While the Quality Score indicates whether your keywords/ads/landing pages are relevant against the competition, it does not tell much about the actual competitiveness of the market. Instead, the average cost-per-click (CPC) can help measure how competitive the auctions are, given that more competition typically causes higher CPCs. That being said, your CPC could be high because of a poor Quality Score, not necessarily because of the fierce competition out there.

In order to isolate the competitiveness component from the equation, one can measure the average delta between the max bid (what you’re willing to pay at the max) and the CPC (what search engines actually charge you). How can this delta be determined, and how actionable is it?

Calculating The Competitiveness In Your Paid Search Account

You can look at it as a ratio; and in theory, it should range from 0% if there was no click to 100% if the CPC was equal to your max bid. Obviously, the grey zone in between is what we are interested in.

Also, for the sake of this analysis, this ratio can be aggregated from the keyword to the ad group, campaign, and even AdWords label level (or any way that makes sense to your business). The aggregation is not pretty as you’ll be summing CPCs and bids; however, from my experience, it does work — especially when weighing by the number of clicks.

What should one expect to see then? Well, say you have labeled your campaigns in AdWords by trademark (your own branded terms), brands (brands that you distribute), generic (short, high-volume, non-trademark, non-branded keywords), and SKU (product-specific keywords). Then, based on multiple paid search accounts organized that way, one can expect:

• Low competition on your trademark terms: Your CPCs should be low compared to your max bids. In the below example, our competitiveness ratio shows 27%. If your max bid is $1, you’ll be charged $0.27 on average.
• Medium to high competitiveness on products-specific keywords: 53% based on this client set.
• High competitiveness on generic and brand related keywords: 66-67% in our example.

Obviously, you can visualize this ratio at the campaign, ad group, and keyword levels. too. For instance, you can look at it for your top 10 keywords from a cost standpoint – where the two first keywords on the left are obviously trademark terms:

Application #1: Bidding Strategy

Being able to determine the average delta between the CPC and the max bid is not just about putting together fancy charts — it can also be pretty powerful and actionable!

Say you’re launching new paid search campaigns, and you’re not sure about the initial bids. All you know for sure is your CPA target and your historical conversion rate for a given product category. Then, given that CPC= CPA*Conversion Rate, you can calculate your initial max bids such as:

Example:

Application #2: Identify Growth Opportunities

To simplify, a straightforward way to identify the room for growth in your paid search account is to spot those profitable keywords not already in rank 1.0 or so. You can then bid up and gain incremental conversion volume as long as profitability remains on target.

A slightly more sophisticated approach to this tactic is to also take the Quality Score into account, since you want to first get more aggressive where your keywords/ads/landing pages are the most relevant.

Similarly, search marketers can also leverage the above competitiveness indicator so that they first get the most out of those keywords where competitiveness is relatively low.

Example: if your target CPA is $40 and the average competitiveness in your account is 43%, then you might want to first bid up on those keywords with a position worse than 1.5, a CPA lower than $35, a Quality Score greater than 5, and a competitiveness lower than 40%. This logic can easily be translated into Excel using pivot table/filters and will help you get the most out of those profitable keywords where the competitiveness is below average.

Conclusions

In conjunction with a Quality Score analysis, analyzing the average delta between the CPCs and the bids can be key to determining the competitiveness in your account, adjusting your bids accordingly, and identifying opportunity keywords you want to bid up first.

While hourly bid multipliers aren’t new, they remain a crucial tactic for optimizing your AdWords campaigns. They work by reducing your ad spend at poor-performing times of the week and increasing your exposure at the best times of the week. Here, I’m going to share the steps you can take (along with a helpful spreadsheet) to determine your hourly bid multipliers for better campaign optimization.

Step 1: Pulling An Hourly Performance Report From AdWords

On the Campaigns tab in AdWords, go to Columns>Customize Columns and ensure that you’ve selected the appropriate metrics. Performance metrics required for the spreadsheet to function properly are as follows: Campaign, Clicks, Impressions, Cost, Avg Pos, and Conv (1-per-click) — all other metrics selected in the screenshot below are optional:

Once your performance metrics have been selected, hit the “Download Report” button. When prompted, add the “Day of the week” and “Hour of day” segments:

This should provide you with all the data you need to analyze hourly performance at the campaign level.

Step 2: Determining Hourly Bid Multipliers

Similar to the template used to determine mobile and geo bid multipliers, I’ve created a basic spreadsheet to help analyze hourly performance and easily determine your hourly bid multipliers. You can download it here.

Copy and paste your AdWords report into this spreadsheet as directed. From here, you can take a closer look at the following:

a. Performance By Day Of Week

b. Performance By Hour

c. Performance By Hour & Day Of Week

If you have collected enough hourly data for each day of the week, you should absolutely make bid adjustments on an hourly basis. This process can be time consuming, as it requires making adjustments on a very granular level, but the results are well worth it.

