Paid Search Bid Concepts and Strategies
Keyword bidding is no easy task. There are several tools and strategies each with their own pros and cons. This is the first in a series of five articles where my colleagues and I will discuss a few bid concepts and bid strategies. We will also address their associated trade-offs to help you choose the best fit for your business.
Before discussing the bid strategies, we will review two key concepts:
1. Key Performance Indicators (KPI) Positional Impact
2. Keyword Level vs. Portfolio Level Analysis
Key Performance Indicators (KPI) Positional Impact
KPIs, such as CPCs, differ across positions. Using Microsoft Advertising Intelligence monetization data, you can see metrics for the average advertiser for the keyword “computers” across the Bing marketplace.
*download Microsoft Advertising Intelligence
Based on the graphs, “computers” tend to have higher volume, bids, and CPCs in better positions, but the slope of each curve differs. In this example, the result is:
· Average position 1 bids ~2.9 times more than position 61
· Average position 1 pays ~2.2 times more than position 6
· Average position 1 gains ~13.5 times more clicks than position 6
If you want to increase volume, positions 1-3 will get you a substantial increase in volume without needing to pay a proportionally equal increase.
Collecting and logging this data via Microsoft Advertising Intelligence and your own testing of keywords can improve your analysis.
Factoring in Conversion Rate2
Knowing the basic search KPIs across positions is usually not enough to make a good bidding decision. We also need to factor in conversion data.
Imagine you bid $1.62 on “computers” with a cost per acquisition (CPA) 3 goal of $25. In this example, your performance report shows:
| Position | Clicks | CPC | Spend | Acquisition | CPA |
| 4 | 208 | $1.15 | $239.17 | 8 | $29.90 |
You are above your CPA goal at $29.90 CPA and need to improve. Using the charts below as guidance, you can see how conversion rates across positions affect your bidding decision. Unlike the basic search KPIs, you cannot retrieve conversion data from our tools. You will need to collect this data through testing because the conversion rates will be different for your set of keywords.
The charts below illustrate how to analyze this hypothetical problem with different conversion rate scenarios.
Scenario A: Flat Conversion Rate
Scenario A uses findings from “An Empirical Study of Search Engine Advertising Effectiveness” which concludes that conversion rates are flat across positions for certain engines. The above charts show that with a flat conversion rate curve, the CPA and CPC curve have the same shape, so CPC acts as a proxy for CPA.
Let’s apply a flat conversion rate assumption to our bid problem. Recall your CPA for “computers” is too high. This data indicates that CPC for “computers” falls with position, so you should bid down to lower your position, which will lower your CPC and CPA.
There is strong evidence that support conversion rate being flat across positions. If you assume a flat conversion rate curve, this can simplify bidding for other cases as a number of keywords’ CPC tend to fall with position, so CPA falls with position for them too.
| Scenario B: Linear Conversion Rate | |
Also, there are articles that suggest you will find a different result for specific groups of keywords. Scenario B is based on a finding from the “The Atlas Rank Report II: How Search Engine Rank Impacts Conversions” article that suggests conversion rate increases with position for low volume keywords. If we apply this assumption to “computers”, the CPA for positions 1-3 is within our $25 CPA goal, and so we should consider bidding up to those positions.
Ultimately, I recommend testing your keywords to understand your landscape. Hopefully this helps illustrate how to analyze your data to make a more informed bidding decision.
Next time we will discuss keyword vs. portfolio level analysis.
Footnotes
1. Calculation: Pos1Bid /Pos6Bid
2. Conversion and Acquisition is used interchangeably
3. CPA = Cost/Acquisitions = CPC/Acquisition Rate
Paid Search Bid Concepts and Strategies (Part II)
posted Fri, Oct 23 2009
Last time we discussed how Key Performance Indicators (KPIs) differ across positions in Microsoft adCenter and how that impacts your bidding decisions. Today we review keyword level analysis, where you optimize individual keywords, and portfolio level analysis, where you optimize the overall portfolio performance.
Tell me about Keyword Level Analysis!
