Keywords are awesome. We can't target an individual person on search, we can target a group of people who share a common characteristic: they've typed in a specific term (or offshoot/variant of that term).
Keywords offer us a lot of granularity (giving us a way to group people into smaller buckets). Keywords also actually tell us something about the person. That's valuable. We all know about the intent-based wonders of marketing on search.
But Keywords Aren't Enough
Granularity is good. So surely more granularity is better? It's true.
There are lots of reasons a person might search for your brand name. Someone might have seen your TV ad and want to buy your product. A job candidate might be preparing for an interview. A delivery driver might be looking for your address.
If I gave you five minutes and asked you to write down every reason somebody might search for your brand term, I'm sure you could come up with dozens. So that's a good example of a keyword grouping that didn't really group people very much.
It works for non-brand keywords too. If I sell flowers, I might advertise on "red roses". Admittedly that's a bit broad, but let's just try the same exercise. Somebody searching for "red roses" might be looking to order a bouquet online, but they might also be after a local store that stocks them, or looking for pictures for an online article, or trying to learn what chemical makes them red.
So a keyword is a good start, but it's not enough to really break people down into potential purchasers versus others.
Adding Granularity Through Targeting
We've been given new buckets. New targeting methods in addition to keywords that we can use to help us add granularity.
At first it doesn't sound like it. The new targets are pretty broad. You can read our analysis of them here.
Time of day, location, device and remarketing status are pretty big buckets. But we're not targeting them on their own. Always in conjunction with a keyword.
What is a Target, Really?
Targets in this sense are any of the features we know about a person. At the moment it's just the list above. But targets aren't helpful to us, what we need are audiences.
Choosing Your Target Audiences
Yes. An audience (by my definition) is a group of targets that specify the smallest possible bucket of people.
Imagine that we only have two devices: desktop and mobile. That immediately splits our "red roses" searches into two smaller groups: mobile red roses and desktop red roses. We don't know much more, but we know something.
If it's fair to assume that each of these groups has a different likelihood of purchasing online from us, then we can use that information (the conversion rate) to finagle our bids.
Now let's break locations down into two: areas I can deliver in and areas I can't deliver in. That immediately splits our two groups into four. That's a lot more information than we had with just the keyword.
Now let's add the time of day (and keep it simple again): when our store is open or when it's closed. Eight groups! From just one keyword!
Many of these combinations carry no special connotations beyond the keyword itself.
If somebody is looking to order something online in the future, then my opening hours probably don't matter. If somebody is on their mobile near my store, then the opening hours definitely do!
So now we need to think about those groups that will definitely behave differently from other groups. Those are my audiences.
People who have purchased from me before, who are in my delivery area (or searching for a location in my delivery area, we have that control). That's a great audience.
People who are looking for generics and are near my store while I'm open. Bingo!
Your homework assignment is to think about the targeting methods available, and choose the target audiences. Those really granular groups that we can achieve by layering all the targets onto each other.
Identify Important Groups
There are far too many groups. Like, a ridiculous number.
We used simple breakdowns in our examples above to look at locations, or time of day. Actually different locations or times or days of the week might have different conversion rates, and deserve different bids.
Once you start adding in all your varied remarketing lists things get really interesting. So you're not talking about eight groups per keyword, but dozens of groups per keyword.
Identifying the good ones is a non-trivial task. But it can be done. Because you only need to identify the really important groups up front, and the rest can wait until you start to see the data – previous customers, people really near you (especially on a brand search), etc.
Think about your own business and create this list.
How We Need to Plan Paid Search Campaigns in the Future
It isn't just about eeking out a little extra performance now. This is a wholesale change in the way you should be thinking about campaigns.
Keywords are still part of it, but you need to constantly be thinking about what every different piece of information you know about a person means for your audiences.
How do you use this to target people? What ads should you be using? How should you be bidding?
For the next year the best paid search managers will be the people who can understand (and explain to others) the core concepts of how to target these people above and beyond keyword groups. When you've got that in your head comfortably you will be able to see how search fits in a modern strategy, and start looking for those same audiences across the web.
Twitter Canada MD Kirstine Stewart to Keynote Toronto
ClickZ Live Toronto (May 14-16) is a new event addressing the rapidly changing landscape that digital marketers face. The agenda focuses on customer engagement and attaining maximum ROI through online marketing efforts across paid, owned & earned media. Register now and save!