Rght now, it is 01:27 AM BST (that’s 00:27 Universal Time) or 8:27 pm Eastern or 5:27 pm Pacific. It’s also July 14 for me, and July 13 for the US. So take a look at this screen shot:

If I select “Today” then the User Interface shows the date as the 14th July and the 13th July. The stats reported are updating for the 13th July, but the graphing tool is showing “Today” as the 14th. There are no clicks in reports or in the UI, for July 14th, only for July 13th. It’ll become even more complicated in a few hours, when the East Coast crosses midnight. Use the wrong mix of Ad Scheduling times and you may never work out exactly when the adverts are actually running, as Google becomes confused about which of the three possible time zones is being referenced, and which tool is being set to use which TZ.

Reporting Delays

Complicating this even further is Google’s little note in faded grey at the bottom of the stats table:

Yup – “clicks and impressions in the last three hours” and “24 hour delay in conversion reporting”.

The good news is that the delays aren’t usually that much. That chart above shows some conversions. It’s probably close enough to right. AFAICS, the stats *USUALLY* take about 45 minutes to stabilise. There’s some delays that can cause the stats to lag for longer, and, more importantly for many users, the numbers may decrease over three hours and even for a matter of months later. That’s because the invalid impression and invalid click detection software (click fraud detection) takes a while to work out that a user has a pattern suggestive of suspicious behaviour. The UI and reporting tools do not report clicks, but *billable* clicks. That means you can see a click, which is later withdrawn. In extreme cases you may see a conversion with neither impression or click – someone, identified by Google as a likely fraudulent user, has purchased a product but their impressions and clicks have been ignored.

So, if you believe anything, believe that the latest updating stats are substantially correct for billing purposes as of about 45 minutes ago. Definitely don’t go adjusting your bids as if the stats were being reported in real time… And it becomes more complex yet when you consider the delays in reporting stats in Google Analytics. A topic for some other time, perhaps. :)

Summary

If you think that you’re confused about when something is happening with when things appear to occur in AdWords, it sometimes appears to confuse and puzzle Google, too. Be aware that you may know more about what time events happened locally, than Google does.

CPA – Cost Per Action

CPA, or Cost Per Action, is commonly used by clients who can’t explicitly track advertising costs into specific user purchases. It is also occasionally used by clients who don’t want the details of their profit margins known, or who don’t want to manage the complexity of identifying varying margins. IME, that’s also fairly common with marketeers who have a sales background, and are used to the “pipeline” and managing the costs of the sales pipeline.

Other examples are businesses that have a “drive to offline” – that is, the initial search takes you to a business that involves professional services, and takes leads either as webforms, emails or phone calls, but where the provision of the service involves steps that completely break the relationship to the advert. In other words, you might be able to track conversion to the point of completing a webform, but no further. You might use a collection of phone numbers, and tweak the webserver to deliver a recognisable phone number, so you can tell when the prospect came from paid or organic search, but you’ll still only end up with an aggregate of calls from paid search, unassigned to particular searches or adverts or even the actual time that the advert was seen.

Assuming that you have some activity that is measurable and identifiable as having a paid search source, you can simply divide advertising costs by the number of leads, to get the cost per action.

Simple as though that sounds, it isn’t an excuse to get sloppy with the account management. Just because the metric is a simple division, the management of adverts to optimise CTR and reduce the AvCPC is still important. Don’t confuse simple measurements with simplicity of advertising. A common mistake that we see is to, for example, use Broad Match only, and huge AdGroups with a single advertisement. This is justified on the grounds that the simplicity of measurement doesn’t justify the complexity to set up in more detail. While you might not be able to track the individual effect of each advert, you can certainly optimise the response to each search and increase the number of clicks for each budgeted activity – allowing you to increase the total volume of business.

I’ll use some non-specific, but representative numbers for the following discussion. Imagine that you have a $10.00 CPA. You could use a single large AdGroup, bid $2.00, achieve a 2.5% average CTR, and pay an average of $1.50 per click. You’ll therefore have to achieve an action with $10.00 / $1.50 = 6 clicks. With the right product and advert and landing page this is achievable – though not for mass consumer brand articles, where you might expect closer to a 1% conversion rate. If you can improve the CTR of the adverts, by splitting the keywords into smaller groups, you might bump up the CTR to 10%. In this case, you might see cost decreases for each click, to something closer to $1.00. That would allow you to increase your bid (and hence your position, and hence your total count of leads, at the same CPA) or you could add more keywords, the ones that you found barely economic previously, and use the spared budget to make some less well converting or lower click rate traffic. Either way, the account will benefit from paying extra attention to improving CTR, even though you can’t measure which advert is generating responses, or the time of day of the search.

There are techniques that will let you get you closer to working out what time of day most affects users, and which adverts are most effective, but most of them involve some hairy maths, and heavy use of the Google AdWords API to automate the various activities. We’ll save the details of that for another time – this article is about CPA and the implications on paid search complexity. :)

Whether this is meaningful depends on the costs of improving the advertising, the time it takes for Google to provide a lower AvCPC in response to a higher CTR, and the size of the advertising budget.

As a rule of thumb, for lower budgets of less than a hundred dollars per day, the additional costs of professional setup may not be justified. Somewhere around a couple of thousand spend per month, the costs of having a professional set up the multiple AdGroups may become worthwhile. Below that, your business probably has other concerns that should be more profitably occupying your time and resources.

Summary

Although Cost Per Action is one of the most simple performance measurement methods for paid search, it doesn’t justify a reduced standard of setting up and managing paid search.

Optimising a CPA measured account can still yield significant improvements in total profit, or decreases in advertising spend for a given level of activity.

Whether to expend the effort or fees to set up an optimised account will depend on the spend and the likely improvement.