A keyword level bidding strategy examines keywords individually, and makes a decision based on each keyword’s data independent of other keywords. Each keyword must meet your CPA goal.
Example: You bid on two keywords “keyboards” and “monitor” with a CPA (cost-per-acquisition) goal of $21. adCenter reports you are in position 1 for both keywords generating large number of acquisitions, but “keyboards” is above your CPA goal.
Pre Analysis
| Keyword | Position | Clicks | CPC | Spend | Acquisition | CPA |
| Keyboards | 1 | 10,000 | $0.22 | $2,200 | 100 | $22.00 |
| Monitor | 1 | 2,999 | $0.81 | $2,440 | 119 | $20.00 |
In keyword level analysis, “keyboards” is evaluated independent of “monitor.” When you calculate the bid price for “keyboards,” you don’t account for the performance of “monitor.” So you bid down “keyboards,” and leave “monitors” untouched as it already meets your goal.
Post Analysis – position 3 for keyboards works!
| Keyword | Position | Clicks | CPC | Spend | Acquisition | CPA |
| Keyboards | 2 | 8,900 | $0.20 | $1,780 | 83 | $21.45 |
| Keyboards | 3 | 3,738 | $0.18 | $672.84 | 40 | $16.82 |
| Monitor | 1 | 2,999 | $0.81 | $2,440 | 119 | $20.00 |
In the next two blog posts, my colleagues will introduce two keyword level strategies that will show you how to calculate a bid price for a case like “keyboards”:
1. Incremental Bidding Strategy (coming soon)
2. Sales Based Bidding Strategy (coming soon)
Although a keyword bidding strategy can be effective, you may want to consider whether a portfolio bidding strategy makes more sense for you.
Does Portfolio Level Analysis make sense for me?
In portfolio level analysis, you make bid suggestions accounting for other keywords with portfolio level goals. Portfolio analysis ranges from very basic to very complex strategies.
Basic Portfolio Strategy Example
1. Aggregate your data for review
| Keyboards Position | Monitor Position | Portfolio Clicks | Portfolio CPC | Portfolio Spend | Portfolio Acquisition | Portfolio CPA |
| 1 | 1 | 12,999 | $0.36 | $4,640 | 219 | $21.19 |
2. Does the “keyboards” and “monitor” portfolio CPA meet your goal?
3. If yes, then no change is needed
4. If no, then reduce the bid of the higher CPA keyword with a keyword level strategy. Repeat until your Portfolio CPA meets your goal
Advanced Portfolio Strategy Example
*This is a simplified illustration to show the concept of an advanced portfolio level analysis – NOT the algorithm behind tool providers.
1. Collect and store keyword metrics such as conversions, impressions, etc across the marketplace
2. Build a mathematical model to forecast data across all metrics, match types, etc
3. Create a programming model to solve for the most optimal solution based on your goal
Post Analysis – sample of result combinations you can bid for “keyboards” and “monitor” below
| Keyboards Position | Monitor Position | Portfolio Clicks | Portfolio CPC | Portfolio Spend | Portfolio Acquisition | Portfolio CPA |
| 1 | 1 | 12,999 | $0.36 | $4,640 | 219 | $21.19 |
| 1 | 2 | 10,950 | $0.28 | $3,112 | 139 | $22.39 |
| 1 | 3 | 10,101 | $0.23 | $2,276 | 109 | $20.88 |
| 2 | 1 | 11,899 | $0.35 | $4,220 | 202 | $20.89 |
| 2 | 2 | 9,850 | $0.27 | $2,692 | 122 | $22.07 |
| 2 | 3 | 9,001 | $0.21 | $1,856 | 92 | $20.17 |
· Found 3 combinations that is below our CPA goal (highlighted in blue) opposed to 1 from the keyword level strategy
· “Keyboards” in position 2 and “monitor” in position 1 generates 202 acquisitions – 43 more acquisitions than our previous result
· Beware “keyboards” in position 2 has a CPA above our goal ($21.45), so your business model needs to accept a CPA of X% above the CPA goal for each keyword
OK, what’s the difference between Keyword and Portfolio Level Analysis?
