Google Ads for B2B

Too much sparse data with high CPCs on your B2B ads? This means lots of room for optimization.

Scaling a B2B company with Google Ads can be very challenging. You would want to show your ads to people who are qualified for business but end up having B2C traffic with very high Cost-Per-Clicks (CPCs). Irrelevant traffic drops conversion rates while you’ll have to deal with sparse data. Instead of waiting for the right clicks to come, you can start optimizing Google Ads for B2B business right away.


Not every B2B environment is comparable to each other. For that reason you should look at your brand traffic to understand what your average customer is looking like. Here are the most important aspects you should check:

What is your brand mobile share?

I know – everybody talks about the importance of mobile – in B2B it can be a different story. Take the mobile traffic share as a good estimation what your non brand mobile share should look like.

What is your brand weekday seasonality?

You will often see a peek in demand on working hours during the week. On weekends the brand traffic is quite low. Always think of your keywords that are also triggering the wrong customers.

Where are your users located?

You will find geographic areas where mostly companies are located. But you will also find the opposite way, where the probability of bringing a B2B lead to your website is quite low.


Conversions events are often very rare in B2B accounts. For that reason it makes totally sense to use some soft conversions, like a sign up, to get more data for judging your keywords. In a lot of cases you will still run into sparse data. Google’s Smart Bidding will probably not be able to handle those areas well. For that reason let’s go back to the fingerprint idea and do not look at conversions for a moment.

We have lots of impression data. Even for detailed views like search query level or very specific locations like the postal code.

Keyword Fingerprint

Take e.g. (impressions(Monday) + impressions(Tuesday)) / (impressions(Monday + Tuesday + Saturday + Sunday))
In other words: The b2b impression share for your keyword. Thesis: The closer the number matches to your brand traffic values the better the traffic will be.

Postal Code Fingerprint

You can use the same business share logic for postal codes. You will be surprised how good this classification works e.g. in the US market where all business districts in inner cities pop up with a very high business share.

  • I do not want go in too much detail. But you can run those impression fingerprints also on search query level or even on n-Gram level to derive some negative keywords of B2C search patterns. For now we just talked about using the week seasonality. Of course you can also do this with the mobile traffic share. When the share is a way higher for single keyword compared to your brand number you will probably trigger more of those bad performing B2C visitors.

But before we are making actions we should look at conversions again and proof our thesis.

Create Impression-Share Buckets

Round the impression share for your B2B customer on 1-digit. By doing that this will result into a maximum of 10 buckets.

Group Performance by Buckets

When doing this you will probably realize huge performance gaps. Now we are able to make good estimations for sparse data keywords.

Focus on B2B Postal Regions

You can use the demand fingerprint right away and start adding location targets on postal code level. Bid up on B2B areas and down where they are very unlikely.

Run your Own Bidding

Congratulations! If the performance gap between the buckets is high, you have found a way to outperform your existing bidding. Use the cluster performance for bid estimations.


Maybe you have used the network domain and the service provider dimension in Google Analytics in the past to figure out which companies visited your website. In 2020 Google deprecated both dimensions. The good news: You can build something more powerful to identify your B2B users.

Use 3rd party tools for enriching the IP

There are some tools out there that are giving you details based on the users IP address. Kickfire or clearbit are some to mention but you can even build something great with a small budget using IP Databases like IP2Location.

Track company name and connection type

Write the enriched data to your web analytics system. After that we are able to use it as a attribute for a user scoring or as a new conversion goal.


Fixed IP Address + Interaction with content

This conversion goal is straight forward. By doing this your conversion goal now is completely based on the B2B user. Because no conversion is needed we have a way more conversion data even on keyword level.

User Scoring

The more sophisticated approach is to use these data enrichments together with site interactions to create a prediction model for the sales probability of a website visitor. Use this score as new conversion goal for your marketing activities.

Do company retargeting

Use the newly created data to run remarketing actions in Google Search or Google Display Network. I highly recommend a scoring based approach for creating the audience lists.

Change your bidding goal

Our cluster approach was nice for creating some good estimations for keyword bids right away. Now we can switch to the new powerful conversion goals.


Use pay for conversion as bidding strategy

If possible use this option to share some risk of display fraud with google. Use an IP related target as conversion – Lead signups can be faked easily and bet this is done. By combining the sign up with the condition that it has to come from a company IP Address it is a way more difficult.

Optimize heavily to get ongoing traffic with pay for conversion

Just setting the bidding strategy to pay for conversions does not mean that your work is done. After Google’s testing I observed a lot of cases where the traffic dropped and never went up. If the Value Per Click on googles side is too low they are sending the traffic to someone else. It is simple like that. For that reason you should optimize like you have to pay for each click.

Block bad placements

You will find a lot of placements that cause a lot of useless clicks. They also often share the pattern that they have Click Through Rates that are a way above average. Block them and all other placements that are related to this website. You can use Publisher Id and IP data of the website to achieve this.

Use location micro targeting

Do you remember the fingerprinting on the postal code we did? Use approaches like that to set up geo micro targeting. It is all about thinking in odds. By doing this location targeting they will increase to trigger your target B2B Users. Google does not know that pattern so we have to tell.