For those times with less traffic, you can still leverage daily and/or hourly trends. For instance, looking at campaign #43 in the attached spreadsheet, it appears that there was not enough data collected on Sundays from 4:00 am to 5:00 am to make a specific bid multiplier suggestion — but you might still want to increase the bids, since the data indicate that both Sundays and the 4:00 am to 5:00 am window perform well in general.

The attached spreadsheet will only address those times of the week with sufficient hourly data, while keeping in mind that “bid adjustments for locations, days, times, and any ad group-level targeting methods can be set from -90% to +900%.” Thus, it can help you to determine relevant hourly bid multipliers between -90% and +900% when there are a statistically significant number of clicks:

Step 3: Implement Hourly Bid Multipliers In AdWords

At the campaign level, navigate to the “Settings” tab; then, go to the “Ad schedule” section. The first step is to specify when you want to make bid changes. Select a day of the week from the drop-down menu:

From there, you can you can adjust the effective hourly bid multipliers, as calculated by the spreadsheet:

Conclusion

All of this is fairly straight-forward; however, your hourly bid multipliers need to be maintained over time, hence the importance of a (semi-)automated process. Also, keep in mind that once set in AdWords, those hourly bid changes do not take into account multiple time zones. For instance, if your AdWords account is set to “(GMT-08:00) Pacific Time,” and you want to increase the bids by 20% at 1 pm, then these bid changes will occur at 1 pm PST across all PST/MST/CST/EST locations. As a result, it makes sense to break down your top campaigns by time zone in order to set more accurate hourly bids.

Anyway, this post is about helping search marketers determine those mobile and geo bid multipliers based on historical data.

1. Pulling The Data From AdWords

In AdWords, first go to the ‘Dimensions’ tab, then select ‘View: Geographic’ and update the column set as follows :

Depending on the size of your account, you might want to use a filter in order to limit the amount of data you want to look at – using a filter such as ‘Impressions>=100’ will help do that:

When downloading the report, use the ‘Device’ segment:

Now, we’ve got all the data we need to analyze device and top locations’ performance.

2. Determining Mobile Bid Multipliers

I have shared an Excel spreadsheet in order to help go through this process. You can download it here: Device and Geo Bid Multipliers.

Provided that mobile bid multipliers can only be set at the campaign-level and are relative to Desktop/Tablets bids, one should first aggregate Tablet and Desktop together, then compare Mobile against it.

One simple way to go about it is to use the formula below  based on the ‘Device’ field: =IF(LEFT(B3,6)=”Mobile,” “Mobile,” “Desktop/Tablet”)

Now, you can look into your historical performance by Desktop&Tablet vs. Mobile, which will help you determine your mobile bid multipliers for each individual campaign:

What is your Mobile CPA target? You can relate it to the Desktop/Tablet CPA target, such as: Mobile CPA target = X *Desktop&Tablet CPA, where X is a multiplier which reflects the role of mobile impressions for your business.

For instance, you might be able to determine that you are ok with your mobile CPA being twice as great as your Desktop&Tablet CPA since those mobile impressions are more about brand awareness, not so much about immediate conversions.

Then, for a given campaign, say your Desktop&Tablet CPA is $29.46 vs. $118.28 on mobile, and you are ok with the mobile CPA being twice as great as on Desktop&Tablet, then your mobile bid multiplier can be defined as Mobile Bid Multiplier = (Mobile CPA Target * 100 / Historical Mobile CPA )– 1, such as 2 * $29.46 * 100 / $118.28 – 1 = – 50%.

As a result, your mobile bid multiplier should be negative 50% for this particular campaign to achieve your mobile goal.

The attached spreadsheet should help apply the same logic to all campaigns with historical data across all devices. Note that you all want to make sure you’ve got enough data to determine those mobile multipliers, hence the ‘Click threshold’ field in the attached spreadsheet.

Note that you should rename your mobile-only and tablet-only campaigns just like your primary desktop campaigns for this tool to work in case you had broken down your campaigns by device.

3. Determining Geo Bid Multipliers

Using the same report and same spreadsheet, you should be able to easily visualize your historical performance by geo – for instance, by country to keep it simple – the ‘Most specific location’ field is way more granular:

Then, the same Excel spreadsheet can help suggest some geo bid multipliers for those locations with a statistically significant amount of clicks, assuming you want each geo’s CPA to converge toward your campaign-level CPA – which basically means you want to invest more where over-efficient, and cut the spend where under-efficient.

Your geo bid multipliers can be determined such as Geo Bid Multiplier = (Avg. Campaign CPA / Geo CPA) – 1.

For instance, if your campaign-level CPA is $18.40, while your New York CPA is $13.27, you might want to invest more in New York and set your New York bid multiplier to ($18.40 / $13.27) – 1 = 39%. And again, you want to make sure each location drove enough traffic to be able to draw any kind of conclusions; hence, the ‘Click threshold’ field in the attached spreadsheet.

