How Big Is Your PPC Data Set?

Big data is all the rage these days. I’ve written about big data in PPC before. This isn’t a post on big data per se – but this week I’ve felt the need to talk about how much data is needed to make good decisions in PPC.

Let’s start on the micro side. A couple years ago, I wrote a post on PPC testing and why days are not data. I stole that phrase from my friend Andrew Goodman, but it’s so good that I find myself using it frequently, including every day this week.

Teaching people and clients about PPC is a passion of mine. I love to help others learn about the career that I love. Sometimes, though, the great aspects of PPC such as quick launches, instant data, and cool reporting get overstated. Suddenly you have a client or boss who wants a daily detailed report on his or her PPC campaign progress.

I don’t advocate that. I talk about that at length in my Days are not Data post, but in a nutshell, a one-day snapshot is full of too many normal fluctuations to be meaningful. Those unfamiliar with the ebb and flow of PPC get too bogged down with the daily deluge, causing unnecessary worry and alarm.

I try to remind these folks that they hired a PPC pro for a reason. We DO watch the data on a daily basis and adjust as needed. We just don’t make pass-or-fail judgments on one day’s worth of stats.

Now let’s look at the big data set of the coin. I wrote a post for Search Engine Watch this week called Do the PPC Engines Reward the Right Behaviors? It was a fun post to write – I’d been mulling it over in my head for literally a year, and finally the time was right to write the post.

In the post, I stated that Enhanced Campaigns are a case of Google rewarding for A while hoping for B – rewarding advertisers with lots of levers, while hoping they’ll create fewer campaigns.

While the number of comments and feedback on the post weren’t overwhelming, they were definitely interesting. Most people agreed that Google made things worse for advertisers and themselves with Enhanced Campaigns.

But Larry Kim disagreed with me. He has been out there trumpeting the nirvana of Enhanced Campaigns ever since they were in beta. Therefore, I wasn’t at all surprised with his stance.

I have a ton of respect for Larry and have no problem with what he said. But I still disagree.

Enhanced campaigns are fine for smaller advertisers with simple settings and campaigns. They’re good for local advertisers who previously had trouble hyper-targeting their PPC.

But for those of us running complex campaigns with diverse goals and objectives, Enhanced Campaigns are a nightmare. Several large PPC companies have written about their tribulations with Enhanced Campaigns, including higher CPCs and worsening performance across millions of dollars of spend. Matt Van Wagner is moving budget to Bing because of them. We’re seeing weirdness with our clients who’ve transitioned, including the same CPC spikes that others are claiming.

I’m not questioning the veracity of Larry’s data. I’m sure it’s accurate for his client set. But for most of us, the PPC big data says that Enhanced Campaigns are bad news.

What do you think about PPC data? When is enough enough? Share in the comments!

Related Posts:

Big Data and Big Goals

This week, in the United States, Barack Obama was reelected as President. It was a close race that really came down to key swing states.

But this post isn’t about politics. It’s not about the good, bad, or ugly of the recent election season and its outcome.  It’s not about who won or lost and why we should love them or hate them.  Rather, it’s a reaction to Time Magazine’s article on the role that big data played in the President’s win. The story was featured on Anderson Cooper’s CNN news magazine last night.

The Time Magazine piece details the various tactics that the reelection team used to target key voters. That’s not the point of this post either, although I find it utterly fascinating.

This post is the product of a conversation I had about the article.  The individual with whom I was conversing expressed dismay that rather than focusing on the issues, the Obama team focused on using and manipulating data (and the voters) to their advantage.  It’s true – there wasn’t much focus on the issues this year (good, bad, or indifferent – remember, this isn’t about politics).

My reply was that communicating the issues wasn’t the goal. The data crunchers did exactly what they were supposed to do – they cranked out models that showed the Obama marketing team exactly who they should focus on, and what tactics they should use to reach them. All of their testing, and they did plenty, was focused on figuring out the best message for fundraising, recruiting volunteers, and getting people out to vote. They used big data to predict voter turnout – which was then used to plan the President’s final campaign visits.  The data was even utilized to decide the best media for ad placement – rather than using traditional optimization tactics, they were able to get laser-focused with ads right from the start.

The reason they did all of this? Because the goal was to get the President reelected.  And it worked.

But could the Obama team have used big data to focus on the issues, as my conversant lamented?  I believe they could have. If their marching orders were to get the word out to voters on key issues, they would undoubtedly have used the data in a different way.

For example, they could have found out who was likely to be concerned about health care, and tailored and tested email messages about the health care initiative to those individuals.  They could have communicated the President’s stance on gay marriage to gays, on women’s rights to women, on business issues to business leaders, etc.  And incentives could have been matched to those audiences – instead of dinner with Sarah Jessica Parker, maybe they would have offered a chance to attend a town hall to ask questions about economic reform.

What if the team didn’t have a clear goal? What if they were let loose to just crunch numbers and make recommendations?

Think about this for a minute.  I know plenty of PPC advertisers who say “we need to start a PPC campaign,” yet they have no idea what they want to get out of it. In fact, this is the most common reason that PPC engagements fail.

Nowhere is goal-setting more important than the application of big data. Data without goals is just worthless gibberish – not to mention a black hole of time- and money-sucking quicksand. But once goals are clearly established, the tactics become clear.

And in the end, big data is like politics. You can’t please everyone. But with clear goals in mind, you can please your boss.

Hot Off the Press! For more on the big data analytics lesson, check out this post from the Harvard Business Review.

Related Posts: