Cold Weather and Cars!
Shortlisted for EConsultancy Most Innovative Agency
Why Your PPC Account May Not Be As Healthy As You Think.
31 Jan 2012
So you sell toasters, and every time somebody makes a purchase, you make £10. Your cost per sale on Adwords is only £6, so you’re making money, you’re happy with your marketing guys and everybody’s happy. But is your PPC really doing as well as you think? Perhaps not…
So you sell toasters, and every time somebody makes a purchase, you make £10. Your cost per sale on Adwords is only £6, so you’re making money, you’re happy with your marketing guys and everybody’s happy. But is your PPC really doing as well as you think? Perhaps not… Just because your overall cost per sale is £6, it doesn’t necessarily follow that every sale you make is profitable. Consider for example, your brand name. Chances are, the cost per click on your name is much lower than for the rest of your account, and there’s a good chance that the conversion rate is higher as well. So what? Consider the impact of splitting this data out: Your performance looks good – you’ve made £5000 (from 500 sales making £10 each) and only paid £3000, so you’re £2000 in profit. But what happens if you split it into brand and non-brand? Interesting… The sales via your brand name have cost £2 each, but your non-brand sales have been costing £10 each. It appears that non-brand conversions aren’t profitable at all! However, it gets worse. Overall, your non-brand sales may be breaking even, but if you’ve ever tried to generate more sales by increasing your budget, you’ll know that this is an over-simplification. Any time that you increase your bids to generate additional sales, your cost per conversion will increase sharply. This is because you will not just be paying more for the incremental clicks that you pick up, you’ll also be paying more for the clicks that you were already getting. Conversely, reducing your bids will reduce your CPA disproportionately. Consider the ‘Other’ campaign above. What would happen if you were to reduce the cost per click to £0.20 (from £0.25)? It’s impossible to tell for certain (since you don’t know whether you’ll drop one place or three, or what impact this will have on your click through rate), but for the sake of argument, we’ll say that clicks also fall by 20%, and that conversion rate doesn’t change. Your CPA has also dropped by 20% (obviously, since your conversion rate hasn’t changed, and your CPC has dropped by 20%), but if you look at the figures more closely, there’s a bit more to it than that. By paying £0.25 per click instead of £0.20, you’ve increased your click spend by £900. But if you look at the sales volumes, you’ve only got 50 more. The additional sales have cost you £18 each – and they are only worth £10! This is the basis of sweet-spot analysis – one way to identify the most profitable position is to estimate the cost per click at which the incremental cost per sale becomes unprofitable. In fact, extending this analysis further, you’d probably want to reduce your bids to just 12.5p – halving them: By reducing your bids by half, you’re now making £625 in profit on this campaign (125 x £5), but even then, you’d be missing a trick. The performance of the different Ad Groups within the campaign may be completely different, so in the same way that your overall figures masked the fact that brand was propping up non-brand, some Ad Groups may be masking the underperformance of other Ad Groups. Continuing the example from above, suppose that having reduced the cost per click to 12.5p as above, you see that your two Ad Groups are performing as follows: Neither of these bids is optimal – increasing bids on Ad Group 1, and reducing them on Ad Group 2 will result in higher profits on both: Theoretically, you could extend this analysis even further, down to keywords, though I would argue that if the performance of keywords in the same Ad Group varies, why are they in the same Ad Group (surely the fact that their conversion rates are different should indicate that the people searching are different, and may respond to different adverts). There are a number of flaws with this type of analysis. Firstly, we haven’t taken into account the average order value. If Ad Group 2 had a much higher AOV than Ad Group 1, then you’d want to pay more per conversion. If this data is available, then you should be using it to calculate the profit for different levels of bids, rather than the overall figure. Secondly, there’s no way to predict accurately the impact on clicks of increasing or reducing your bids. Since this is the case, I would advise against making dramatic changes to your bids – changing them slowly over a period of time will allow you to assess the impact on clicks with more confidence, and reduce the risks inherent in basing your bids on estimates. Another concern is the conversion rate. Most of the time, the conversion rate isn’t affected that much by the position in the search results (something Hal Varian at Google acknowledged in 2009), but there are cases when the conversion rate can be affected significantly. In these cases, the effect of bid adjustments cannot be estimated with the same degree of confidence, and you should rely on empirical data far more. However, perhaps the biggest assumption in this analysis was that your brand conversions weren’t caused by non-brand searches initially. People searching for you by name must have been led to your site by something, and whilst some may be via word-of-mouth or other advertising, much of it may be people that have visited your site previously via a non-brand search, and are returning to make a purchase. Clearly, these sales are at least partially (almost entirely) due to the non-brand clicks, and excluding their value from non-brand understates the true value of these clicks. On this last point, the solution is multi-touchpoint conversion attribution. By assigning part of the value of the sale to the first visit, and part to each subsequent visit, you’ll get a much truer idea of how your Adwords campaigns are performing. The two points to take away from this blog are these:
- Don’t rely on the overall performance of your account. Look at the performance of the individual Ad Groups and campaigns (particularly non-brand campaigns or Ad Groups). If you aren’t being given this information by your marketing team or agency, request it.
- Don’t be distracted by the total cost per sale – consider how much the incremental sales have been costing. Could you make more money with a higher or lower bid?
- Use conversion attribution to understand the true value of your PPC campaigns (and the rest of your marketing), and attribute your sales appropriately. This data is now available in Google Analytics, so your marketing team or agency should be able to analyse it for you easily.