That said, from speaking to other publishers who use SEM extensively in support of their ePN campaigns, at a high level most associate each SEM keyword they purchase to drive traffic to their site with a Custom ID, then track through the Transaction Download Report the ePN conversions associated with each Custom ID. Tying that back to the cost incurred on the relevant keyword gives you a good idea about the efficiency of each SEM keyword purchased in driving ePN conversions.

One technique which I used constantly when managing SEM campaigns was to plot daily spend against revenue in a scattergraph at whatever grain I was looking to optimise (be it keyword, adgroup or overall portfolio). Through these daily spend versus revenue points I’d plot a line of best fit to get the most accurate view of how much revenue I’d get for a given level of spend (and therefore what my profit would be). In most cases, the ‘power’ trendline in Excel, which gives you a curved line through the points, gave the best representation of the declining marginal profit as you increased spend.

Once you have this curve plotted, you can actually reverse-engineer your way to the point where marginal ROI is equal to zero, and you have effectively maximised profit. To do this, follow these steps:

1. Plot your daily spend vs revenue scattergraph and add the ‘power’ trendline through the points.

2. Go into ‘Format Trendline’ and under Options, select ‘Display Equation on Chart’ and ‘Display R- squared on Chart’. The R-squared value shows you how well the changes in revenue reflect the changes in spend. 1 is the maximum value – the lower the value, the less confident you can be in the line of best fit. The equation on the chart should read ‘y = _x^_’, where the ‘_’ symbol represents a value (for example, ‘y = 29.422x^0.6382′). The first number in the equation is the Coefficient, the second is the Exponent.

3. You can then use the values in the equation to predict how much revenue you will get at a given level of spend. To do this, use the formula ‘Revenue = (Spend ^ Exponent) * Coefficient”. For example, to find out your daily revenue when spending $10,000 per day, using the example above your formula would read ‘Revenue = (10,000 ^ 0.6382) * 29.422′. This would give you a daily revenue of $10,506.84.

4. To take this a step further and answer the question of what spend level maximises profit, make a table with three columns. The first column should list different possible daily spend levels (e.g. $10 through to $10,000, in $10 increments). The second should show the predicted revenue amount at each of these spend leves by replicating the formula above (e.g. ‘= (A2 ^ 0.6382) * 29.422, where cell A2 is the daily spend). The third and final column should show the extra profit yielded by each spend increment (e.g. [cell B3 - cell B2] – [cell A3 - cell A2]).

5. When looking at your table, whichever cell in the third column is closest to 0 gives you the optimal spend level. In the example above, this would be a daily spend of $3,320.

These 5 steps could doubtelss be automated relatively quickly – for example, a quick search of Microsoft Excel help forums should reveal the formula for automatically generating the power curve without needing to plot a scattergraph.

As to your point about a second custom field for the transaction download report – that is definitely something we will take into consideration when planning out future functionality.

Hope this helps,

Chris

Since the conversions are taking place outside of our tracking realm, is there any way to directly link a PPC ad to a conversion on eBay? Or even goals that could be set up in Google Analytics that would shed some light on the productivity of a PPC ad?

What I’m particularly interested in is price elasticity of profit – finding the optimum CPC that yields the most profit.I’m not concerned about budget or margin percentage, just absolute profit.

If I pay more for my clicks I get more traffic and more revenue but my costs are higher – will the incremental revenue (more than) offset the incremental cost or not? That’s the question I’m grappling with daily. Haven’t found a solution as yet, other than trial and error and a certain amount of ‘feel’. Also, as you say it changes over time. Any tips for maximising the absolute profit return?

I would find a second custom field in the transaction download very useful as I write more data to this field than just the keyword so I have to parse the keyword out of this data which is a pain. A second custom field would certainly allow me to analyse better and I suspect many people would find it useful.

Regards,

Tim