Lesson 1 of running a PPC campaign: track your conversions. A simple rule that we all try to stick to. But sometimes we can't.
Not all campaigns work so easily. It may be that conversions are too few and far between to make a difference. The website may convert everything over the phone through a system to complex/entrenched to track back to keyword level.
Whatever the reason, there are two techniques you can use to help get your PPC account delivering more value, even if you can't track it.
Option 1: Engagement Metrics
This one is hopefully going to seem like an obvious first step to most of you. If you have Google Analytics linked to your AdWords account, then you can create "engagement goals" and import them into AdWords as a conversion.
Engagement goals consist of tracking visitors who meet a threshold time on site or pages visited. In the screenshot above you can see how a goal would be set up to create a pages visited requirement of greater than 1. Doing this would let you see a goal in AdWords for any visits that did not bounce.
Based on the type of website you're running you really need to think carefully about which metric (time on site or pages visited) is going to be more useful for you, and what your requirement is going to be to log it as a conversion in AdWords.
Remember! Google Analytics does not track the time the user spent on the final page of their visit. If the user read that page for a long period then left the site, you won't see how long they spent on that final page. This can make time on site a slightly misleading metric for websites where visitors won't view many pages.
If you want to log pages per visit, then decide what kind of visitors qualify as "engaged" enough to be worth logging. You will find that many visitors will either bounce, or look at everything on the site. Averages suck. Your averages are being skewed by bounces and by visits with many many pages viewed, so don't base your assessment on logging visitors who visit more pages than your site average.
Visiting more than one page (i.e., not a bounce) can be a good signal. If it takes at least two clicks to reach a product page or view a price, then your ideal threshold may be greater than two pages viewed.
Option 2: Relative Weighting
For some sites you can't even accurately track the above behavior. One page visits might still be worth a lot to an information or blog site. You may have organizational reasons that you can't use Google Analytics or other web analysis packages.
You need to be more creative in these situations.
Take your top 20 keywords by traffic. In most cases these will account for the majority of your total clicks, but if not then take 30, or 40. Don't take too many at this stage, or you're making a lot of extra work for yourself for not too much value. Exclude any of your brand keywords from this list.
Rank these keywords by a "likelihood of conversion" factor. You need to know your business really well here, but you have to put your keywords into order sorted by how relevant and targeted they are. A particular keyword might suggest that everybody searching for it is looking for exactly your product. Alternatively it might have other meanings or be more generic. Rank these keywords by their "awesomeness" factor.
Give each of these keywords a score from 1 to 10 for relevance. The best and highest should be 9 or 10 (this won't always be the case, if none of your best keywords are in your top 20 by traffic) and work from there. Don't feel like you need to go all the way down to 1. If the lowest keywords have maybe half the likelihood of conversion of the highest keywords, then respective scores of 5 and 10 would be appropriate.
Add a column to your list that indicates the potential value of a conversion. This is only really necessary if different keywords imply a different value (e.g., they might be searches for a different product). If your website sells just one service, or value is not related to search term at all, skip this step and just add "1" to this column next to every keyword.
Multiply these scores together for each keyword, and divide each one by the total sum of all scores added together. What you have now is an estimate for what proportion of your budget you should be spending on each keyword.
Remember that these keywords didn't make up your total spend. If the top 20 keywords spent 80 percent of your budget, then these proportions you've found are proportions of the 80 percent.
Deciding the Bids
You should now have each of your top 20 keywords with a score from 0 to 1 representing the proportion of your spend that should be going to this keyword. Achieving this takes some time. You're going to need to analyze your spend levels on a daily basis, and make changes accordingly.
Each day, download the keyword data for the previous day from the account. For each keyword in your list, work out how much it spent (proportionally) compared to your target. If the keyword's spend is too high, reduce the bid by 10 percent. If it is too low, increase the bid by 10 percent. Set a sensible level of tolerance on this, or you'll be changing your bids even if your keywords are reeeeeeally close to perfect.
The process that will occur is this: every day each keyword that is spending too much will have its bid reduced a little. If the next day it is spending correctly, then you'll stop making that change to that keyword and it's nicely on target. If it still hasn't reached target then the process will happen again.
The Inevitable Complication
You're working from assumptions based on gut instinct and guesswork. These aren't accurate and they never will be.
Don't go crazy looking to make each keyword fit your target perfectly, since your target is an educated guess at best. The process outlined above will give you a bidding strategy to meet your target but it doesn't validate it if you were wrong.
The second issue to be aware of is that every time you change your bids, the proportion of budget spent by each keyword will change. But it will affect the proportion spent by all your other keywords too.
If we imagine a situation with two keywords, "keyword a" and "keyword b". "Keyword a" spends $10 per day and "keyword b" spends $20 per day. If I conduct this analysis and increase the bids on "keyword a" so that it spends $20 per day, then "keyword b" is now spending a lower proportion of my new, higher budget. And if my budget is limited, then I may need to reduce "keyword b" to compensate.
Each action on a keyword affects the proportional spend on the others. Don't be surprised if your keywords overreact.
Having a lack of good conversion data is no excuse for not optimizing bids in a campaign. The campaign can still perform better and deep down you know it!
The two methods in this article will give you some pointers to go about pushing your spend towards your best converters and most profitable keywords, but your overall budget must still be determined by the profitability of PPC as a whole.
Re-assess your assumptions regularly. Don't take your first best guess as fixed. Make changes and improvements over time as your knowledge grows and you can keep making the campaign better and better.
Optimising Digital Marketing Campaigns with Search, Social and Analytics
At SES London (9-11 Feb) you'll get an overview of the latest tools, tips, and tactics in Paid, Owned, Earned, Integrated Media and Business Intelligence to streamline your marketing campaigns in 2015. Register by 31 October to take advantage of Early Bird Rates.