What PPC Impression Share Really Tells You

By , Online Marketing Manager at ImmobilienScout24

218 SHARES

Impression share metrics are often used as the key indicator of one’s market share. A 100% impression share would indicate that you are reaching the entire traffic volume. But this metric would be misleading.

 

Our case study reveals that total impression share can be a quite misleading metric for determining your share of total voice. The metric is a better indication of your share in auctions. Moreover, we show that your bid influences the number of auctions you participate in.

 

The Case Study – And How We Were Fooled By The Impression Share Metric

 

A keyword’s ad rank is determined by multiplying quality score and the max CPC:

 

  • Ad rank = Max CPC x Quality Score

 

Therefore, the keyword with the highest bid and quality score is ranked highest. Based on this model, we proposed the following hypothesis:

 

If we are the only bidders on a certain keyword and have a constant quality score of 10, we can set our max CPC to a minimum and are still ranked on position 1. Thereby we can reduce our price per click and our costs.

 

As a secondary condition, we assumed that our market share remains constant. Thus, we gradually lowered the max CPC for an exact keyword which has the following attributes:

 

  • High impression volume
  • Low competition
  • Average position 1
  • 98.01% total impression share (1.99% were lost due to rank)
  • Constant quality score of 10
  • Not targeting mobile devices
  • Google Search only
  • No budget restrictions

 

An astonishing thing happened. As expected, the CPC decreased by 40% compared to the previous period, but the total numbers of impressions dropped dramatically (-95.46%). The following graph shows the number of impressions dropped significantly on March 31st.

 

Image of index impressions vs. index CPC

Development of total impressions and CPC

 

But looking at the impression share, we still nearly reached the entire audience (impression share 97.96%). Additionally, we continued to be ranked at position 1 and the competitive situation stayed constant.

 

Image of Search IS and Position

Development of Impression Share and Position

 

We first concluded that the total market volume declined. Let’s do a small calculation to see what that would mean:

 

  • Total market volume = (Impressions / Search Impression Share)  x 100 %

 

So the total market volume dropped by 95.44% within one week? That seems inaccurate. Thus, we assumed that the impression share and the CPC must be dependent, so increasing the CPC would recover our impression volume. And that’s what happened: our total impression volume nearly recovered (there is still a seasonal decline due to Easter holidays):

 

Image of index impressions vs. index CPC graph

 

In the end, the total impression metric was not reliable and seemed to be dependent on the CPC.

 

But How Big Is My Piece Of The Pie? Can I Determine The Market Volume?

 

To see how the traffic volume really developed we looked up the keyword in Google Trends.

 

Image of Google Trends data

Google Trends: Search volume development for the respective keyword

 

The data proves our latest assumption: the total market volume did not diminish that harshly. The decline in search volume can be explained by Easter holidays in Germany. Unfortunately, Google Trends only provides index values but not total numbers.

 

Another slightly better tool is the Adwords Keyword Planner. Since the latest update, it provides retrospective search volume data and some real numbers.

 

Image of Google Keyword Planner

Google Keyword Planner: Retrospective search volume for the respective keyword

 

So let’s compare the keyword planner data to our monthly data. We assumed that the values in the keyword planer are displaying the whole market volume. Note, that we have excluded the mobile search volume. The search volume share is calculated the following way:

 

  • Search Volume Share = Impressions / (Total Search Volume – Mobile Search Volume) x 100 %

 

Our search volume share in March was 86.80% and 66.10% in April. That differs greatly from the impression share metric in the campaign interface, which was 97.81% in March and 97.43% in April. So we lost market share in April.

 

Although this tool provides better estimates for your market share, it has some heavy drawbacks:

 

  • Figures are rounded up
  • Data is accumulated across all devices
  • Data is only available for last 12 months
  • Broad and exact match keywords search volume is the same

 

It is difficult to determine the real search volume with Google tools, but at least you can get a rough estimate.

 

What The Impression Share Metric Might Really Tell You – Be Aware Of The Minimum CPC

 

You might wonder how the discrepancy between the total search volume and our total number of impressions can be explained and what the impression share metric is really about.

 

It is quite possible that not every search request leads to an ad auction. For example, people use ad blockers or the Google algorithm classifies keywords irrelevant for a certain user history. That’s why your impressions mostly differ from the total search volume, even if you have 100% impression share.

 

But there might be two other explanations:

 

  • A simple assumption seems to be quite reasonable: Google utilizes a quite uncommon pricing model. You have to pay an additional charge for a higher volume of traffic instead of getting volume discount. Consequently, the impression share would be directly related to your max CPC.
  • The more plausible explanation states that a minimum CPC is required to enter the auction. Google confirmed that some years ago (read here). The minimum CPC is automatically determined by the Google algorithm. At least 0.01€ is required, but it is higher the lower your quality score is. Furthermore, you have to consider that the real quality score range differs from the quality score in the interface. Even a quality score of 10 in the interface does not mean that you have the highest possible quality score in the Google system. Moreover, the quality score fluctuates from auction to auction. Assuming that our quality score is not the highest value possible and our minimum CPC is not 1 cent. By reducing our bid it was more likely that we underbid this minimum CPC in more auctions than before.