In a nutshell, Enhanced Campaigns require some in-depth analysis by device and geo — hopefully, this post will help you get started, or at least put you on the right track if you were not sure how to go about it.

Obviously, those device and geo bid multipliers will need to be updated on a regular basis as consumer behavior evolves over time.

Now, we can address those paid search campaigns capped due to insufficient rank – and there are going to be more assumptions involved since all metrics are going to be impacted: the rank obviously, then the CTR, CPC, and pretty much everything else as a result.

This article goes beyond impression share (IS) analysis and is actually mostly about regression analysis. Given the number of assumptions involved, the numbers we’ll come up will be closer to directional estimates rather than actual predictions; although in some cases, this methodology turns out to be very accurate.

How To Mitigate Lost IS (Rank)?

The Lost IS (rank) metric is an interesting, however, vague metric – officially it is “the percentage of time that your ads weren’t shown due to poor Ad Rank.” Hence, a couple of initial comments:

• A “poor” Ad Rank is usually due to two main factors: a low Quality Score or a low bid. AdWords does not provide the details, while Bing Ads does, thanks to a couple of additional useful metrics: “Impression share lost to bid,” and “Impression share lost to keyword relevance.”

• A “poor” Ad Rank actually means your average rank is not 1.0 across the board. That’s where it gets tricky since rank 1.0 is just not achievable across the board, especially for phrase and broad generic terms, even with insanely high Quality Scores and bids since Google keeps testing and rotating different ad copies all the time and tends to reward advertisers using exact match type.

• If rank 1.0 was achievable in phrase and broad match types, the traffic quality (understand conversion rate) would most likely decrease because of even broader queries being associated with your keyword list when bidding is up.

The bottom line is that advertisers who want to scale up their PPC program should first focus on ad relevancy, then aggressive bid management, while keeping in mind that rank 1.0 is just not achievable in most cases. For the sake of this article, I’ll focus on the impact of a more aggressive bidding strategy – i.e., I’ll keep relevancy optimizations aside.

Estimating Incremental Available Impressions

Similar to the way one can estimate overall available impressions based off Impression Share (IS) data, one can estimate missed impressions due to rank based off Lost IS rank data. More specifically: Missed Impressions(rank)=Impressions*Lost IS(rank)/IS, hence the below sample table.

Those are all missed impressions because you were not in rank 1.0, which in not achievable, anyway, but at least, we’ve got some hard limits that we can use for reference.

Analyzing The Relationship Between Rank, Impressions, CTR & CPC

From my experience, the best way to go about it is to first explore multiple scenarios so you get a good understanding of the relationship between rank, impressions, CTR, CPC, ad spend and revenue.

You basically want to collect enough data to measure the elasticity between those metrics and answer the following question: seasonality aside, what would happen if I spent $X more? This is precisely where a regression analysis can help figure this out.

Also, just looking at the relationship between ad spend and revenue is usually not insightful enough to actually draw any sort of actionable conclusions. In other words, you need to first understand the relationships between rank, impressions, CTR and CPC, then determine the ad spend vs. revenue numbers.

Then, comes the question of which data model to use: linear, logarithmic, exponential, polynomial, etc. A fairly simple way to go about this in Excel is to:

• Add a trend line to your scatter chart

• Try multiple regression types

• Consider the model with the greatest R-squared (determination coefficient) since the model with the greatest R-squared is the most reliable one.

Below is an example of what it looks like in Excel – similar scatter charts should be put together for the relationships between rank vs. impressions, and rank vs. CPC:

A “good enough” R-squared is usually greater than .85; however, in all reality there might be a lot of statistical noise, and you’ll most likely see much lower R-squared values. This can definitely be a major limitation when the data does not look good and your determination coefficients are low. Other than that, from my experience, the most relevant regression type is usually the exponential model.

Now, you should be able to establish the relationship between rank, impressions, CPC, and CTR such as:

• Impressions=a*Exp(b*Rank)

• CPC=c*Exp(d*Rank)

• CTR=e*Exp(f*Rank)

…where a, b, c, d, e, and f can be determined using Excel formulas based off your historical data. Then, you can simulate different scenarios by giving different values to the rank variable, from better ranks to worse values, and come up with the below chart.

Putting The IS & Regression Analysis Together

Getting back to our campaign level data, we can now leverage those regression models and apply them to each individual campaign (or ad group, if you’re looking at ad group level IS). Basically, it all depends on the current rank for each campaign or ad group. Say your current rank is 3.29 and you found that the relationships between rank, impressions, CTR, CPC were as follows:

• Impressions = 812,454 * Exp( -0.15 * Rank)

• CTR = 0.04 * Exp ( -0.44 * Rank)

• CPC = 4.20 * Exp (-0.20 * Rank)

Then, we can establish the estimated impressions, CTR, CPC at multiple ranks:

Then, just make sure the greatest estimated number of impressions is lower than the total number of available impressions calculated earlier, otherwise, it means your data model is a bit “optimistic” and you need to consider another data model or perhaps dig into the data and exclude a couple of weeks of data with seasonal peaks or so.

