PPC Bid Management in a nutshell

One thing that I have never blogged about here is the bid management part of pay per click marketing. I just did a search in Google to see what others have written about PPC bid management and the SERPs are dominated with details and reviews of PPC bid management tools.

I found this surprising as I believe that for the vast majority of companies there is absolutely no point investing in a PPC bid management tool. Why? Because there will be so few keywords in their account that have high enough search volume and conversion data to allow a bid management tool to run with any degree of accuracy.

This post will be taking you through manual PPC bid management. Why should you read this? Because if you are clear on how to manage your bids manually then you will be able to get 90% as far as a bid management tool will and perhaps just as far in the case of most small businesses but without the costs to your business. Understanding PPC bid management will also put you in a stronger position for speaking to any PPC agencies that are touting bid management tools as one of their selling points.

Understanding your business goals through PPC

The first thing that you need to get clear is what constitutes success for your PPC account – this is a line in the sand that is your break even point; go over and you are losing money.

The best way to define this line is to understand the lifetime value of each of your customers/members/supporters. This is the profit that you will realise from each customer on average over their lifetime as one of your customers.

Here is a formula to use that will help you calculate the customer lifetime value of your business.

The cost of sales relates to your margin – here we have put in a 70% cost of sales for the first year and then assumed some economies of scale have been achieved down the line so profit margins increase to 35%.

Marketing costs decline over time for the same customer, as they become more loyal it gets easier to keep them.

The discount rate relates to interest rates as the value of future profits will not be as much in terms of today’s values. We have set interest rates incredibly high here so that we have an absolute worst case scenario, at least if you are living in the developed world (not that anyone is safe anymore from macroeconomic forces).

This figure will keep rising as the years go by but if we assume that the lifetime of a customer ends after 3 years then we know that any additional marketing we do that brings in a new customer for less than £51.94 is profitable in the long run.

Once you have this figure clear then you are pretty much set up for all of your marketing decision making. The difficulty with applying this figure to Google Adwords is that Google Adwords conversion tracking does not have the ability to split out new and existing customers. With Adwords you will often find repeat customers coming through your branded keywords and also a range of generic keywords too. Therefore there is an additional exercise for you to do and that is to ensure that you are tracking the source of all of your orders in your database.

This could be achieved through using custom variables in Google Analytics. This allows you to track customer specific information e.g. whether they are a new or returning customer and also if you have customer groups then you could split the data based on these too.

Alternatively you could put tracking tags on all your PPC destination URL’s that tie the traffic back to the ppc channel e.g. www.mywebsite.com?aff=adwords tagged onto the end of all of the destination URL’s in your PPC account will mean that the users all arrive with that tag so you know that they came from Google Adwords. If you can set up your database to capture this tag on all sales then you will be able to count the number of sales that relate to new and existing customers in your database. But to tie them all back to the individual keywords you will need to tag all KEYWORDS (not ads) destination URL’s with a unique ID e.g. www.mywebsite.com?aff-adwords123456 or www.mywebsite.com?aff-adwords123457 etc.

Then you will be able to run a keyword report in Adwords to get your costs and then match this to your sales data from your database to see which keywords are bringing you your new business customers and which are bringing returning customers and which are doing diddly squat. Matching this data up in Excel is pretty easy as you can match on the individual tags using a vlook up function.

Now I am going to take you though how to create a PPC bidding model based on your customer lifetime value that will help you achieve constant growth in new customer sales for your business.

1) Set a target cost per new business customer

Your customer lifetime value is what you can afford to spend up to, assuming you have the short term cash flow , to bring in each new business customer. Therefore at an aggregate level if you are pulling in new business customers from Google Adwords at an average cost of £50 (from example above) then things are all OK. However, with almost all Adwords accounts you will find that not all keywords bring you conversions; some will convert every month consistently whereas others will bring you a sale one month and not the next, some will just fail. On the other hand you will also have keywords that will keep bringing you new customers at below the customer lifetime value so therefore giving you more budget to spend on other keywords.

You need to factor in for fluctuations as well as budget for testing new keywords and ad groups.

There is going to need to be some tweaking as you go but if you start off by setting the target cost per new business customer (CPNBC) at about 80% of your customer lifetime value then analyse after 1 month what the aggregate cost per new business customer has been and if it is over your customer lifetime value then you need to decrease this target CPNBC at a keyword level.

Decide on some rules around how you are going to treat keywords based on their performance

You have probably gathered by now that the basic premise is that if a keyword is good then you can invest more in it (bid up), if it is bad then you reduce spend on it (bid down)or kill it completely (pause/delete). But to what extent?

You can use the cost per new business customer at a keyword level to calculate this.

