, resulting in higher budget efficiency. Now I can go give her a high five. Plus, it took me less than a minute to run through this entire process. I could scrub the entire account at the ad group level (30 AdGroups) in this manner in less than 10 minutes. With that said, looking at year over year data isn’t necessarily enough to have a full understanding of the accounts performance. From here, I’ll go into a six month trend analysis, starting with:
Clicks Versus Cost Per Click:
Now I’m at the account level and I’ve stretched my date range back about six months. The first thing I want to see is how the account was managed from peak season, which on this account is August, through their slow season, which on this account is December. I already know that on this account, budget is reduced through the slow season and they are limited by budget so there is no shortage of opportunities to find clicks. What this report is showing me is that cost-per-click shrank in relation to overall clicks. I assume that this was a tactic to mitigate a loss of traffic by lowering position and driving more volume for each dollar. During the high months, I’d imagine position was raised to gain more exposure and relevance when people are buying. I’ll confirm this by swapping cost per click for average position, which I did and the graph confirmed my assumed results. Now I want to know if the strategy paid off.
Conversions versus Cost:
Here’s where things get interesting on this account. As you can see in the business months at the start of the graph, the relationship of conversions and cost was close; high conversions and high cost. In the slower months a gap forms where conversions are peaking out considerably higher than cost. This shows better efficiency in the slow months, which we already noticed on the AdGroup graphs earlier. You can also see towards the end of this graph that the trend converges again as we head into January, which is one of this clients busier months. What this arms me with, is the ability to realign strategy with the client and walk them through the progression of their account. The best part is, I don’t actually push buttons or pull levers in this account and I’ve got the trend and strategy over the past six months pegged in about 15 minutes worth of research.
Performance on this account was outstanding so there wasn’t much for me to dig deep into but a similar analysis using only graphs on another account could uproot some items that need immediate attention. When I find trends that don’t look great to me, I zoom in on the date ranges in question and start doing a comparative analysis of the time period directly preceding the goof. It’s then that I’ll start looking at overall metrics to see what shifted and I’ll analyze change history reports to see if it was user caused. If so, we can easily go back and revert to the previous state and pick up with a new optimization. Analyzing accounts doesn’t always take a long time but you should make a habit of analyzing your accounts on a short-term and long-term basis regularly. If you only stay within the day, week, or month, you might be losing site of just how much your account is changing.