Once this is done, you can determine all other metrics, append your conversion rate and average order value (AOV) numbers and come up with your final ad spend vs. revenue numbers, such as:

Note that I am assuming that both the conversion rate and AOV do not vary much with position, which makes sense to me for exact keywords, but not so much for phrase or broad keywords for which the associated search queries tend to be even broader when increasing the bids hence a dip in traffic quality.

Conclusion

This methodology presents several assumptions or defects, such as:

• Lost IS rank = 0% is only achievable when rank = 1.0 which is simply unachievable most of the time, except for exact branded keywords.

• “To combat Lost IS rank you have to bid up,” which does not address missed impressions due to poor Quality Score.

• “The relationships between rank, impressions, CPC, CTR are stable over time,” which is not true because of changes in the competitive landscape.

• Most of the times, you’ll get low R-squared values when producing data models in paid search due to a lot a statistical noise.

• “Conversion rate and AOV do not vary much with position,” which is  mostly true in exact match type, but mostly untrue for phrase and broad keywords.

However, a best practice to mitigate all those assumptions and defects is to only apply this methodology to a subset of keywords in your program:

• Non-branded keywords not already in rank 1.0-2.0, since branded keywords typically do not have a lot of potential.

• Relevant keywords, i.e., those already optimized from an editorial standpoint with an ok Quality Score.

• Converting or assisting keywords. That’s where the opportunity lies anyway, so you can filter out those keywords which never assist nor convert.

In this particular case, search marketers can actually come up with actionable insights in terms of what keywords to jack up the bids on, and what a good target rank and CPC is for those guys to generate incremental and profitable growth.

There are lots of different ways to go about this – one way I’d like to cover in this post is to leverage impression share data. We’ll see how to aggregate and dig into impression share data, and how to estimate incremental revenue opportunities based on those impressions missed due to insufficient budget. Due to the length of this post, I thought I’d cover those impressions missed due to insufficient rank in a separate article.

Note that a similar logic can be applied to AdWords impression share data and Bing Ads share of voice data.

Getting A High-Level View Of Your Impression Share In Paid Search

Impression share data is currently available at the campaign/ad group levels in AdWords, and account/campaign/ad group/keyword levels in Bing Ads. Whatever level you are interested in, the same logic can be applied to aggregate impression share data up to the account level or any label/dimension you like.

The metrics I will refer to in this post will be:

• AdWords Search impression share/Bing Ads impression share (%) = IS
• AdWords Search Lost (budget)/Bing Ads Impression share lost to budget (%) = Lost IS (Budget)
• AdWords Search Lost IS (rank)/Bing Ads Impression share lost to rank (%) = Lost IS (Rank)

Say we want to analyze the below campaigns in AdWords – and we want to determine what is the combined impression share. We need to aggregate the impression share data by weighting the IS data by the number of impressions of each individual campaign.

More specifically, the aggregated IS in this example is: (120,457*98% +58,789*68% +78,456*71%) /(120,457 +58,789 +78,456) =83%

Generally speaking, if n indicates individual campaigns (or ad groups), the formula to determine the aggregated IS is:

The same thought process can be applied to the aggregated Lost IS (budget). The aggregated Lost IS (budget) is then (120,457*0% +58,789*68% +78,456*71%) /(120,457 +58,789 +78,456) =2.7% Hence, a similar formula to determine aggregated Lost IS (budget).

Same thing for the Lost IS (rank). The aggregated Lost IS (rank) is then (120,457*2% +58,789*20% +78,456*29%) /(120,457 +58,789 +78,456) =14%

Now that we can aggregate IS data, we can visualize the data at different levels — for example, by country and by product category/country:

Determining The Room For Growth

One can easily determine the total number of available impressions in the marketplace if you were not missing any impressions due to neither insufficient budget nor rank. This provides some hard numbers as of the actual room for growth.

For instance, if your IS was 68% and you got 58,789 impressions, then the number of available impressions in the marketplace was 58,789 /68% =86,454.

