PPCThe Bid Management Recipe

The Bid Management Recipe

If you want to start applying bid management to your AdWords campaigns to optimize for conversions, you're going to need certain ingredients and a framework before you start. This information will give you all the basic starting points.

Bid management is a catch-all term for a group of technologies that will look after certain “automateable” portions of PPC campaigns for you. For very large campaigns this may be the only way to give it the attention it deserves, but even smaller campaigns will generally benefit if the system is set up well.

Bid management systems can be expensive. Off-the-shelf systems are available and some agencies will use their own technology. The information here will give you all the basic starting points you’d need to be able to understand what’s involved and what you’d need if you wanted to set up a system yourself.

The Ingredients

ingredients-bid-management

To make this work you’re going to need:

  • Conversion data in AdWords
  • A database or spreadsheet to store your data and apply formulae
  • An ability to get data out of AdWords, and put new bids in on a regular basis

The first of these comes from conversion tracking. Make sure you’ve got this set up and running well for at least a month before you start. The more conversions you get per day, the better this sort of system will perform.

For most campaigns, Excel will suffice for the second requirement. A good bid management system will have its own dedicated database, storing daily stats over a long timeline. But at the most rudimentary level you’ll be able to do a csv export from AdWords Editor.

The final section will also depend on either AdWords Editor, or using Google’s API. If you use AdWords Editor then you’ll need to be willing to go through this process manually on a regular basis.

The Instructions

At the most basic level, there will be a formula that will give you an optimum cost per click to pay.

Assuming: CPA = Cost / Conversions

Then: CPA = (Clicks x CPC) / (Clicks x Conversion Rate)

Then: CPA = CPC / Conversion Rate

So: CPC = CPA x Conversion Rate

The amount that you can afford to pay per click on a keyword is the conversion rate multiplied by your CPA target. If your CPC is any higher, then you’ll go above an acceptable CPA. If your CPC is any lower, then you’ll miss out on sales that would be within CPA target.

If you run an e-commerce campaign, then you may prefer to target on a Cost of Sale rather than a fixed CPA target. Each keyword will have an average order value associated with it. Choose your Cost of Sale percentage (e.g., 15 percent) and adjust the formula to the below:

CPC = Average Order Value x CoS x Conversion Rate

You’ll want to consider how long a date range you use for the conversion rate statistics. Too long and you lose the effect of changes in conversion rate (e.g., you may have improved your landing pages or it may be a better time of year). Too short and the data quality may be less acceptable.

The Difference Between Average CPC and Bid

Due to the nature of Google’s auction, there may well be a difference between the CPC you pay and the amount that you bid.

Never worry about this.

auction-bid-managementThe nature of an auction (even a quality score influenced one) is that your optimum bidding strategy is always to bid your true value. Consider the scenarios:

  1. You bid your true value, but the competition means you pay less. In this situation you’ve paid lower costs than you intended, so you assume that by increasing the bid you might get more traffic and still pay an acceptable amount. But…
  2. You bid more than your true value expecting the amount you pay to rise accordingly. The amount you pay will only increase if you go above another advertiser that you were previously below. You were below them before with your true bid, so to go above them now will mean paying above your true bid, which means you’re now losing money.

Auctions are very good at making sure this all ends up right, so your bid should equal your target average CPC.

Bid = Average Order Value x CoS x Conversion Rate

Low Conversion Rate Keywords

Not all of your keywords will have enough data to make this work. Choose a boundary level of monthly conversion volume, and apply this formula to keywords above that level. This will help avoid the situation of tripling bid on a low volume keyword just because it converted once, and will avoid setting bids to 0 for keywords with no conversions in a period.

You need to compensate for these keywords. Just because they have a low total conversion volume doesn’t mean they don’t work. So now your formula gets split into two parts: high volume and low volume keywords.

First you need to filter in AdWords for just your low volume keywords (based on this conversion volume threshold – what number you choose is up to you and will depend on the size and volume of your campaign).

Look at your overall CPA (or CoS) for these keywords. They may be better than your high volume terms or they may be worse. Your high volume terms should compensate accordingly.

