From experience, the most common performance metric is revenue. The agency earns a percentage on each sale they generate from the paid search campaign. So for example, the agency generates £10,000 in revenue during a calendar month.

They are paid 15% on that revenue, making the fee £1,500.

As an aside, I have heard of situations where the agency doesn’t just buy the media but actually pays for it.  Effectively, they work as an affiliate, ‘owning’ the account with little client input. I’ve never heard of an instance where this model has proved successful (although I am happy to be corrected!)

Therefore, for the purposes of this article, let’s assume the client is paying for their own media.

Is this model right for me? Start by working the numbers…

If you hire an agency to manage your paid search campaign, it goes without saying that you will be paying for their time in some form (hey, we have to make a living!)

With percentage of spend and fixed fee models, the math is pretty straightforward. However, with a revenue based performance model, there is a little more to think about.

The direct link between the revenue generated from the campaign and the cheque you hand over to the agency at the end of the month makes working out how much can you afford to ‘give away’ to generate a sale absolutely essential.

With this in mind, the metrics to hone in on are:

  • Sales volume (the average number of transactions on a monthly basis).
  • Average order value (AOV).
  • Margin.
  • Agency effort (the amount of time actually required to manage the campaign).

A performance based model may not be suitable if:

  • Your transaction volumes are very high.
  • Your AOV is low.
  • Your margin is small.

Let’s use an example to illustrate the point:

Your AOV is £20. The agency drive 10,000 sales in a month, of which you pay 15% in commission on the total revenue (£200,000). The monthly ‘fee’ is therefore £30,000 (nice work if you can get it by the way!)

But, what if your margin is only 20% to begin with? If so, you’ve just given a pretty large chunk of that away.

Another point to consider is the amount of agency effort required to drive those sales. Let’s say the campaign requires 50 hours per month management time. At £100 per hour, you might be looking at around £5,000 per month working to a flat fee model.

Clearly, your profitability will be negatively impacted to a much greater degree by ‘giving away’ £30,000 per month compared to paying for the time actually invested by the agency.

This is both purposely simplistic and extreme but illustrates a key point: you must know your numbers in the first instance (I’m amazed at how many businesses don’t) and then work them… hard.

The agency should be part of this process. Not only does this help assess the viability of the entire model, but will also allow you to negotiate a fair rate of commission should both parties agree pay-on-performance is the most appropriate way forward.

Other things to look out for

Some of the considerations raised in my last post are also relevant to pay on performance models (see the comments around set up fees, ‘set it and forget it’ and brand term bidding). However there are also some other considerations specific to pay on performance that come to mind, as follows:


In almost any sector, but especially retail, consumers use multiple channels to research, compare and ultimately make their purchase.

Therefore, for a pay on performance model to be fair to both parties, the following needs to be in place:

  1. An agreed attribution model. For example, if somebody visits your site initially via an organic listing but converts on a paid search ad, who gets the sale? In turn, do you pay the agency commission on this? Or do you pay part-commission?
  2. Appropriate tools and processes to track the above… with as much accuracy as possible.

The attribution model and commercial arrangements need to be set in stone from the outset. This ensures that the role PPC has played in generating sales is fairly rewarded (or, on the other hand, not rewarded).

Driving traffic is just the start

Buying traffic via Adwords or Bing is very easy to do, almost too easy in fact (I refer to the actual act of creating an account and handing over your credit card details, not the significant amount of skill required to actually drive a return).

However, turning that traffic into revenue is an altogether different challenge. 

Whilst it should always be in both parties interest to ensure as much of the traffic driven from the paid search program converts, the interest is understandably heightened from the agency’s side with a pay on performance model employed.

Therefore, there needs to be a clear focus on the entire customer journey and a clear commitment to conversion optimisation.

If this isn’t clearly outlined and agreed from the outset it can inevitably lead to frustration for both sides later on. A poor user-experience and low conversion rate undermine the paid search campaign, no matter how well thought out it is. 

Ownership of the account

The issue of who ‘owns’ the paid search account(s), client or agency, isn’t restricted to pay on performance models. However, it is something that I come across more frequently when a pay on performance model is employed.

Recently, we took on a client where their previous agency had complete control over the Adwords account in terms of strategy, structure, keywords, messaging and so on. The client had virtually no input and certainly no access to the account.

At the end of each month, the client was billed for media, plus 20% on any sales generated (needless to say, this was not a great deal due to the non-collaborative approach).

Worse still, when it came to a parting of ways, the agency claimed ownership of the account citing that it was their intellectual property. The client had to start again from scratch.

Take note; just because the agency knows PPC, it does not mean they know your business better than you. Therefore, don’t ‘hand the keys over’ and simply let the agency get on with it. Not only does this run the risk of a poor performing campaign but if things do go wrong you might find valuable Adwords data is lost because you didn’t read the small print.

When does this model work best?

As I cite above, the success of this model boils down to the numbers. Therefore, as a buyer, you need to be willing to share sales volumes, AOV, margin and other key financial metrics with an agency.

If you don’t know them or don’t have enough historical data, it may prove difficult for the agency to make an informed decision on the viability of the model from their side.

An appropriate amount of analysis, scoping, due-diligence and forecasting is therefore required prior to contracts being signed. Jumping in blind to this kind of arrangement is dangerous for both agency and client.

It then comes down to the size of the carrot you are willing to dangle in front of the agency. Assuming a set-up fee is not being charged, you are asking the agency to invest their time and energy to determine the strategy, create the account and then manage it on an ongoing basis.

There needs to be enough of a financial incentive to have the agency go the extra mile. Without it, corners will inevitably be cut. Ultimately, this will come at your expense.

In some cases, this model may be more appropriate a few months into the relationship when there is greater understanding of the business and trust has been developed between agency and client.

Issues such as stock availability also come into play with the model. Out of stock items limit the agency’s opportunity to generate sales. More widely, it could be argued that a general failure to put the customer first will undermine not just PPC (and this model in particular) but all marketing efforts. 

Lastly, flexibility in media spend – nothing will be more frustrating for the agency if market demand exists but there is not the adequate media budget to realise it (and in turn, earn their commission).

Two models down, one to go

Next time, fixed fee comes under scrutiny.

In the meantime, please share your own experiences of pay on performance models. Have you come across any weird and wonderful variations to a revenue based approach?

What worked? What didn’t?

Image courtesy of opensource.com