Skip to content
uclic
FRAudit
Confidential client5 weeks (ongoing)Google Ads

E-commerce: from 2 to 11 Google Ads sales in 5 weeks

Taking over a Shopping account with no signal: measure on real purchases, cut what doesn't sell, refocus budget on the only lever that converts.

Get the same diagnosis
Key figures

Key results

-88%

Cost per purchase

+450%

Purchases (2 → 11)

-63%

Shopping CPC

×2

Shopping click-through rate

5weeks

Measured engagement length

01

Context

An online retailer selling a premium, high-ticket product asks Uclic, in autumn 2026, to take over its Google Ads account: a single Shopping campaign and no usable measured sales.

From November 2025 to April 2026, the account had spent about €3,100 for 2 purchases, a cost per purchase of about €1,550, far above the break-even threshold agreed with the client. Shopping paid €1.04 per click at a click-through rate of about 0.94%, no campaign covered the brand, and bidding had no purchase signal to learn from.

02

Objective

The goal is not yet a profitable ROAS, but to move the account from "no signal" to an account that sells.

  • Get measured sales on an account that barely produced any.
  • Measure on real purchases, sent server-side from the store, not on cart steps.
  • Get below the break-even cost per purchase agreed with the client, without raising the budget.
03

The Uclic approach

Purchase measurement first, then brand coverage, decisive cuts and a scoped expansion of what sells.

  1. 01

    Measure on real purchases

    Server-side purchase sent from the store, set as the only primary conversion. Add-to-carts and product views moved to secondary.
  2. 02

    Cover the brand, test competitors

    Dedicated Brand Search campaign (€10 a day, capped CPC), then a Competitors campaign after analysing search volumes and bids per brand.
  3. 03

    Refocus budget on what sells

    After the first Shopping sale, the generic Search budget was cut and the Shopping budget was doubled. A week later, the generic and Competitors Search campaigns, with no sales, were paused.
  4. 04

    Clean up Shopping queries

    48 negative keywords (competing brands, foreign languages, entry-level accessories), on queries that accounted for about 25% of the week's spend. Brand queries kept: they carried the sales.
  5. 05

    Expand with a scoped Performance Max

    PMax limited to the main product, optimised on purchases only, brand excluded. After 5 days, 32 negatives added: nearly half of visible spend went to entry-level accessories.
04

Results

Same conversion over both periods: the server-side purchase. Limitations: only 5 weeks and 11 purchases, periods neither contiguous nor of equal length, and a simultaneous increase in Meta campaigns that feeds brand searches. The reported value understates actual revenue. Ongoing engagement.

Google Ads accountBefore the takeoverWith UclicChange
Spend≈ €3,100≈ €2,100−33%
Clicks≈ 3,000≈ 3,700+23%
Server-side purchases211+450%
Cost per purchase≈ €1,550≈ €190−88%
Shopping CPC€1.04€0.39−63%
Shopping click-through rate≈ 0.94%≈ 1.95%×2
  • 11 server-side purchases in 5 weeks, versus 2 in almost 5 months of spend before the takeover.
  • Cost per purchase −88% (≈ €1,550 → ≈ €190), below the break-even threshold agreed with the client.
  • Shopping CPC −63% (€1.04 → €0.39) and Shopping click-through rate ×2 (≈ 0.94% → ≈ 1.95%).
  • 711 changes made to the account in 5 weeks; October, the first full month, accounts for 10 of the 11 purchases.
05

Key takeaways

  • An account with no sales isn't optimised, it's rebuilt: without a purchase signal, the priority is the first measured sales, before chasing ROAS.
  • Refocus on the only lever that sells: on a limited budget, cutting campaigns with no results frees up spend for the one that converts.
  • Read search terms before expanding: the first Shopping sales came from brand queries. Negatives and a scoped PMax aim to make generic demand convert too.
  • Measure server-side: a purchase sent from the store is more reliable than a browser conversion, and cart steps are never counted as sales.

Limitations: only 5 weeks of work and low volumes (11 purchases). The two periods compared are neither contiguous nor of equal length. The client's Meta campaigns grew sharply over the same period, which also feeds brand searches: part of the increase in Google Ads sales probably comes from them. Finally, the reported value understates actual revenue, so the ROAS shown in Google Ads is not meaningful here. Ongoing engagement.

All case studies
Related case studies

A similar challenge

All case studies
DeepkiEnergy / ESG real estate

B2B Acquisition in the Energy Efficiency Sector

Deepki aimed for a 3% conversion rate on its paid campaigns + an outbound structure targeting 10,000 QHSE decision-makers in companies with 250+ employees and €50M+ turnover.

10000+

QHSE/CSR Decision-Makers Targeted

3%

Paid Conversion Rate

Read the case study
ConfidentialGoogle Ads

Self-storage: −36% cost per lead on 2 Search campaigns, in 30 days

Takeover of a Google Ads account (Search + Performance Max) for a self-storage operator. By cutting ad groups with no signal and lowering target CPAs in steps, the cost per qualified lead of the two drifting campaigns fell 36% in 30 days (−11% across the account), with 12% more qualified leads.

-36%

Cost per lead, 2 campaigns

+12%

Qualified leads, 2 campaigns

Read the case study
ANSSIPublic sector / Cybersecurity

ANSSI — MesServicesCyber: A Data-Driven SEO & GEO Strategy to Make State Cyber Expertise Visible

The MesServicesCyber portal (ANSSI) rendered most of its content invisible to search engines (100% JS rendering). Uclic conducted the technical audit, then built a data-driven SEO/GEO content strategy: 8 clusters, validated by real demand, a 12-week calendar, and 29 answer-first briefs.

8

Semantic clusters mapped

29

Answer-first article briefs

Read the case study
Last step

You do not need more channels.
You need someone flying the plane.

Free audit · 48hSee the Uclic deck

FreeResults in 48hNo commitment

06 17 12 54 285.0Google4.96Sortlist4.3Trustpilot40+ B2B clients