Hiatt Hardware: 87%+ in revenue over 18 months

Partnership Duration: 2 years and counting
Annual Budget: 7 figures
Channels: Google Ads, Microsoft Ads
Services: Paid Media Management, Conversion Rate Optimization, Data Modeling & Machine Learning (AI)
Location: UK

Overview & Audit

Hiatt Hardware came to our team looking to grow their paid media footprint. First, our team started with an audit to find any issues, quick wins and help us determine a long-term strategy. In the audit, we identified the following points:

• Fixed conversion tracking
• Added GAds tag and set GA4 tag to secondary
• Fixed and enabled dynamic remarketing
• Product feed optimisation
• Building audiences
• Identified 1 performance max campaign with 3 asset groups targeting all products (over 2000 SKUs)
• Found that 19% of products drove 81% of revenue.

We then had a discovery call to go over our findings and establish goals, which were to grow total revenue as much as possible while keeping to the blended ROAS target they’ve previously set.

Launching & Optimising

Once the foundational groundwork was completed and goals were set, like optimising the feed alongside the client, we started working on the account. First, we moved away from 1 performance max campaign targeting all products with three asset groups to a category based campaign structure with all relevant asset groups needed per category.

We excluded branded terms from all the new and old shopping/pmax campaigns. We did this to start collecting more relevant, accurate data, like brand versus non-brand and product category performance, and to give more overall exposure to products. We then slowly shifted over daily budget and campaign targets not to shock the account and switch the account into the new structure.

Data-Driven Account Structure

After 90 days of data, we pulled out all the data and compared click-through rates, cost per clicks, conversion rates, return on ad spend and more on the product level. Using this information, we rebuilt the account again utilising a performance based structure. This included:

• Top performers: products at or above the “best” ROAS target
• Medium performers: products between the best and benchmark ROAS target
• Low performers: products below the benchmark target
• Bin: products that spend but are way below benchmark
• No exposure: products that don’t get impressions

Following the same concept as above, we moved the account over a four week period from the category based structure to the performance based structure, to ensure business continuity. The above structure enabled us to spend budget where it was most effective, while working on and analysing performance of the products which we’re performing.

Dynamic, Feed-based Search Ads

In the background, we also started to launch feed-based search ads. These use dynamic data from the feed that enabled us to include price, stock and product descriptions alongside our ad copy, to create highly-targeted search ads. Since starting with us in 2023 to the end of the year, results were as follows:

Key Results (2023)

• 26%+ in CPC YoY
• 4% above blended ROAS target
• 263% increase in total revenue
• 7% increase in average order value

Testing New Products on Shopping

With 2023 behind us and as we entered 2024, we looked at our overall performance and realised that with the release of new products and a growing inventory, not all of the new SKUs were getting impressions. Our winners went from 19% to over 31%, with many more products in the above benchmark range. That said, over 50% of the products were still getting low impressions. We countered this by creating something we called product warmers – campaigns with no targets that spend on a group of products that have gotten low or no impressions. This was to force the system to spend on underserved products so we could re-categorise them and push more of the overall inventory. This looked as follows:

• Campaign 1: Products with 0 to 25 clicks
• Campaign 2: Products with 25 to 49 clicks
• Campaign 3: Products with 50+ clicks

This helped us push more of the inventory during low season, finding 6% more products that were categorised into Top or Medium performers, which helped us weather the lower demand and increasing YoY CPCs. To be exact, we’re paying 36% more per click compared to 2023. In any case, when we compared YTD figures for 2024, we landed on the below:

Key Results (2024 YTD)

• 36%+ in CPC YoY
• 7% decrease but above target
• 97% increase in total revenue
• 11%+ in average order value

Expanding Their Product Footprint

Given the client’s ambitious goals, we needed to re-evaluate our growth strategy. By June, 500 new products were added to the line-up. To effectively categorise these new products, we utilised the campaign warmers warmers. At this stage, we decided to upgrade our data and performance analysis methods. Working closely with a web development company and a statistician, we incorporated data clustering and machine learning techniques to develop historical models of the account, aiming to identify the most efficient structure. We created three distinct product groups, categorising them as outlined below:

• Top Performers
• Medium Performers
• Low Performers
• Bin
• Performance Warmers

Furthermore, there is a detailed subdivision within these product groups based on various statistical models that are fully customised for Hiatt.

Conclusion & Growth Steps

This collaboration is helping the client invest their money more efficiently. By leveraging advanced statistical methods and AI-driven analysis, facilitated by our partnership with web development experts and a professional statistician, we are able to more accurately classify and manage our product offerings, enhancing our ability to optimise the client’s investment and achieve their objectives.

Most importantly, throughout the whole process, Hiatt’s team have been very involved with all strategy decisions and have suggested different approaches that helped us grow. The overall closeness of our partnership has greatly supported our team, which made this all possible.

We’ll be running this for the next two months to measure performance as we head into peak season, and will add to this case study (or create a version two) following on from this. Stay tuned!