If your low volume keywords have a worse CPA than your high volume terms, then you need to target a lower CPA on your high volume terms, to give yourself some leeway with spend. If you run your high volume terms exactly on target CPA, then your low volume material could pull you over that limit. So knock the bids down on your high volume stuff until the low volume areas perform correctly. Use a multiplier in your formula to do this. To reduce by 10 percent, use a multiplier of 0.9

Bid = Average Order Value x CoS x Conversion Rate x Multiplier

Handling the Low Volume Keywords

Dealing with these keywords is something you’ll need a completely different style of system for. You can consider position targeting or CPC targeting, but the aim is still to get CPA in total as close as possible to target.

You won’t be able to do so on a per-keyword basis for these, so one option is to use the formula we’re building, but do the calculation across all these keywords, rather than keyword by keyword. This should give you aggregated stats that keep these keywords showing, but get you close to your CPA target.

Predicting Conversion Rate

In some markets conversion rate will be static. In others it won’t. You will need to have some idea how your campaign is likely to behave.

Most retail campaigns can expect conversion rates to rise just before Christmas. Most B2B campaigns can expect better conversion rates (assuming that you can track them!) shortly before a new financial year begins. So your conversion rate over the last month may not necessarily be a solid predictor for your conversion rate over the next month.

Similarly, average order values change. If you go into sale, then the average order value from the last 30 days may be too high compared to what people will likely pay over the next few days.

Predicting average order value isn’t too tricky: you know your schedule of offers and promotions. If you know you’ve got 10 percent off across all products, you can simply reduce the average order value by 10 percent until the figure compensates. Use a factor I’m calling VA (Value Adjuster).

Bid = VA(Average Order Value) x CoS x Conversion Rate x Multiplier

Conversion rate is a trickier beast to handle. Unless you have several years of data it could be really difficult to predict. So now you need a new function, CA (Conversion Adjuster). This function has to do a couple of things:

  • Take last year’s conversion rate over the coming week and index it against that year’s conversion rate over the previous 30 days. So if last year saw that next week’s conversion rate was 20 percent higher than the previous 30 days average, then the CA function should provide you with a 1.2 multiplier.
  • If last year’s stats aren’t available, it should provide the same analysis across similar campaigns in similar markets.
  • If there are no similar stats to use as benchmarks (e.g. if you build this system in house) then you’ll need to use a best estimate. The trend of conversion rate can be very similar to the trend of search volume in some markets. Use Google Insights to get that annual trend (you won’t be able to do this keyword by keyword, so choose some high value representative keywords) and make the trend changes a little more conservative (better safe than sorry when guessing these numbers). You won’t be able to automate this process, so I suggest having a secondary database (or spreadsheet) with a weekly or monthly index.

Bid = VA(Average Order Value) x CoS x CA(Conversion Rate) x Mutliplier

In Use

Getting every major keyword into the highest position that you can afford will work wonders on your campaign. Whether you do this via spreadsheet calculations and upload into AdWords Editor, or with a web-based application integrating with the API, if your campaign has a lot of conversion data then consider implementing a system like this.

If you buy an off-the-shelf system or sign up with a technical agency you’ll get a more sophisticated system than the one I’ve described here. The predictive powers will be better, and it will better deal with low volume keywords.

Techniques like using different timelines for different keywords, or predicting stats for new keywords by using similar ones that already exist in the campaign will all help to improve performance even further. Even if you don’t have access to these, do your best to get these techniques in place.

I’ll leave you with a chart showing the results of these techniques in action. The line represents revenue / spend, and the bid management was implemented at the start of November on a campaign that was already well run manually.

results-bid-management

Resources

The 2023 B2B Superpowers Index

whitepaper | Analytics The 2023 B2B Superpowers Index

8m
Data Analytics in Marketing

whitepaper | Analytics Data Analytics in Marketing

10m
The Third-Party Data Deprecation Playbook

whitepaper | Digital Marketing The Third-Party Data Deprecation Playbook

1y
Utilizing Email To Stop Fraud-eCommerce Client Fraud Case Study

whitepaper | Digital Marketing Utilizing Email To Stop Fraud-eCommerce Client Fraud Case Study

1y