You can generalize the calculation to determine available impressions — for instance, by country and product:

Estimating Incremental Revenue Opportunities Based Off Lost IS (Budget)

As for those campaigns capped due budget, the logic is relatively simple. Assuming you are not changing your bids/creatives/landing pages in the meantime, one can expect stable CTR, CPC, conversion rate, and average order value (AOV) for those incremental impressions:

• Incremental Impressions (Budget) =Available Impressions * Lost IS (budget)
• Incremental Clicks (Budget) =Incremental Impressions (Budget) *Stable CTR
• Incremental Cost (Budget)= Incremental Clicks (Budget) *Stable CPC
• Incremental Conversions (Budget) = Incremental Clicks *Stable Conversion Rate
• Incremental Revenue (Budget) =Incremental Conversions *Stable AOV

Getting back to my initial example, if a campaign generated 58,789 impressions and had a IS of 68% and a Lost IS (budget) of 12%, then the total number of available impressions is 86,454, and the corresponding number of incremental impressions due to budget is 12%*58,789 =10,375. This particular campaign would generate an additional 10,375 impressions if opening up the daily budget.

Assuming the CTR and CPC are stable on those incremental impressions, then the number of incremental clicks due to budget for this campaign would be 10,375*3.4% =353. This campaign would generate an additional 353 clicks.

Assuming the conversion rate is stable on those incremental clicks, then the number of incremental conversions due to budget would be 353*2.1% =7 (after rounding this value). We can also assume the AOV is stable at $612 and come up with hard incremental revenue numbers.

At a high level, you can use the same methodology across all campaigns, aggregate the data, and come up with incremental conversion estimates by country or any dimension you are interested in.

Also, beyond the raw potential for conversion growth or revenue wise, search marketers typically look for the most efficient growth possible, which one can determine by looking at the incremental ROAS for this incremental revenue.

This is a first step to impression share analysis and potential budget reallocation. In the above example, the UK market seems to have lots of potential in terms of revenue volume, however the CA and FR markets seem more profitable, so you might want to invest more there first.

Again, since the same methodology can be applied at the country, product category, brand vs. non-brand levels, or any dimension you’re interested in – this analysis can help go beyond traditional paid search and make better business decisions based off not only current performance but also future growth.

In this article, I’ll share a couple of thoughts which I hope you’ll find helpful when testing multiple ad copy and landing pages for your PPC program.

Get Your Campaign & Ad Group Ready

Once you have a fairly mature keyword list, you should be able to easily identify those top cost keywords with a Quality Score lower than 7/10. From my experience, 7/10 is definitely decent for non-branded keywords, while anything below 7 usually results in greater CPCs and greater first-page bid estimates.

By top cost keywords, I basically mean generic keywords – those guys where any slight Quality Score improvement can make a huge difference. You might want to isolate these keywords in specific ad groups so you can fine-tune specific ads and landing pages specifically for them – and you want to take care of this before getting the A/B test started.

This test might be a good opportunity to break down your top campaigns by device and/or geo so you can potentially look into device and/or geo specific data and come up with the best creatives not only for each top keyword, but also by device and/or location. This breakdown might be time-consuming and dilute the data; that’s why you might want to tier your campaigns.

Last, but not least, you want to make sure you’re using the “Rotate evenly” feature at the campaign level to give equal preference to all active ads in your ad groups. Keep in mind that “if ads in an ad group are unchanged for 90 days, the ad rotation in this ad group will automatically begin to optimize for either clicks or conversions” (see AdWords page).

Watch The Competition

As a numbers person, I wish I could come up with a formula to determine the best ad copy based on historical data and combinations of headlines, descriptions, display URLs and landing pages. The reality is that your search program success not only comes from your ability to draw conclusions looking at numbers, but also from your awareness of the competition.

More specifically, you should definitely search for your top keywords in Google and Bing on a regular basis, and see what the competition is doing in SEM and also SEO, then identify opportunities in terms of unique positioning: price, quality, free shipping, etc. You can use AdWords AdPreview tool to see “unbiased” search results.

Set Up New Ads & Landing Pages

This is definitely not a one-time thing, but rather, an on-going process with a 90-day limitation coming from AdWords.

1. Start with significantly different ad copy; each ad should boast a unique differentiator (price, quality, etc.)

2. Once you’ve determined the best axis of communication, fine-tune your headline without touching the description and display URL

3. Now that you’ve determined the best headline, test different descriptions

4. You can now focus on testing different display URLs

The scheme below represents this process in a simplified way:

Those tests will impact your CTR, Quality Score, rank, average CPC. They might impact your conversion rate, too. Ideally, you want your test to be as clean as possible, which means that you want to touch just one component at a time: headline, description, display URL, or landing page. This will help isolate every factor when determining what the champ ad is.

However, because of the relationship between CTR and conversion rate, you might want to test multiple ad copy and landing pages at the same time. You can rotate four ad copies per ad group with two unique ads and two unique landing pages so you can still get a good sense of the ad copy with the best CTR and conversion rate.

Top vs. Right-Hand Side Distribution

When testing multiple ads and landing pages, you definitely want to keep an eye on the distribution by “Segment: Top vs. Other” in AdWords, which simply indicates whether your ads are shown as the top of the search results, or on the right-hand side. In the example below, we tested different strategies for one of our clients from a CTR to a conversion rate focus, until we found the balance between “good enough” CTR and strong conversion rate.