Keyword level analysis strategies typically can be implemented by a single individual – usually with a simple formula in Excel or other spreadsheet applications. The cost includes the spreadsheet application and the time it takes to implement the formula.
Advanced portfolio level analysis requires a lot of resources. Imagine looking at 1,000 keywords – that’s 8^1000 position combinations alone! It would be difficult to examine each combination on your own. To simplify the task, there are many companies that offer a tool to do this in the backend – list of bid management tools1, but you may have to pay hundreds of dollars.
Summary Table
| Criteria | Keyword Level Analysis | Advanced Portfolio Analysis |
| # of People Required | 1 | 1 + Tool Provider |
| Complexity | Low | High |
| Combinations to review for 1,000 keywords | 1,000 | 8^1000 or more depending on number of factors modeled |
| CPA | Each keyword must meet CPA goal | Portfolio CPA meets CPA goal. Each keyword CPA can be X% above the CPA goal. |
| Price | Free – given you own a spreadsheet application | Free – hundreds of dollars |
So is keyword or portfolio level analysis better for your business?
· If you are the only marketing manager with limited bandwidth and budget, utilize the keyword level strategies.
· If you have a large budget with complex campaigns and a business model that accepts +-X% CPA goal for each keyword, investigate tool providers.
Part 3
Our previous post discussed keyword level and portfolio level analysis. In this post, we will dive into a keyword level strategy called incremental bidding. Utilizing this strategy is a simple way to optimize each individual keyword and achieve a lower cost-per-acquisition (CPA).
Tell Me about Incremental Bidding
With incremental bidding, you compare each keyword’s CPA to your overall CPA goal. Based on how the keyword performs relative to the overall CPA goal, you can adjust the bid, which affects the position. The objective is to test different positions until you find the keyword’s optimal position. Below is an illustration of how to analyze performance and adjust accordingly:
Give Me a Real Life Example
We are running a search campaign selling TVs with a $50 CPA goal. We selected $50 as the CPA goal since it is close to our profit margin. Let us look at how we would adjust the keyword bids based on the performance using the incremental bidding strategy:
| Keyword |
Bid |
Average Position |
Cost |
Actions |
CPA |
| Flat screen |
$3 |
1.2 |
$500 |
20 |
$25.00 |
| LCD |
$5 |
5.0 |
$600 |
25 |
$24.00 |
| Plasma |
$4 |
3.0 |
$380 |
8 |
$47.50 |
| LED |
$3 |
2.0 |
$800 |
14 |
$57.14 |
| Tube |
$2 |
2.0 |
$1,000 |
4 |
$250.00 |
Keyword CPA < Goal CPA
- Flat screen is performing well at half the goal. The keyword is close to the top position and increasing the bid will have little impact. As a result, we should keep the existing bid.
- LCD is also performing well, but is currently in position 5. With a higher bid, this keyword can rank higher on the page, drive more clicks, and possibly bring in additional sales. We should try increasing the bid by 10% – 30% to see how it affects performance. This range is just a suggestion. You can experiment with the percentage increase depending on how aggressive you are.
Keyword CPA close to Goal CPA (+/- 5%)
-
- Plasma is at 95% of the CPA goal. Since it is performing well with its current bid and position, we should leave the bid as is. If we increase the bid to improve the position, the keyword will most likely exceed the CPA goal.
Keyword CPA > Goal CPA
-
- LED is at 114% of the CPA goal. By lowering the bid by 10% – 30%, this keyword CPA may fall below the CPA goal.
- Tube is performing at 5 times the CPA goal and needs immediate attention. We can look at adding negative keywords, improving landing page relevancy, turning off phrase or broad match, or drastically reducing the bid. If none of these work, we should consider pausing or removing the keyword altogether.
I like the strategy! How do I implement?
Implementing this strategy is simple, which means you can start optimizing your keywords immediately!
To get started, import your keyword performance data from Microsoft adCenter into Microsoft Excel. If you are tracking actions or conversions through adCenter, you should include these in your report. If not, you will need to match your conversions with your keywords separately.
