How do you assess PPC ROI?

[CMC-SD]

I thought this would be pretty straightforward, but I'm trying to put together my first PPC ROI spreadsheet and it's turning out to be much more complicated than I had anticipated. Would anyone be willing to tell me what you look at to determine the return you're getting on PPC -- and if you're using AdWords and Analytics, what screens/reports you rely on? It seems to be more complex than simply comparing Paid Search Revenue to spend ... but I could be overcomplicating things.

[john4math]

I built my own Google Analytics custom reports which I export and add a few equations to in order to track ROI. I do it for a subscription site so it's not quite as complicated as it would be for an eCommerce site, where each purchase can have different items, margins, cost of goods sold, etc.  What are you looking to see beyond paid search revenue to spend?

GA may be making this easier in 2013 with their "Universal Analytics" offering: http://analytics.blogspot.com/2012/10/google-analytics-summit-2013-whats-new.html.

[AWCthreads]

Know your margins and watch the cost per conversion for your products.

Eventually, you'll get a good idea of what I call a "conversion threshold" which tells me what I can expect to spend to convert that product to a sale. From my experience, once the conversion threshold is determined, it won't deviate far from it over time.

If my net on the product is $20 and my conversion threshold is $12 then I'm making money (excluding cost for shipping, handling, promo discounts etc).

If the threshold his $18 or $22 then I re-think what to do with the product.

I'd put together a report that shows product cost, retail price, margin, net and cost per conversion. It should be quite telling.

[alinkfred]

I sell a downloadable product and a significant amount of my revenue is from Google adwords. We track the conversion rate of every keyword. An important factor is the click through ratio. You can get a lot of clicks and few orders. The important thing is to get the balance right. Any keyword with the word free will get a lot of clicks but very few orders.

Another important factor is to ensure that you are using negative keywords properly.  On my site we sell generic set of speeches. We would have a negative keywords for therapy or disorder. Somebody looking for content to help with a speech is not interested in speech therapy or speech disorders.

Another area is to check your ROI in different countries, and only bid for clicks in countries where you can make an adequate ROI

[CMC-SD]

I'm looking to compare spend to revenue from transactions that are directly attributable to that spend. So if I know how much I spent between Oct 1 and Oct 31, I want to know how much revenue resulted from those clicks. Not how many transactions occurred during that period, since that might include clicks from before Oct 1.

The problem is, AdWords includes conversions where the paid click might not actually be responsible for the transaction. If someone searches [red widgets], clicks an ad, comes to our website, leaves without purchasing, searches [blue widgets], clicks an organic result, and purchases, Analytics will correctly attribute that sale to organic. AdWords, however, will include that conversion in its cost/conv metric. Skewing the cost/conv number lower, I might add.

I've spoken with several AdWords/Analytics googlers about this, and they don't seem to know of a report that does what I want. I guess I'm wondering if others are trying to measure the same thing, or if they're comfortable with the less precise AdWords-supplied cost/conv figures.

[CMC-SD]

Do you rely on AdWords' cost per conversion numbers?

[AWCthreads]

Yes. Then you'll have to factor in the other costs associated with the sale like shipping, handling, discounts, etc.

There are some unknown variables such as repeat sales that are acquired from the initial conversion and/or volume purchases made after the initial conversion (assuming there are any).

[CMC-SD]

How do you determine which revenue is directly attributable to AdWords?

[AWCthreads]

I'm not sure you can cover every angle or scenario.

We've found our data to be reliable enough to make good decisions.

[john4math]

To be fair in your example, the Adwords click did (on average) have some effect on the ultimate sale.  If the customer never came through the door looking for red widgets, they may not have purchased the blue widgets from you?  There are a few different reports in Analytics that can help you see this:

• Conversions > Multi-channel funnels > Attribution Modeling Tool (I think this is available to everyone now).  If last-click interaction is by far the most important for you, you can select this in the tool and select the primary dimension of Source/Medium.   Drill into Google / cpc and you should see the data you're looking for there.
• Conversions > Multi-channel funnels > Top conversion paths.  Select primary dimension Source/Medium Path.  This will should you the different parts that Adwords played in each conversion.  You can also pick a secondary dimension of Adwords Campaign Path to see more info.