Home › Case studies › Amazon DSP Case Study: Branded Search and Revenue
Case study

Amazon DSP, branded search and revenue: a US bedding brand

Amazon DSP · March 2023 to February 2024 · Brand anonymized
The short version

Over 12 months, a US bedding brand's branded searches and revenue rose and fell with its Amazon DSP spend (correlation r ≈ 0.82). Each extra $1 of DSP spend went with about 0.067 more branded searches and about $10.73 more revenue.

The brand and the question

A US bedding brand selling on Amazon, in a crowded category, wanted to know whether its Amazon DSP budget was building awareness that showed up in sales. We used the account's own data from March 2023 to February 2024 and measured two things:

  • Branded searches on Amazon, as the measure of awareness: a shopper who types the brand's name already knows it.
  • Revenue: total sales in US dollars.

The brand is not named here. We never publish a client's name, logo or products.

How we measured it

We fitted two simple linear regressions on the account's data for the period: DSP spend against branded searches, and DSP spend against revenue. A regression draws the straight line that best fits the points and shows how much one number tends to move when the other does.

Scatter chart of DSP spend against branded searches, showing a positive correlation of about 0.82

DSP spend against branded searches: a positive correlation, r ≈ 0.82.

What we found: awareness

Branded searches moved closely with DSP spend: the correlation was strong, r ≈ 0.82. The fitted model puts each extra $1 of DSP spend at about 0.067 more branded searches, and predicts around 1,283 branded searches at $15,000 of spend.

What we found: revenue

Revenue moved with DSP spend too. The second model puts each extra $1 of DSP spend at about $10.73 more revenue, and predicts about $158,513 in revenue at $15,000 of spend.

How to read these numbers

A regression shows how closely two numbers moved together over a period. It does not prove that DSP alone caused the change: seasonality, sponsored ads, price changes and promotions can move spend, searches and sales at the same time. So we treat a result like this as a strong signal, then measure what DSP adds on its own with a holdout before scaling a budget on it.

To run that test on your own account, start with holdout testing on Amazon DSP and how to tell whether display is incremental.

What it meant for the brand

For this brand, DSP spend and branded demand moved together closely across the year, and revenue followed. Branded search was a useful early read on whether the upper-funnel spend was working.

Want the same measured on your account? We run Amazon DSP as a managed service and show the method before the number.

Book a DSP discovery call
Figures come from the client's own Amazon account data, March 2023 to February 2024, and are labelled with the period they cover. The client is anonymized at its request and under our policy: names, logos and products are never published.