That’s not meant to scare you… but one of the best tips to optimising Adwords is precise, controlled and measured decisions. Changing things willy-nilly based on a hunch will turn your results into an out of control roller coaster. Plan your changes, monitor the results, and eventually you’ll end up with a campaign giving you a great return-on-investment.
So how do we go about this?
The first thing you have to do is create a baseline of results so you know whether future changes are an improvement or not. This baseline has to be as segmented as possible so you can narrow down performance to certain variables. A lot of new Adwords users will have 1 campaign with 1 ad group covering every option Google offers – this is quick to setup but will cause headaches down the road.
I like to break my campaigns up initially into something like this:
- Search Network Desktop – Consists of search network, desktop device traffic only.
- Search Network Mobile – Consists of search network, mobile devices only.
- Display Network (New Traffic) Text Ads Desktop – Consists of display network, text ads,desktop device traffic only, targeted to new traffic.
- Display Network (New Traffic) Image Ads Desktop – Consists of display network, image ads,desktop device traffic only, targeted to new traffic.
- Display Network (Remarketing) Text Ads Desktop – Consists of display network, text ads,desktop device traffic only, remarketing traffic.
- Display Network (Remarketing) Image Ads Desktop – Consists of display network, image ads,desktop device traffic only, remarketing traffic.
- Display Network (Similar Audience) Text Ads Desktop – Consists of display network, text ads,desktop device traffic only, similar audience traffic.
- Display Network (Similar Audience) Image Ads Desktop – Consists of display network, image ads,desktop device traffic only, similar audience traffic.
Having the campaigns broken up like this lets me see how the different channels and devices perform. I create different mobile and desktop campaigns due to their known performance differences.
The keywords in each campaign should also be broken up into very tight ad group, each shouldn’t contain many keywords and they’ll go from broad to very specific. I’ll cover creating quality ad groups in detail in a future post but for now break them up as much as possible. The reason you want your ad groups broken up is so you can create an ad that talks directly to that ad group and any changes can be tied to that set of keywords.
If you were a Brisbane mobile truck mechanic you would break the keywords up as such:
- Mechanic (+mechanic, +mechanics)
- Brisbane Mechanic (+mechanic +Brisbane, +mechanics +Brisbane)
- Truck Mechanic (+truck +mechanic, +truck +mechanics)
- Mechanic Truck Car Brisbane (+truck +mechanic +Brisbane, +truck +mechanics +Brisbane)
- Mobile Mechanic (+mobile +mechanic, +mobile +mechanics)
- Mobile Brisbane Mechanic (+mobile +mechanic +Brisbane, +mobile +mechanics +Brisbane)
- Mobile Truck Mechanic (+mobile +truck +mechanic, +mobile +truck +mechanics)
- Mechanic Truck Car Brisbane (+mobile +truck +mechanic +Brisbane, +mobile +truck +mechanics +Brisbane)
Next you would create two different ads for each ad group. Why two? You create two ads so you can see which one generates both the highest click-through rate and most conversions. Make these ads totally different from each other to learn what your market responds to. One might mention the price while the other doesn’t; one may ask a question while the other simply states your service as-is. You should set your Adwords Campaign to rotate the ads evenly and indefinitely to best measure this.
You can now run the campaign for the first time to start collecting performance data, enough data to be able to make statistically significant decisions. The length of time this takes depends on the search volume for your market and the conversion rate of your site. If the conversion rate is low then the more traffic you’ll need to be able to make an educated change to the campaign.
Once you are happy with the collected baseline data you can start trying to optimise the campaign. A great place to start is looking at the performance of the ads in each of the ad groups. Go through the ads and pause the lowest performing ad; now create a new ad to replace it. The new ad should be based off the best performing ad but with a change to one element – the headline perhaps. Run the campaign for the same period of time as before and then re-compare the ads performance. Keep trying to beat the best performer.
If you want to get more clicks on your ad then you need to test the bid price and how it changes the ad position. A higher position doesn’t always lead to higher conversions so before making a bid increase, note the current bid, average position, conversions, cost per conversion and conversion rate. Increase the bid and let it run for a period of time to collect enough data to see what effect it had. If it lead to more conversions and the cost per conversion (CPA) is still within your acceptable range then keep increasing and checking until you are maximising the conversions for an acceptable CPA. Eventually you’ll hit a point where the bid price makes the CPA uneconomical – that’s when you back-off back to the sweet spot.
This same process can be repeated for things like time of day, day of the week, devices etc.
What’s important is you don’t make a lot of changes at the same time that will all affect one another. If you change the bid price but also change the schedule then the data will not give a true indication of either change. If the change in schedule causes a drop in CPA but the bid change leads to a significant increase in CPA then you might think the schedule change was a bad thing. If you did one at a time and acted on the results then you’d know that the schedule change was good but the bid adjustment was bad.
If you conduct your Adwords optimisation with the discipline of documenting current performance, making an adjustment, measuring the new performance and concluding whether it was good or bad; you’ll eventually end up with the best performance possible. Making random changes hoping to get lucky will just leave you running around in circle.