 

To undermine the last hypothesis we used the bid estimator within the keyword planner keyword. The following graph illustrates that a minimum bid is required (in this case more than one cent) for the respective keyword. Above this threshold the number of impressions stays constant. Moreover, it shows that the potential audience is around 4,100,000 per month if you bid more than the minimum CPC.

 

Screen Shot 2014-08-22 at 10.39.45 AM

 

Finally, we conclude that the impression share refers to the number of auctions in which you have competed, not your share of the whole market volume. There is no direct relationship between market share and CPC, but between the numbers of auctions you participated. But the number of auction is indirectly related to your bid. The number of auctions in which you are competing, depends on you overcoming the minimum CPC requirement.

 

Image of CPC graphs

 

What Can PPC Hero Readers Learn From Our Case Study

 

  • Be careful when using the impression share. It only tells you the share of auctions you competed but not your share in the market.
  • If your number of impressions drops and you cannot explain it by seasonal effects, remember the minimum CPC hypothesis: You need to overbid the minimum PPC to enter the auction.
  • There might be different reasons why you are below the minimum CPC
  • Last but not least the keyword planer tool might help you to find the minimum CPC and give you a rough estimate of the total market share.
  • Always be sure that you bid at least the minimum CPC.

 

The case study leads to further questions: Are there ways to detect your real market share?  How can I calculate my minimum CPC? If you have ideas I would be happy to discuss!

Get more weekly links with our Fast Five newsletter! Five Fast Links in Your Email Every Friday.

Also send me a daily RSS digest

Social Advertising Toolkit

Twitter Facebook LinkedIn Google+ Email Print More
  • Sam Owen

    Hey Carina,

    This is really interesting stuff.

    I was wondering if you could share what you changed your bid from and to? Was it €5 to €0.05 or more like €2 to €0.50?

    Have you tried slowly bidding up / down at all to find where that cliff is?

    • Carina Schwarzmüller

      Hey Sam,

      Unfortunataly, I cannot tell how high our bid was.

      In the beginning we cut our bid by around 50% because the max CPC was much higher than the average CPC. Then we bid down in smaller steps until we reached the theshold.

      In our case the threshold was identic to the minimum bid displayed in the keyword planner tool.

  • disqus_p7okpEispV

    I’ve talked with Google on an aspect of this…
    Q. When there’s nobody else bidding on a phrase how come there’s a (high) minimum amount to bid to achieve first page results?
    A.In the absence of other bidders, Google introduces algorithmic benchmark ‘quality’ competition – in the real auction world this is called ‘pennies from heaven’ or imaginary bids. In Google’s world it’s called ensuring quality (in Google’s profit?).

    • Sam Owen

      Thanks, that’s really interesting and something I’ve always suspected happens.

  • http://www.iConversing.com iConversing

    Google help mentioned : “It includes all auctions where your ad showed or where the system estimates that your ad was competitive in the auction. For example, it could include auctions where your ad could show at twice its current bid, but could exclude auctions where your ad is estimated to need a 1,000% bid increase in order to appear.”

    Hence I would say its not exactly the share of auctions competed, but rather as google mentioned where the ad was competitive (which is why there is lost IS due to budget).

    Also i think the minimum bid is really what is playing a part here as you mentioned. Even with little or no competition, there is the minimum bid. And it also differs for various keywords i.e. keyword with “insurance” in it will have significant minimum bid. I didnt get a good answer from the google support specialist, our conclusion is it is another “special algorithm”, which likely depends on the nearest keyword and your quality score.

  • Ben

    Hi Carina,

    A few points on this article:

    1. Impression share only shows you your IS% in auctions you are eligible for.

    - That’s pretty much always been the case because if you’ve added a negative, or reduced your ad scheduling to not show at certain times of the day, or excluded mobile then you were never eligible for those impressions. Otherwise you would have campaigns where you were achieving 100% impressions share for all your are eligible for. But <10% in the interface because you have reduced your scheduling, or added negatives.

    It would make no sense for it to tell you how much total impressions you could get if you removed every possible form of targeting. Impression share is based on your targeting and thus what you are eligible for.

    2. Impression share and average position stayed constant yet CPC's continued to drop.

    -There are quality thresholds for ads to serve. Therefore, even if there was no competition (which is highly improbable), a keyword can lose impression share due to rank if it does not meet the quality, relevance, and ad rank thresholds to serve. This means that the first assumption raised is faulty; you can not reduce your bids indefinitely and still show for every auction.

    What I suspect has happened here is changes to the account outside of you just lowering or increasing a bid.

    For instance, if you had 80% impression share for campaign X it could be that a broad match keyword alone is responsible for the loss of up to 20% in the campaign (versus say your exact match keywords with 100%). By adding that broad match as a negative or reducing it's match type you would naturally increase/stay level in impression share without theoretically increasing or decreasing your bid. In fact, with a decreased bid you could still achieve 100% as the exact match keywords (with typically a higher QS) would maintain their rank fairly consistently while the broad match would be more susceptible to a drop in impression share given the sheer number of extra keywords a broad match encompasses.

    I think in order to confirm if your suspicions are true we would need to see the full change history on the account and campaign in question. Because reducing bids ultimately reduces ad rank, as your ad rank decreases at some point so to will your impression share. Unless you decide to make other campaign changes which would invalidate the point of the test.