The Challenge

The client, a leading network of UK holiday parks, sought to better understand the incremental value of their marketing efforts in order to optimise their advertising budget.

Despite a constant, year-round advertising presence, running both performance marketing and off-line communications, there was a lack of confidence in how to allocate spend effectively, with budget decisions based on anecdotal “sources of truth” such as historical performance, live trading figures, and Google Analytics last-interaction reports.

Furthermore, fragmented channel objectives and KPIs caused inconsistencies in their overall media strategy.

The Solution

The client needed a unified measurement and reporting infrastructure that enabled them to confidently set channel level budgets and swiftly adjust them to meet weekly and monthly booking targets.

We began by conducting a comprehensive data & martech audit to identify gaps and inconsistencies in current data capture across both on-line and offline channels. From there, we developed a robust measurement framework powered by Meta’s Robyn, which was combined with geo-targeted incrementality testing for further validation.

This allowed us to accurately assess the incremental value of each marketing channel. Additionally, we introduced a unified set of KPIs for reporting overall media performance and used budget optimisation to optimise budget allocation across channels based on robust data-driven insights.

The Impact

Following implementation, the client saw significant improvements across key performance metrics. Overall budget wastage was reduced by 20%, customer acquisition cost (CAC) decreased by 15%, and return on ad spend ROAS improved by 17%.

These results were achieved through a more strategic and data-driven approach to setting and adjusting channel spend, grounded in validated incrementality rather than anecdotal evidence.

Key Takeaways

• Data and MarTech Audit - A thorough audit identified gaps, leading to the development of a precise measurement framework combining MMM and geo-targeted tests.

• Unified KPIs - We established consistent KPIs across channels, eliminating fragmented objectives and providing a clearer picture of media performance.

• Scenario Planning - A custom-built scenario planner allowed for optimised budget allocation, reducing waste and improving return.

• Business Results - Optimised channel spend led to a 20% reduction in budget wastage, a 15% lower CAC, and a 17% improvement in ROAS.

Tools and Techniques

• Marketing mix modelling
• Geo-based incrementality testing
• Data & martech audit
• Budget optimisation for bespoke scenario planning
  
• SQL - data extraction and transformation
• R - modelling
• Snowflake - data warehousing
• Tableau - reporting

Related services

The Challenge

The client, a large international online recruitment platform, had built a strong model in its home market, especially the US, and tried to apply the same media strategy and optimisation logic across the British Isles.

Despite sustained investment, the UK and Ireland business could not match the same levels of efficiency or growth. Spend was generating top-of-funnel volume, but not enough high-quality applications or consistent employer acquisition. Differences in labour market dynamics, hiring seasonality, candidate behaviour, and media consumption meant the US playbook did not land as expected.

Because the business depended on both applicant quality and employer-side outcomes, the client needed a clearer view of what actually drove value in the British Isles before rising acquisition costs put more pressure on spend.

The Solution

We ran a full marketing mix modelling engagement focused on identifying the true drivers of performance in the British Isles. The framework measured impact across the applicant funnel, from registrations and completed applications through to employer sign-ups, while controlling for seasonal hiring patterns, economic conditions, competitor activity, job vacancy trends, inflation, and hiring confidence.

The work quickly showed that US assumptions did not hold. While the US model leaned heavily on performance search and high-frequency retargeting, the British Isles market responded more to strongly to brand-led comms during key seasonal windows (e.g. The New Year Peak & The Autumn Surge job switches). Employer-side outcomes were also more sensitive to perceived platform quality, making visible presence in trusted channels more influential than expected.

Our recommendations centred on a rebalanced media mix, improved demographic targeting by industry vertical, and changes in campaign timing to better reflect local hiring seasonality.

The Impact

The new strategy delivered clear performance gains. CPA for qualified applicants fell by 20% within the first quarter, helped by reallocating budget away from over-saturated low-impact channels and into more responsive formats.

Employer sign-ups increased by 15%, while ROI improved by 18% as stronger candidate quality and more visible brand presence built confidence during key hiring periods.

Beyond the numbers, the engagement gave the client a market-specific blueprint for growth in the British Isles, grounded in the behaviours and signals that mattered to British Isles users and actually drove revenue.

Key Takeaways

• Local market dynamics mattered – US-led strategies did not transfer cleanly to the British Isles.

• MMM clarified real drivers – The model showed what actually influenced qualified applications and employer-side value.

• Budget reallocation improved efficiency – Spend moved towards more responsive channels and local hiring cycles.

• A market-specific view improved decisions – The client gained a stronger basis for planning and future market localisation.

Tools and Techniques

• KPI framework spanning job seeker and employer outcomes
• Marketing mix modelling
• Market-specific budget optimisation and campaign timing
• Data & martech audit
• SQL - data extraction and transformation
• R - modelling
• Snowflake - data warehousing

Related services

The Challenge

The client, a US-based language learning app expanding across Europe, had built a strong performance engine but struggled to justify investment in upper-funnel activity such as Video and CTV.

