PPC Campaigns Blog Posts

One Adgroup Fits All: Should I split my keywords down into one per Adgroup? This has been a topic of much debate with even the most experienced PPC advertisers coming to blows. This blog doesn’t aim to solve the mystery once and for all, but simply to look at the pros and cons in a few situations. Answering such a question is very much circumstance based; there is almost certainly no straightforward answer for most Adwords users. Let’s take a look at the following two scenarios:

1. Find-me-a-fast-car A large company that wishes to advertise second hand sports cars nationwide. Conversions are not made online, but over the phone. The aim of the campaign is to generate as many clicks as possible for their budget.
2. Creepy Cravings A small company that sells food and equipment for pet spiders, snakes and reptiles. The customer can order online for delivery.

Find-Me-A-Fast-Car

We are talking about a large campaign here where tens of thousands of keywords are delivering hundreds of thousands of clicks. It’s impractical to break the keywords down into individual Adgroups. The user might group keywords together, based on brand names like this

Ferrari (clicks/day)		Porsche				Lada
[ferrari] (100)			[porsche]			[lada]
[testarossa] (11)		[boxster]			[riva]
[testerosa] (6)			[boxter]			[used riva]
[ferrari  testarossa] (10)	[porsche boxster]		[lada riva]
[used testarossa] (8)		[boxster porsche]		[used lada riva]

Second-Hand Ferraris 			Used Lada Supercars
Live The Dream 				Live Your Grandparents’ Dream
Call Us Now For More Info 		Offers Of £50 Or More Accepted
www.find-me-a-fast-car.com 		www.find-me-a-fast-car.com

The question is, should they be splitting these 3 Adgroups down into 15? Let’s say the user currently spends 1 hour a week writing new adverts and adjusting bids. By splitting the keywords up, the user must now spend 5 hours writing new adverts and adjusting bids. For a large campaign like this, we’re probably looking at a more realistic figure of 50 car manufacturers, with 20 keywords per group minimum. That’s 1000 Adgroups, testing two adverts each, which is … a lot of hours. Ok, so our Adwords Expert is a dab hand with Adwords Editor and manages to fit all this advert writing in to their busy schedule, what do they gain? – Let’s see. Splitting up [ferrari] and [testarossa] would almost certainly improve the campaign’s click-through rate with better targeted adverts being allowed, but what about [testarossa], [testarosa], [ferrari testarossa] and [used testarossa]? Consider adverts for [testarossa] with the following titles:

The adverts have been running for 2 days but require another 25 days to reach a 95% significance level! (www.splittester.com) That’s nearly a month to make a small step forward. So what if we leave [testarossa], [testarosa], [ferrari testarossa] and [used testarossa] in the same Adgroup:

The same two days generates enough information to be 80% sure the adverts are different, and only 6 more days to get a 95% significance. It may be that grouping these keywords together has a slight detrimental effect on the clickthrough rate – a few of the keywords may have worked better on the old advert – but the potential benefits of the faster turnaround on the advert test will generally outweigh this. Even if the clickthrough rate on the new advert had only been 11% (39 clicks), the test would still have ended more quickly. But being able to run more tests will lead to more improvements in the clickthrough rate over a period of time, and the time taken to write hundreds of additional adverts could be better spent elsewhere. In this instance, splitting out the keywords would involve four times the amount of work, for just a few additional clicks…

Creepy Cravings

This campaign is clearly different, as the number of keywords is limited and we can see which ones convert. So let’s begin with a clustered set-up:

Spiders 			Snakes 				Iguanas
[spider] 			[snake] 			[iguana]
[spider food] 			[snakes] 			[iguanas]
“spider tank” 			“snakes 			“iguanas”
“spider” 			“snake” 			“iguana”

Because it would be very difficult to write varied adverts that are relevant to all the keywords in these Adgroups, we should certainly split them down:

Spider Tanks
[spider tank]
[spider tanks]
[cheap spider tanks]

This is much better – we can write adverts relevant to the search queries. These are a few of our ideas for adverts, and statistics after a week:

(1) Spider Tank 		(2) Spider Tanks 		(3) Cheap Spider Tanks
Looking For Your Spider? 	Looking For Your Spider? 	Looking For Your Spider?
Put Him In A Tank 		Put Him In A Tank 		Put Him In A Tank

CTR = 13.3% 			CTR = 14.4% 			CTR = 8.3%

Being the experts in campaign optimisation we conclude that advert 3 doesn’t work and it gets the sack. However, are we really making the most of our campaign? Instead of deleting adverts, we run some tests and find that when we move [cheap spider tanks] into its own Adgroup with advert 3 and then delete advert 3 from our old Adgroup this is what happens:

(1) Spider Tank 		(2) Spider Tanks 		(3) Cheap Spider Tanks
Looking For Your Spider? 	Looking For Your Spider? 	Looking For Your Spider?
Put Him In A Tank 		Put Him In A Tank 		Put Him In A Tank

CTR = 14.3% 			CTR = 15.7% 			CTR = 20%

So now we’re smugger than Larry when he said “What about calling it Google?” But have we gone the whole hog? Why don’t we take what we’ve learned and split up [spider tank] and [spider tanks]:

(1) Spider Tank 		(2) Spider Tanks 		(3) Cheap Spider Tanks
Looking For Your Spider? 	Looking For Your Spider? 	Looking For Your Spider?
Put Him In A Tank 		Put Him In A Tank 		Put Him In A Tank

CTR = 20% 			CTR = 20% 			CTR = 20%

So now you’re thinking of changing your name to Mr. Smug and buying the biggest house in Smuggleton. You’ve taken what would have been a very satisfactory Adgroup with a 14.4% CTR and split it up into three Adgroups with 20% CTRs.

Conclusion

In the first campaign there are more clicks to be had than we could possibly afford, and so our aim was to get as many cheap ones as possible. This means we have a huge keyword list which we expand regularly, with Search Query Reports, Google suggestions, etc. The bids have to be increased or decreased every time a change is made to the campaign, which means that trying to tweak every keyword becomes a full time job. We can increase the clickthrough rate on one keyword by writing an advert to match it and testing until significant. The effect may only be to decrease the average cost per click of the whole campaign by a fraction of a penny. This is a lot of effort for such a small gain. However, Adgroups with large numbers of keywords would be able to test many more adverts in that time and perhaps improve the clickthrough rate by almost the same amount, but with much more impact on the campaign. In the second campaign every click has a potential to change the success of the campaign by converting. It is much more important in this case to squeeze out every relevant click. So, in this type of campaign, investing more time into individual Adgroups will yield a greater increase in the profitability of the whole campaign. In a real campaign the choices and results will generally be less obvious, but with work the results can be very worthwhile. For such a niche campaign it’s likely that you’ll have budget to pay for every available click and so getting them is a priority. The cost of each click in such campaigns is less important, so long as the cost per acquisition remains profitable. These are two extreme examples. In the average campaign the choice of whether and how much to split down the Adgroups will be dependent on 2 or 3 factors; How much time you have & how much traffic you get (& if you can’t think of ideas for adverts).