The Challenge
Galeries Lafayette started scaling investment in Meta as part of this shift towards full- funnel acquisition. Early results supported the direction, but Last Click reporting captured only part of Meta's contribution. This made it difficult to justify a faster increase in investment and to understand its impact on acquisition performance at portfolio level.
The objective was therefore to put in place a measurement framework suited to a full-funnel strategy: one that could provide a more unified view across channels, help the team compare different performance signals and track how changes in Meta investment affected overall revenue, efficiency and customer acquisition. Fospha was selected to provide this cross-channel perspective alongside existing data from GA4, platform reporting, Last Click attribution and internal business data.
"This was a main pain for us. We have numbers from the platform and we have numbers from GA4 and it's a mess, so we have to find the sweet spot between both systems." — Emmanuel Guillo, Head of E-Business
For an enterprise retailer, the stakes go beyond the numbers. Aligning multiple decision-makers around a new trusted measure of performance - is itself a challenge, as is building organizational confidence in a new way of measuring. Earning that buy-in took a structured data validation stage, reconciling Fospha's outputs against the business's own data, and clear model explainers that made the methodology transparent - so the team understood exactly why the numbers could be trusted.
“I'm really glad with the Fospha support we're having, because the team always helps us to move forward and learn." — Regis Pennel, eCommerce Director
The Solution
Establishing a measurement framework for full-funnel investment
Fospha's measurement system gave Galeries Lafayette visibility into which channels, campaigns, and ads were actually driving revenue, letting the team optimize across the full mix. Galeries Lafayette used this view alongside GA4, Last Click, platform data and business KPIs to monitor the acquisition portfolio as a whole. This helped the team assess Meta beyond its directly attributed conversions and established a framework for confidently deciding how far and how quickly to scale investment.
Fospha’s measurement gave the team confidence to progressively increase Meta investment while monitoring blended revenue, channel efficiency and customer acquisition triangulated across multiple measurement sources. The rebalanced budget saw total spend increased by just 10% resulting in +32% blended revenue growth.
Acting on the insight
Insights extended beyond the Fospha dashboard's through Fospha's Model Context Protocol (MCP), enabling Galeries Lafayette to query data directly and pull answers on demand, before acting immediately on the insights surfaced. Stakeholders at every level, not just the analytics team, could ask questions ahead of meetings, planning sessions, or boardroom discussions, with measurement always at their fingertips. That accessibility drove adoption: Fospha stopped feeling like a separate platform to log into and became an always-available source of truth for daily decisions - and Galeries Lafayette soon became one of Fospha's heaviest MCP users.
"When Fospha first announced they were launching the MCP, I was very excited about the idea, because it helped provide quick summaries alongside the dashboards." — Regis Pennel
Geo-holdout validation
To validate the reallocation, Galeries Lafayette ran a controlled geo-holdout test - applying the new spend mix in one set of markets while holding a comparable set unchanged, then measuring the difference. to validate whether Fospha's model had accurately reflected real-world incremental return. The test isolated the impact of the spend change across comparable geographic markets and confirmed it: the new mix significantly outperformed the prior allocation. This validation proved the model, giving the team a reliable framework for future allocation decisions.
The Results:
The Results
During the 30-day comparison period, Galeries Lafayette recorded the following year-on-year outcomes across its blended paid-media portfolio:
- +32% blended revenue growth YoY, on +10% total spend increase
- +20% blended ROAS uplift YoY
- +20% increase in new customer conversions YoY, with CAC falling 8%
- 19% reduction in cost-per-paid-visit YoY, reflecting improved efficiency across the portfolio
Model validation
Meta ROAS remained stable as investment increased, indicating that the channel had not reached immediate saturation at the new spend level. Last Click aligned showing paid net revenue up 26% YoY and conversions up 22%. The directional consistency between Fospha’s outputs, Last Click data, business KPIs and the geo-holdout test validated the decision to scale and increased confidence that Meta could be scaled further while maintaining the overall efficiency of the acquisition portfolio.
"We had a working media mix, but we didn't have confidence that it was the right one. Fospha showed us where the real return was, and we acted on it. The geo-holdout test we ran afterwards validated the model, and that gives us the confidence to keep making decisions this way." — Regis Pennel
The Bottom Line
Galeries Lafayette's results demonstrate what becomes possible when measurement moves beyond what platform attribution alone can see. Fospha's incrementality model surfaced an efficiency opportunity invisible to Last Click reporting, the team acted on it, and a geo-holdout test run afterwards confirmed the model was right. The result was meaningful revenue and efficiency gains at minimal incremental spend, and a proven framework the team can apply to future allocation decisions.
Footnotes
- Impact type: Blended and Paid Media (including Google, Meta, TikTok, Bing).
- Time range & period comparison: Period-on-period analysis, 21 February – 21 March 2026 vs. 21 February – 21 March 2025, YoY.
- Markets: France (FR).
- Geo-holdout test design: 3 geographically comparable regions compared with 1 holdout and 2 test groups. Test groups contained different spend mixes. Groups were compared and concluded one test group displayed measurable uplift relative to holdout and other test group.
.jpeg)