The bottom line is that a successful A/B test is a blend of “appealing enough” ad copies to secure a high Quality Score and top positions as a result, and a strong conversion rate. Search marketers should work on determining what is the lowest CTR to secure premium positions, then maximize the conversion rate while maintaining this minimum CTR level.

If you are an online retailer and your business is significantly dependent on product level keywords and ads, you might be interested in hearing about a couple of ways those processes can be improved – your priority being to offer the most relevant results for product-related queries.

Controlling Product Level Keywords/Ads In Traditional Paid Search

There are many ways to build out paid search campaigns based off a product-level feed; however, most of them are not scalable with many feed updates. One is to use MS Excel to generate keyword strings and ad copy based off product-level information. I have personally utilizes this strategy before and ran into three main challenges:

• Generate relevant keywords and ad copies. There are so many scenarios you need to anticipate, such as, some product names are too long for the ad coy headline, or even too long for the ad copy description. In case you’re using trademarked terms you might need a back-up ad copy, etc.

• Create new ad groups for new products each time a new product becomes available.

• Pause existing ad groups for out-of-stock products.

There are a couple of technologies out there. If you search for [inventory driven search], [dynamic inventory management paid search] or similar queries in Google or Bing you’ll definitely get to know more about tools that can help you address those challenges.

As I mentioned in a previous post, my company’s solution does a pretty amazing job at tackling those issues. We recently launched 40 fully-automated campaigns based off a 200,000+ SKU feed; hence, 200,000 ultra-targeted ad groups with unique keyword combinations, relevant ad copy, and deep landing pages.

Without getting into more details, I believe it is fair to say this is the ultimate solution to automatically manage an inventory of that scale, especially since ad groups get created and paused daily based on feed updates. This solution is also the best set-up in terms of control since one can have 10-20 exact keywords for each individual product.

Controlling Product Listing Ads Campaigns

A strong PLA strategy is critical for retailers… especially since Google Shopping is no longer free! Etailers definitely need to allocate more resources to this channel and explore new ways to maximize the channel.

A couple of interesting articles about PLAs do mention that, ideally, you want to build one ad group by product category. In other terms, a catch-all ad group which covers all auto-targets (=all products) is not optimal.

More specifically, advertisers can even go further than the product-category level and actually build out one ad group by product (=unique auto-target). Most advertisers will find that this way requires too many ad groups to create and effectively manage. Product-level ad groups definitely are more time-consuming without a tech solution; however you can then:

• Break down your top products in separate campaigns for budget allocation purposes

• Set specific bids (or CPA targets, depending on the pricing model you’re using) for each individual product

• Customize ad copies by product

Also, while the challenges of successful PLAs are similar to traditional product-level paid search campaigns, it is worth noting that:

• Like traditional campaigns, product-level PLAs require 1) new ad groups and auto target for new products and 2) ad groups to be paused when out of stock

• Unlike traditional paid search campaigns, PLAs don’t require keywords; that’s one less challenge

• Unlike traditional paid search campaigns, URLs are set within the Google Merchant Center feed and not in AdWords

On a side note, it also makes sense to collect all traffic and revenue data for both non-PLA and PLA campaigns in one place; so, search marketers can potentially leverage any transition paths from non-PLAs to PLAs campaigns and vice versa.

Long story short, online etailers should definitely look into technologies which can help build out both product-level traditional paid search and PLA campaigns. Not only is the methodology to get the most control over your PLA campaigns similar to traditional paid search, but the technology required to generate highly-targeted ad groups for traditional paid search and PLA is very similar as well.

Cross-Domain Clicks & Revenue Attribution

This is the key — you want to be able to allocate the end-revenue numbers back to every single touch-point across multiple websites, so you can adjust your marketing mix accordingly.

For instance, say you have multiple AdWords and Bing Ads accounts for your websites A, B and C. Those might offer the exact same products across multiple domains, or similar products with unique quality, positioning and pricing.

In both cases, we’re interested in diving into cross-product and cross-domain selling patterns. More specifically, online marketers should definitely approach their businesses as a whole, and make sure the tracking solution they are using allow them to take into consideration consumer behavior across multiple products and domains.

Cross-Domain Insights In AdWords

Unfortunately, AdWords does not allow users to look into cross-account consumer behavior. Only cross-campaign/ad group/keyword data are available, and not across multiple AdWords accounts.

However, if you have just one AdWords account where you manage multiple domains (which is unlikely but still possible) then you should be able to analyze cross-domain transition paths.

Tracking Challenges

Similar to the challenges of cross-device tracking, it can be tricky to track cross-domain conversions. Not for the same reasons, though.

On one hand, measuring conversions across multiple devices (desktops, tablets, mobiles) is difficult when using cookie-based tracking solutions because of the unique cookies set on each individual device/browser.