In the example below I have assigned a keyword level target cost per new business customer of £20. I have then simply split the keywords into 3 groups based on their performance:

• Winners = keywords that have a cost per new business customer that is less than the target.
• Losers = keywords that have had new business orders, but the cost for gaining each new business customer is greater than the target.
• Big losers = keywords that have not had a new business sale in the period.

One thing that you will notice here is that I have been super strict based on new business orders; even when a keyword has brought in returning customer sales I have still bid it down. In your own model this is something that you should think about as keyword 2 has brought in 7 sales (the highest of all) yet we are bidding it down because it did not perform in terms of bringing in new customers, but retention is also important. In reality, in most businesses I would also reward keywords that bring lots of returning customers back.

In the final column you will notice that I have assigned a new bid to each of the keywords. I did this by creating a rule for each of the 3 groups; the rules were as follows:

Winners

New bid = (1+(avg pos-1)/10) x Max CPC

For example with keyword 10 the formula would be:

(1+(3.42-1/10) x £0.47 = £0.58 so we increase the bid from £0.47 to £0.58

The theory is this:

• I want to move the keyword up into position 1
• I therefore need to find the variance of it’s current position from position 1. (avg pos – 1)
• I am estimating that a bid change of 10% will move the keyword up 1 position – this is why we divide by 10
• With keyword 10 we are 2.42 positions from position 1 so need to bid up 24.2%

Losers

New bid = (1-(5-avg pos)/10) x Max CPC

For example with keyword 1 the formula would be:

(1-(5-2.78/10) x £0.62 = £0.48 so we decrease the bid from £0.62 to £0.48

The theory is this:

• I want to move the keyword down into position 5
• I therefore need to find the variance of it’s current position from position 5. (5-avg pos)
• I am estimating that a bid change of 10% will move the keyword down 1 position – this is why we divide by 10
• With keyword 1 we are 2.22 positions from position 5 so need to bid down 22.2%

Big Losers

New bid = (1-(8-avg pos)/10) x Max CPC

For example with keyword 13 the formula would be:

(1-(8-3.02/10) x £2.99 = £1.50 so we decrease the bid from £2.99 to £1.50

The theory is this:

• I want to move the keyword down into position 8
• I therefore need to find the variance of it’s current position from position 8. (8-avg pos)
• I am estimating that a bid change of 10% will move the keyword down 1 position – this is why we divide by 10
• With keyword 13 we are 4.98 positions from position 8 so need to bid down 49.8%

There you go; now you have your own bid management tool and it hasn’t cost you £5000/month. Yes, it’s simple but the key thing is that you can modify and make it as complex and as specific to your business as is necessary – it will not be a one size fits all but completely bespoke to you.

My main problem with PPC bid management tools is that for them to automate bidding based on rules they need a large data set and most keywords just will not provide this; they are low volume with just a few clicks per week (assuming you haven’t just gone and set up an account with a few catch all broad match keywords – if so then good luck!).

The above is just an example of one possible way to treat keywords – you may want to create more groups based on more complex rules or you may opt for just bid up / bid down / pause. It really depends on what you are trying to achieve – if you are cash rich and just want market share then you will be less strict whereas if you are a start up on a low budget then you will want to tread more carefully.

Keywords that do nothing consistently

You need to have a method of keeping an eye on keywords that bring you no orders over a long period of time or rather over many clicks (regardless of the time frame). If you run reports regularly then there will be lots of keywords that fall into the no orders bracket so you need an idea of what to do with them; do not let them fall under the radar.

Set a rule that any keywords that reach a certain number of clicks and do not achieve any sales at all get killed – it is a good idea to double check the assisted conversion report in Google Analytics before pausing just in case the keyword is bring you assisted orders down the line through other channels.

The number of clicks that you set your rule at will spend on your website conversion rate. For example if your average conversion rate is 5% then you can expect a conversion for every 20 clicks, so if you have a keyword that has had 80 clicks and no orders then pause it. always leave in a bit of fat – e.g. if the average conversion rate is 5% then let a keyword go to 30 or 40 clicks before pausing just so that you can be satisfied it had a good run.

Making decisions on keywords before you have enough data

This is also something to watch out for; for the majority of your keywords you will probably have no conversions and only a few clicks so what do you do? Until they prove themselves as either good or bad I think that you need to give them a chance – assuming that they do in theory have a close relevance to your products/services. You can do this by running them in positions 5-8 where they will get traffic but at a lower CPC and then once  the performance of the keyword becomes clearer over time you can react accordingly.

Having a bidding model is essential when you are spending £1000 per month and up on PPC as you need to know that you are investing wisely. But you should also be running regular ppc ad reviews and improving your PPC ads. You should also be fine tuning your keyword lists through using search query reports.

If you have any questions after reading this then please feel free to email me if you want to speak privately or leave a comment below if you want to speak more generally.