The challenge

An online distributor that sells aesthetic medicine products only to clinics and licensed professionals. Its Performance Max campaign ran at 10–11x ROAS in 2025, but from April efficiency fell sharply and never recovered for long. The client needed a diagnosis and a clear structure going forward, with a target of 5–6x ROAS.

What I found

• April: feed and listing issues in Merchant Center. Conversion rate fell and ROAS dropped from 9.67x to 5.71x.
• June: the algorithm lost its signals. Purchases became the only primary goal and Add to cart was removed. Clicks spiked, conversion rate fell to 0.32% and spend drifted to Discover with no return.
• Budget was going to the wrong audience. Broad, low-intent searches were bringing in consumers who can’t buy from a B2B-only store.
• Where the value really comes from. Segmenting PMax by network showed that 94% of purchase value since June came through the Search Network.

What I recommended

• Restore signals: bring back Add to cart as a secondary conversion with a symbolic value.
• Protect learning: keep Maximize conversion value until the campaign holds 3x+ for a full month, then move to target ROAS and raise it gradually, about 10% every 10 days.
• Clean the traffic: recurring search-term reviews and negative keywords to cut expensive non-converting queries and keep spend on clinics and professionals.
• Adjust budget without breaking anything: a gradual 12% daily cut to offset earlier overspend, instead of a sharp drop.
• Escalate what needs a specialist: a technical review of attribution and sales tracking.

Testing the alternative

To check whether PMax was still the right structure, a Search and Standard Shopping test ran in parallel for a week. PMax was the only campaign that closed purchases, so investment stayed consolidated there. The test still gave a useful lesson for the next step:

Next decision

The client was very satisfied with how the account has stabilized. My recommendation for the next phase is to move to target ROAS, so the campaign stops drifting to cheap traffic that doesn’t convert. The client’s main concern is losing volume, so the plan is to start the target close to the ROAS the campaign already achieves (around 3x), raise it in small steps and track conversions and value week by week.

The question

Using Google’s public GA4 demo account for the Google Merchandise Store, I built a dashboard to answer three questions: which channels bring in revenue, where traffic is being wasted, and what the data can’t tell us yet.

Channel efficiency

Purchase rate = ecommerce purchases ÷ sessions. Site average: 1.0%.

What I found

• Paid Search is the least efficient channel. It brings 17% of the traffic but only 3.7% of revenue.
• Organic Search punches above its weight. 22% of the traffic generates 38% of revenue.
• Referral and Organic Social convert best (around 3%), but their volume is still very small.
• Traffic peaks don’t always mean sales. On 26 October sessions spiked with no matching rise in revenue.
• A data quality gap. Roughly a third of product revenue has no category (“not set”), so category decisions can’t be trusted yet.

What I’d recommend

• Audit Paid Search: search terms, targeting and landing pages, before adding budget.
• Invest in what already works: SEO content, and partnerships that grow referral traffic.
• Fix product tracking: make sure every item sends its category, then revisit the product analysis.
• Watch traffic quality, not just volume: investigate where the 26 October spike came from.

How I built it

Six KPIs at the top, a daily trend, then the detail by channel, product and device. One lesson along the way: GA4’s default channel grouping is attribution-based and only counts sessions with key events, so I switched to the session-scoped version to keep every total consistent with the KPIs.