On the other hand, measuring conversions across multiple sites (sites A, B, C, for instance) is complex because of the unique cookies set for each site, even though Web users are using the same browser.

Long story short, the data get compartmentalized by site and by device; and, unless you’re using advanced tracking solutions, you might not be able to get the full picture.

Connecting The Dots

There is Google Analytics. It’s free, simple to use, and fairly reliable. You can get insights into cross-channel transition paths, and look into cross-domain conversions only if you have thoughtfully tagged your marketing campaigns using one of the dimensions available in Google Analytics.

However, just like in AdWords, you won’t be able to actually allocate the revenue back to each touch-point since this feature is not fully integrated in Google Analytics. Nor will you be able to manually or automatically adjust the bids based on those multiple touch-points.

In short, if you are ok manually analyzing the data, Google Analytics might be good enough. Otherwise, you are going to need a more sophisticated solution in order to automate the revenue-allocation process and then adjust the bids accordingly.

In a nutshell, a successful multi-domain strategy relies on a robust tracking solution. Marketers need to be able to put together multiple data-sets by domain so they can actually make business decisions from a holistic standpoint. The technology is already out there, and becomes even more relevant when measuring online marketing returns for multiple domains across all major online advertising networks.

It’s all about engineering a consistent conversion chain from search to conversion, which implies high resolution targeting.

In an ideal world, search marketers should break down their paid search accounts by user intent, geo, language, device, and more… as long as they see some kind of gain targeting wise.

That all makes sense so far, except that you might end up with several thousands of campaigns, duplicate ad copies and keywords across those campaigns, etc., and gradually lose control over your paid search program.

More specifically, you might run into a scalability issue, which raises the following question: where is the middle ground between high resolution targeting and manageability of your paid search effort?

In this article, I’ll review the main segmentation possibilities in paid search, then identify those that always make sense (=”Must-Have”), and those who might need more in-depth thinking before rolling them out across the board (=”Advanced”).

Segmenting Options: By Account, Campaign & Ad Group

Let’s start from the beginning. Search marketers have lots of segmenting options at their disposal; let’s start at the account level, down to the campaign and ad group level.

You might want to have multiple AdWords accounts if you need to:

• Set up multiple billing settings (for different businesses for instance)
• Manage user access by account
• Build more than 10,000 campaigns (current AdWords limitation)
• Isolate keywords with a poor (or strong) Quality Score, since the account history is one of the major Quality Score components.

If you don’t need any of these, then you just need one account.

Then, you might want to create separate campaigns if you need to:

• Allocate separate daily budgets
• Segment Search vs. GDN (Google Display Network)
• Segment Desktop vs. Tablet vs. Mobiles
• Segment by geo (state, city, zip code, it can get incredibly granular)
• Segment by language (English vs. Spanish, etc…)
• Exclude terms for certain campaigns only
• Create sitelinks or any other type of ad extensions

Finally, the ad group level breakdown is really about relevancy and negative keyword management in AdWords, since they are no ad group-level targeting settings (while there are some in Bing Ads).

Must-Have Segmentation #1: Search vs. GDN

While some advertisers still set campaigns to target both the search and GDN networks, it’s been a couple of years since the whole industry has seemed clear about this – you definitely need to split your campaigns by search network vs. GDN.

There are multiple reasons for this:

• GDN campaign structure is completely different, since you can not only bid on keywords, but also on, placements, audiences, topics.
• GDN ad copies should be CTR-focused, since the historical CTR is an even more important Quality Score on the GDN.

Must-Have Segmentation #2: Brand vs. Non-Brand

For many reasons, you definitely want to isolate your branded keywords:

• To maximize your impression share for your branded keywords (in practice, you want to make sure this campaign is never capped due to budget while maintaining high Quality Score and aggressive ranks at a low cost per click).
• Set up branded site links.
• You might want to exclude your branded terms from your non-branded campaigns to silo impressions and make sure you’re measuring brand vs. non-brand performance, as opposed to having everything mixed together in one place.

Advanced Segmentation #1: Device Targeting

For lots of advertisers, this is actually a must-have segmentation, since you most likely want to:

• Allocate device-specific budgets
• Measure clicks-to-call and other device-specific metrics
• Set device-specific bidding strategies
• Adjust your keyword list by device
• Write specific ad copies by device (even by operating system or carrier)
• Use device-specific landing pages.

However, if you are already managing 100 campaigns and need to update your ad copies on a regular basis (special offers, holiday season, etc.), perhaps you should not have 300 campaigns instead of 100 because of a scalability issue.

Will the gain in performance outweigh the loss in manageability?

There is no generic answer to that, and it all depends on the paid search platform that you’re using (for instance the ability to easily duplicate campaigns), although it is fair to say that one should take care of branded and the top revenue keywords first.

