Man Vs. Machine In SEM
There seems to be an ongoing debate in the SEM world on the role of people vs. technology. Of course, it’s not an either/or argument. As my main man T-Bo likes to say, companies must have the right balance of people, process, and technology. And he’s got the P-P-T.ppt to prove it!
I’ve worked on both the agency and tech side of SEM, and I can tell you that there is certainly a role for both people and technology -- and some process, too. The key is knowing which parts are best left to (wo)man and which are best left to machine. Here are a few examples.
1. Goal-setting and KPIs. Are you trying to increase market share? Generate profitable sales? Improve margins? These are questions only humans can answer, and these answers will vary by organization, brand, and product. After choosing top-line business goals, the next step is setting KPIs -- and, again, this is something that people should handle. What metrics should you hold your marketing and, in turn, your SEM campaigns accountable for? Impressions and clicks? Conversions and sales? CPA? ROI? Without clear goals and KPIs, even the best technology will fail.
2. Brand positioning. Are you the high-end brand or the discount player? Are you competing on value or price? What’s your key differentiator? What do your customers expect from you? What does your brand stand for? These are critical decisions that need to be made (and tweaked!) by people who’ve done the proper research and truly understand their audience. Technology can help you monitor competitors and recommend optimal pricing and merchandising models, but the guts of your brand persona must come from a person.
3. Analysis and insights. What do the latest CPC trends mean for your business or vertical? How can you improve your website based on the latest SEM conversion rates? What does it say about your target audience if your overall ROI is stronger on Android devices than iOS? Machines can track activity and surface data points at an unprecedented rate, but only mankind can make the giant leap to conclusions.
1. Keyword selection and expansion. What keywords should you buy via PPC or build assets around for SEO? What are all the ways consumers might be searching for your products and services? When it comes to building comprehensive targeted keyword lists, machines routinely outperform people. Technology solutions can take product feeds or website maps and correlate them to relevant keywords including head, torso, and tail terms, misspellings, geographic concatenations, translations, and even other categorical taxonomies. Doing this manually might leave less room for error but is not scalable when you’re dealing with a robust product catalog or high volumes of individual campaigns.
2. Bid optimization. What impact will raising your bids on keyword X have on your overall goal? Where should you reallocate budget from in order to invest further in keyword X? Is keyword X the best place to spend your next dollar, or should it be keyword Y? If you spend more on keyword X, what will happen to keyword Y? Advanced bid optimization requires millions of interdependent calculations for each keyword in each portfolio each minute of each day. Not even Ken Jennings could compete with that.
3. Reporting and attribution. How many new customers have you gotten from Yahoo and Bing vs. Google? Are people engaging with your brand on Facebook before searching for your products and buying? How should you distribute the credit among all ads in a purchase path? Tracking and reporting are obvious candidates for technology automation, but trusting a machine to do cross-channel attribution can take some getting used to. Fortunately, sophisticated SEM technology platforms can provide attribution model simulators to show you what your overall results would look like depending on what settings you choose. Once you’ve found the model that best reflects your business environment -- a task I’d reserve for humans as part of goal-setting with the occasional recalibration -- you can let Deep Blue go to town.
When it comes to SEM, in whom do you trust?