Platform reports showed reach, view-through, and attributed conversions, yet the commercial team could not confidently link that spend to incremental sign-ups, especially with seasonality and ongoing performance marketing in the mix.

The client needed a reproducible way to prove the true incremental value of Video and CTV and reduce budget waste as investment scaled.

The Solution

We ran geo experiments across the client’s three largest European markets, increasing CTV weight in treatment regions while keeping control regions business-as-usual.

The design used three windows: a 20-week pre-test period to establish baseline behaviour, an 8-week intervention period where spend changed, and a 6-week cooldown period to capture the delayed conversion response that often follows video exposure.

We also ensured that major confounders, including pricing, promo cadence, and ongoing performance marketing, remained stable so the readout reflected the effect of increased CTV. Success was measured through incremental paid sign-ups and first-month revenue at country level.

The Impact

Treatment regions delivered +3.1% incremental sign-ups versus control, with a further +1.4% during the cooldown period, bringing total measured lift to +4.5% across the full test readout. That matched how the category typically converts after video exposure.

Incremental CPA came out at €24, versus a blended paid-search CPA of €19 in the same period. That gap was expected, because CTV was not competing on last-click efficiency. It was creating incremental demand that later converted through other channels.

The client kept CTV in the plan, but with clearer deployment rules, using geo-based lift rather than platform reporting as the proof point and scheduling repeat tests each quarter as creative and inventory changed.

Key Takeaways

• Upper-funnel Video and CTV are hard to value using platform reports alone – Impact is often delayed and diffused across channels.
• Geo experiments can prove true incremental value using first-party data – Especially for upper-funnel activity with longer pay-off windows.
• A carefully structured geo-based experiment improves interpretability – It captures both immediate and delayed conversion response.
• The real output is a decision rule – Where spend creates incremental sign-ups, how far it scales, and where marginal returns flatten.

Tools and Techniques

• Geo-lift incrementality testing
• Data & martech audit
• Matched treatment-control region design
• Difference-in-differences lift estimation
• SQL - data extraction and transformation
• R - modelling
• BigQuery - data warehousing

Related services

The Challenge

The client, a UK-based ferry operator serving routes between Great Britain, Ireland, and mainland Europe, wanted to improve the efficiency of its marketing spend across services with very different demand patterns.

Budget decisions had historically been made at a national level, with limited visibility into how performance varied by route, season, or departure port. Offline media, especially Regional Press and Out-of-Home, was concentrated around ports such as Dover, Liverpool, and Calais, based on legacy assumptions about traveller proximity.

With operational costs rising and price sensitivity increasing, the business needed a clearer view of how marketing investment translated into bookings, revenue, and load factor at route level, and where spend could be redeployed more profitably.

The Solution

We implemented a marketing mix modelling framework to measure the incremental impact of both online and offline media across key routes, while accounting for seasonality, pricing, and external market forces.

The model combined media inputs such as TV, print, paid search, paid social, discounts, and promotional pricing with control variables including fuel prices, GBP–EUR exchange rates, weather, and macroeconomic indicators like inflation. Operational costs were also built in, allowing performance to be assessed against breakeven load factor thresholds rather than media metrics alone.

A detailed geographic analysis of booking data showed that offline spend was overly concentrated around ports, while bookings were actually clustered around major urban centres such as London, Birmingham, Manchester, and Leeds. These insights supported a strategic redeployment of Regional Press and OOH into urban catchments aligned to route-specific demand patterns. We then used scenario planning and budget optimisation to determine the minimum investment needed to sustain load factor above 67% across routes with different seasonal profiles.

The Impact

The revised media strategy improved both efficiency and commercial performance. After some offline media was reallocated into urban centres, passenger and vehicle bookings from Regional Press and OOH increased by 14%, with a corresponding 11% uplift in revenue attributable to those channels.

At total media level, marketing ROI improved by 19% as spend moved away from low-return placements and towards routes and periods with stronger marginal returns. Average customer acquisition cost fell by 12%, and budget efficiency improved enough to keep load factors above the 67% breakeven threshold for nine months of the year, up from six previously.

Route-level optimisation proved especially valuable in shoulder seasons, where differences in price sensitivity and demand elasticity helped the client reduce unnecessary spend on highly seasonal routes while sustaining profitability on year-round services.

Key Takeaways

• Route-level MMM – Revealed meaningful differences in media effectiveness that national-level reporting was hiding.

• Geographic booking analysis – Challenged long-held assumptions about proximity-based offline media placement.

• Commercial ROI framing – Incorporating operational costs reframed ROI around load factor breakeven, improving decision-making.

• Budget optimisation – Aligned marketing investment more closely with seasonality and route-specific demand patterns.