Advanced Segmentation #2: Geo Targeting

Similarly to the device segmentation, you might be tempted to build out a huge number of campaigns by geo, knowing that geos can be pretty much as granular as you want them to be: by country, state, city, latitude/longitude, zip code, etc.

It makes a ton of sense when wanting to:

• Allocate geo-specific budgets
• Set geo-specific bidding strategies
• Adjust the keyword list by geo such as [keyword] + [city], etc…
• Write geo-specific ad copies
• Leverage location extensions if applicable

Again, if you already have 100 campaigns, do you really want to multiply this number by 50 to target each individual US state?

How manageable would 5,000 campaigns be vs. just 100? Same piece of advice here, one should align the effort with what it is worth.

More specifically, what you should really care about are just those couple of top revenue geos (whatever they are: states, cities, etc…) where you are seeing a significantly lower or greater ROAS (return on ad spend) – then you should definitely adjust your spend and strategy by geo for those guys in order to re-allocate your budget from low-performing geos to high-performing geos.

In short, you don’t really need to go through that process if your current ROAS by geo is the same nationwide.

Addressing The Scalability Issue

Beyond the above segmentation possibilities, there are many other ways to break down your paid search program (by language, Google and Search Partners vs. Google only, by time zone for more granular hourly bids and more…), and search marketers will always run into some kind of scalability issue at some point.

So far, I have mostly referred to scalability issues as maintenance/manageability issues, however there is another issue involved with building out high resolution paid search account: data dilution. This can get a serious issue and paralyze your bid decisions if you just don’t have enough data by geo, language or device.

In a nutshell, it is every search marketer’s role to draw the line between granularity and manageability. It depends on the resources available, the maturity of your program, and the platform you’re using.

A general piece of advice from my experience would be to always tier your account by whatever targeting settings you’re considering, and then get more granular just for your tier 1 campaigns (and maybe the tier 2 campaigns as well if reasonable).

To clarify, tier 1 campaigns are typically those accounting for 50% of the conversions/revenue.

As a result, a granular while manageable account structure could be:

The bottom line is that it might not be worth treating tier 2 and 3 campaigns like tier 1 campaigns.

No big surprise here, however, I believe it makes sense to keep this in mind when building out your account. That should help you scale up your account while keeping control over it.

In other terms, now that you are most likely finding that a great percentage of your conversions occur after more than 1 just impression or click, let’s focus on your paid search program and define some simple action items based off a higher resolution consumer behavior analysis.

Identify & Leverage Top Assisting Keywords

You have certainly heard about “upper funnel keywords”. Those keywords are usually generic, non-branded, and play an essential role in your paid search program – and more generally in your online marketing efforts.

Upper funnel keywords assist other keywords in your program and don’t necessarily get the full credit for it, depending on how you look at the numbers and what platform you’re using.

In AdWords, those assisting keywords can be found in:

‘Tools & Analysis>Conversions>Search Funnels>Assist Clicks and Impressions’

You can then set the primary dimension to be ‘Keyword’ and sort the table by ‘Assist Clicks’.

Since assisting keywords are usually generic, they are high-traffic keywords. Since they are non-branded, you can most likely demonstrate that they tend to generate a high proportion of new customers vs. repeat customers – which basically means that the lifetime value (LTV) for those keywords is much greater. You might want to adjust your efficiency targets (CPA, ROAS, etc..) accordingly as a result.

Long story short, attributing the full credit to upper funnel keywords while baking the LTV component into it are two essential conditions to scale up your paid search program to the fullest.

In practice, you might want to:

• Set a separate bidding strategy for those top assisting keywords if the platform you’re currently using does not allow you to easily see the “true” revenue numbers.

Say you’re using AdWords – then you can identify top assisting keywords but you cannot set bid rules based off multiple touch points. You’ll have to analyze the data offline, then come up with some ideal bids which take into account consumer behavior. You might also want to simply target a certain rank to secure impression volume andboost customer acquisition on those competitive search queries.

• Apply a custom revenue attribution model (if the platform you’re using allows it) in order to display the “true” revenue numbers and make more informed decisions.

Pause Keywords Which Never Convert Nor Assist

As a matter of fact, some keywords never convert nor assist, they only generate ad cost. You definitely want to pause those; however, you might want to make sure the numbers you’re looking at are statistically significant.

For instance, you might consider that 100 clicks are required to be 95% confident that your results are significant, in which case you should:

• Pause those keywords which generated a statistically significant number of clicks, did not convert nor assist
• For low-volume keywords: Pause those ad groups which generated a statistically significant number of clicks, did not convert nor assist

Repeating this process on a regular basis (at least once a year if not more frequently) typically saves around 10% of your overall ad spend with zero impact on revenue.

While there is much more to say about how to leverage the search funnel to the fullest, these two simple tips are from my experience, a good starting point as they are quick to implement and pretty impactful.