Performance Marketing Optimization

Introduction

The PHH Group was formed by merging Lithuanian Pigu. lt and Finnish retail giant Hobby Hall. The group operates online shops and marketplaces in Finland, Estonia, Latvia, and Lithuania. Over 4,000 merchants sell on the group’s marketplaces like Pigu.lt, 220.lv and Kaup24.ee. The PHH group owns five eCommerce properties across four EU countries that handle 400M+ in GMV (gross merchandise value) combined.

The PHH Group recently crossed the one-year anniversary of our partnership with Isima and the use of its bi(OS) real-time data platform.  During this time, we went through a war in our backyard, an upheaval in the marketing landscape, and a self-serve bi(OS) experience.  Through this journey, our commitment to Isima increased by 3X. Now, bi(OS) manages real-time product feeds for 4M+ SKUs across Google Ads, Google Shopping, and Facebook.  I will highlight the MarTech use case that relies upon bi(OS) and conclude with a qualitative commentary about our journey.

The state before bi(OS)

I headed Growth and Performance marketing for the PHH group. One critical aspect of performance marketing is managing ROAS (Revenue on Ad Spend) effectively. It requires iterative ad campaign management that relies on the cleanest and most real-time product catalog (e.g., the latest prices). Our IT teams compile this product information into files “pulled” by Facebook and Google Shopping at a defined frequency.  

This approach had the following flaws:  

• Product information (e.g., prices) could only be updated once daily/hourly. 
• Due to bugs, incorrect product information would be caught hours or days late.
• As the product catalog grew, uploads became brittle. Due to a large number of products, we experienced time-outs when trying to upload.

Even when it worked flawlessly, this approach suffered from the following shortcomings¹: 

• False marketing complaints: Savvy consumers noticed price differences between ads on Facebook/Google and our eCommerce properties. This would result in complaints filed to authorities, increased customer support calls, and a sub-optimal customer experience.
• Increased CAC²: We removed prices from ads. This had the impact of higher click-through rates but reduced conversion. In other words, consumers clicked only to see the price, but many dropped off as soon as they saw it.

We felt the pinch when the Ukraine war came to our doorstep, and geopolitics impacted everyday lives. Imagine a ~30% drop in customer activity in ~ five days.

The bi(OS) Difference

bi(OS) and its MarTech pack can be used to build a real-time product catalog from 3rd-party SaaS sources such as GTM (Google Tag Manager) and internal IT sources.   Changes to product attributes are detected and sent in real-time to Google Shopping, Google Ads, and Facebook.  A detailed audit log of changes from sources and destinations is also available within bi(OS).  It also allows for a real-time³ and unified view of marketing metrics such as CTR (click-through rate), CPC (cost per click), and CPM (cost per impression) across Facebook and Google.  It was achieved by providing simple OAUTH access to bi(OS)’s MarTech pack⁴.  

This approach has the following advantages:  

• All ad network updates happen holistically within minutes of detecting change.
• The integration with ad networks is “push-based,” allowing for granular updates.
• All business rules⁵ are applied holistically across all sources.

We use this re-targeting approach, constituting 40%+ of our digital marketing budget.  For prospecting⁶, we have integrated a feed from our IT systems with bi(OS). bi(OS)’s MarTech pack combines these two product catalogs – the one generated by GTM and the other provided by our IT systems – and feeds them holistically to various Ad networks. 

Current State and Future Possibilities

For Google Display Network, we rolled out the bi(OS)’s MarTech pack across all products for four eCommerce properties⁷ across three countries.  Within Finland, where Google shopping is available, management of the entire product catalog has been migrated to bi(OS).  For Facebook, we have created a bi(OS) real-time feed for each of our five web properties covering ~4M+ products and are creating Ad Sets to run targeted campaigns.  Achieving this between June and November of 2022 has been a great achievement.  In all cases, bi(OS) helped the PHH group minimize the risk of false marketing across all channels.  We also saw a credible increase across all metrics – traffic quality, CTR, and conversion rates.

Within marketing, we plan to use bi(OS) for advanced ML/AI towards:  

• Optimizing the product catalog dynamically for campaign management
• Performing cross-channel optimization across Facebook, GDN remarketing, and Google Shopping.

We are also exploring how bi(OS) can help other departments at the PHH group for real-time personalization, product bundling, and near-real-time alerting.

Conclusion

I have desired real-time MarTech capabilities throughout my career and have always been met with – “IT can’t do it” or “It’s expensive”.  When I first met with Isima, I was sold on the “real-time feedback loop” vision.  Many naysayers said it couldn’t be done. Well, we have proven them wrong.  We put it into production within X weeks, with no added hardware, handling 4M+ product updates across Google Shopping, Facebook, and Google Ads. It has worked better than expected and even helped us discover data quality issues in our IT systems.   These solutions allow us to grow even under the shadow of war.  bi(OS) is how [real-time marketing] data platforms should be built, and I look forward to more eCommerce vanguards adopting bi(OS).

Sign up here to take bi(OS) for a test drive.