Hiscox and Croud supercharge PPC results through machine learning

May 2018

The financial services paid search landscape is increasingly competitive, so organisations need to be much smarter in their PPC tactics and leverage better targeting to mitigate costs. Premium home and business insurance provider Hiscox enlisted digital agency Croud to reduce their cost per action and increase their conversion rate. Croud embraced Google’s machine learning capabilities, which allow advertisers to be highly targeted and win the right types of auction.

About Hiscox

Premium home and business insurance provider

Home insurance cover for over 60,000 homes

Tailored business insurance for more than 150,000 UK SMEs, professionals and consultants


About Croud

Digital marketing agency

Specialists in PPC, SEO and content, programmatic, paid social, creative and analytics

Offices in UK, US and Australia


Remain competitive in increasingly aggressive marketplace

Increase volume of conversions from paid search

Decrease overall cost per click and cost per action


Used Google’s machine learning products including Dynamic Search Ads, Smart Bidding and Smart Lists

Adopted a test-and-learn approach

Gathered insights to inform and optimise full rollout of each feature


Dynamic Search Ads increased click-through rate by 8% and reduced cost per quote by more than £10 compared to non-brand search

Smart Lists converted 62% better on average than standard search

AdWords Smart Bidding improved conversion rate by 48% and reduced cost per quote by 39%

Working closely with Croud, Hiscox is leveraging some of Google’s most powerful machine learning tools – Dynamic Search Ads, Smart Lists, and Smart Bidding – to drive superior results across its paid search account.

Dynamic Search Ads

Dynamic Search Ads use machine learning to pick up long-tail, low-volume search traffic that might otherwise be missed, creating bespoke ads and selecting the most relevant landing page based on the search terms used. Hiscox and Croud implemented Dynamic Search Ads over an initial four-month trial period and saw impressive results, including an 8% increase in click-through rate compared with the rest of the account and a cost per quote that was more than £10 lower than across non-brand search.

Following the successful trial, Dynamic Search Ads have now been rolled out across the entire Hiscox account, running at an average cost per action that’s 5% cheaper than non-brand search overall. Not only that, but using Dynamic Search Ads has provided Hiscox with valuable insight into consumer behaviour by surfacing long-tail search data. This has enabled the insurance provider to identify new opportunities, such as targeting people searching for personal camera insurance using specific long-tail keywords.

“Machine learning-based tools have proven particularly invaluable in keeping cost per action down, whilst also giving us great insights into our own users’ search behaviour”

- Stuart Mahoney, Head of Acquisition Marketing at Hiscox

Smart Lists

Another key initiative driven by machine learning has been the use of Google Analytics Smart Lists. Using dozens of signals, including location, device, browser, referrer, session duration and page depth, Analytics identifies users that are most likely to convert, and dynamically manages the remarketing audience to focus on those users, adding or removing users based on a constantly evolving model. Hiscox has seen great results from this, with Smart Lists converting 62% better than standard search.

Smart Bidding

AdWords’ Smart Bidding capability enables Hiscox to optimise towards specific goals, using information around campaigns, ad groups, keywords and additional data, such as device, day of week or time of day and location. When applied to Hiscox’s Dynamic Search Ads campaign, this produced a 48% uplift in conversion rate and a 39% decline in cost per quote.

In summary, the power of machine learning has enabled Hiscox to outperform the competition and stay ahead in a highly saturated marketplace. Through these tools, Hiscox has been able to lower the average cost per quote, as well as drive clicks and conversions, all whilst gaining useful user insights.

“At Hiscox we’re regularly seeing increases in competition across volume-driving head terms, which increases our average cost per click and cost per action”, explains Stuart Mahoney, Head of Acquisition Marketing at Hiscox. “We work closely with Croud to make sure we’re exploring every opportunity for driving cheaper actions, from long-tail terms to ad copy and landing page testing. Machine learning-based tools have proven particularly invaluable in keeping cost per action down, whilst also giving us great insights into our own users’ search behaviour. We’re excited to continue testing Google’s latest machine learning innovations!”

Automation is the key to taking your search marketing from good to great