Once your data is in Excel, you can analyze your keywords using Excel filters. You will need to add two columns, one titled “CPA” and another titled “New Bid”. In the CPA column, divide the keyword cost by the number of keyword conversions or actions. You can now filter by following the chart above. First, filter by the CPA and then filter based on the position. Once you finish filtering a group, take your existing max bid and either increase or decrease it based on how the keyword performed. Enter this value in the New Bid column.
After determining your new bids, implement the changes and check back later to see how performance changes.
Pros and Cons
There are pros and cons to the incremental bidding strategy. The advantage is that you can optimize your keywords to achieve a lower CPA. The disadvantage is that it can take weeks to months before you reach close to an optimal position for your keywords.
Due to the constant change in the marketplace, you will need to continue using this strategy to make sure your campaign performs efficiently. The frequency depends on how many impressions and clicks your keywords receive. If you have several high volume keywords, you will have enough data to run through the analysis every 1-2 weeks, while if you have low volume keywords, every 3-4 weeks is sufficient.
Part 4
In Part 1 of this series, we cited “An Empirical Study of Search Engine Advertising Effectiveness” which concludes that conversion rates are flat across positions for certain engines. Taking this study into account, you should not determine bids by the position they will appear on the page, but rather the amount of money you are able to spend per click and remain profitable.
Every keyword in your account will drive users to a different shopping experience and therefore, drive a different sale amount. Managing your keywords to the sales amount will help your company spend your search dollars more efficiently and drive more profitable orders.
One example of this bidding strategy would be:
Sales Per Click*Advertising to Sales = Bid [SPC*A/S=Bid]
In the table above we’ve run the keywords through this formula, putting 75% of sales back into advertising (A/S). You should work with your financial department to figure out what A/S percentage will be most profitable for your account. Sales per Click is calculated by taking revenue divided by clicks.
These new bids will push well performing keywords into higher positions, giving them more visibility and giving you the opportunity to drive more orders. At the same time, keywords that are spending too much will be lowered to a more cost-effective level.
Based on the data you receive in your account, you should build reports for three different time periods (typically: 7, 14, and 30 days) to analyze keywords. By looking at different time periods you will be able to better account for seasonality and recent trends. This will also prevent a quality keyword that has had a bad week from being pushed down for too long.
One downside to this style of bidding is it does not take sales margins into account. If you are able to pull in margin-level data per keyword, you can take this strategy one step further and adjust your bids based on the profitability of each keyword.
Part 5
Keyword bidding can seem like an overwhelming task: How to bid, what to bid, and when to bid – there are many variables at play. In this series (I, II, III, and IV) we’ve outlined how to start looking at your keyword performance and use key performance indicators (KPIs) to adjust your bids:
1. Using Microsoft Advertising Intelligence monetization data, you can see the average advertiser data across keywords in the Bing marketplace. Seeing the variance in KPIs can help you plan your bids for high volume keywords.
2. Keyword-level analysis can typically be done by a single person, whereas advanced portfolio-level analysis can require a lot of resources.
3. When making bid changes, keep your goals in mind: Are you trying to maintain a Cost Per Acquisition (CPA), Return on Ad Spend (ROAS), or another goal?
4. By using all of the data available to you, you can adjust your bids based on performance – making sure you are getting the most return on investment for your search advertising campaign.
Now we’d like to open up the topic of bidding concepts and strategies to discussion:
- How are you optimizing your bids?
- In the past when I was working for a retail company, we optimized bids based on a CPO goal. Later, the company moved to margin based bidding.
- How are you accounting for seasonality?
- Before peak seasons I like to take a look-back at what keywords performed best the prior year. Where are those keywords today? Do I need to increase bids on them?
- What do you with keywords that have no conversion data?
- A question that plagues many of us in search engine marketing. You don’t want to kill a keyword too soon, but you also don’t want to lose money letting it run.
Also, please let us know if there is a specific topic or strategy you’d like us to delve deeper on and write about!
Source: http://community.microsoftadvertising.com/blogs/advertiser/archive/2009/12/18/paid-search-bid-concepts-and-strategies-part-v.aspx