Tools and Techniques

• Geographic booking analysis and catchment mapping
• Marketing Mix Modelling
• Budget optimisation and scenario planning
• Data & martech audit
• SQL - data extraction and transformation
• R - modelling
• Snowflake - data warehousing

Related services

The Challenge

The client, one of the UK’s largest supermarket chains, needed a more actionable measurement framework to understand the impact of its marketing activity.

Its incumbent partner had built a technically impressive store-level modelling suite, but it was rarely used in practice because media planning was not done at store level. As a result, the outputs were difficult to activate and had little influence on planning.

The brief was to design a more strategic solution that retained analytical rigour, could be refreshed quarterly, and aligned with how the business actually made decisions.

The Solution

To ensure long-term relevance, we designed a marketing mix modelling suite structured around the client’s internal Commercial Planning Divisions. That meant Fresh Produce, Chilled & Frozen, Groceries & Dry Goods, Household Items & Clothing, Health & Beauty, and General Merchandise each had access to insights that reflected their own trading patterns, seasonality profiles, and promotional cycles, while still supporting a joined-up view of total marketing performance.

We also built in automated quarterly refreshes so analysts could update models and rerun outputs with minimal friction. Broader business initiatives, including the client’s Price Match programme, were incorporated so the framework aligned more closely with internal commercial analysis.

The result was a measurement framework that fit how the business actually planned, making outputs easier to trust, use, and maintain whilst keeping high technical and analytical rigour.

The Impact

Within six months of deployment, the new MMM framework was already influencing media planning across three divisions. Marketing teams were able to shift investment using clearer incremental ROI estimates, improving allocation across both traditional and digital channels.

Overall ROAS remained in line with the client’s previous modelling suite, but the new structure made reporting far more actionable. Aligning the framework to Commercial Planning Divisions drove a 21% improvement in overall budget efficiency, measured through internal ROI benchmarks and externally validated econometric outputs.

Crucially, because the system matched how commercial teams already structured planning and forecasting, insights were easier to embed into day-to-day decision-making.

Key Takeaways

• Division-aligned MMM – Structuring outputs by Commercial Planning Division increased relevance and adoption across the business.

• Reduced complexity – Moving away from store-level modelling improved clarity and made the framework easier to use.

• Regular refresh cycles – Built-in updates made the solution more sustainable for the client’s analytics team.

• Commercial fit – Incorporating initiatives like Price Match strengthened credibility and stakeholder trust.

Tools and Techniques

• Data mapping to commercial planning divisions
• Marketing mix modelling
• SQL - data extraction and transformation
• R - modelling
• Snowflake - data warehousing

Related services

The Challenge

The client, a fast-growing EU home-cleaning brand selling primarily through a major online retailer, had scaled sponsored products and on-site placements aggressively. Platform ROAS looked strong, but the commercial team suspected that much of this spend was simply shifting credit from organic and repeat buyers.

In marketplace environments, sellers can end up paying for conversions they would have happened anyway, creating a cannibalisation effect rather than genuine growth.

The client needed a causal answer on the true value of sponsored products and premium placements, and a practical way to stop waste as spend scaled.

The Solution

We designed a matched product-group holdout inside the retailer account, paired with a regional split where feasible. One set of products remained fully exposed to sponsored products and premium placements, while the matched holdout group had those placements suppressed for a 4-week test window.

Because this was a holdout design, the goal was to measure what happened when paid retail media was withdrawn from comparable products. That gave us a clearer read on how much of the reported platform performance reflected real incremental demand versus displaced organic or repeat demand.

We also segmented results by product maturity and customer type, allowing us to separate displacement on high-ranking, high-repeat SKUs from genuine incremental contribution on more competitive terms and products with room to expand reach.

The Impact

The holdout showed clear cannibalisation at the current spend level. While attributed orders remained strong, total sales barely moved in the exposed group relative to holdout, indicating that a large share of paid conversions were replacing organic demand rather than adding net growth.

Across the test window, Sponsored Products and placements drove +16% attributed orders but only +1.1% lift in total units, implying roughly 70% cannibalisation at current spend levels. The story diverged by segment: mature hero SKUs showed close to flat total lift, while mid-ranking products on competitive queries delivered +4.2% total-sales lift and +19% new-to-brand orders.

As a result, the client introduced guardrails for retail media, capping spend once products hit saturation thresholds and reallocating budget towards segments that showed measurable total lift and new-to-brand growth. Reporting also shifted from platform-attributed ROAS to incrementality-led KPIs.

Key Takeaways

• Retail media can look strong in-platform while adding little net growth – Especially at high saturation, where paid demand substitutes for organic demand.

• Matched product-group holdouts provide a practical causal read – They measure total outcomes rather than relying on attributed conversions alone.

• The commercial unlock is guardrails – Budget should concentrate where it grows category or new-to-brand demand, and pull back where it mainly shifts credit.

Tools and Techniques

• Matched product-group holdout testing with placement suppression
• Data & martech audit
• Segmentation by product maturity and new-to-brand
• SQL - data extraction and transformation
• R - modelling
• BigQuery - data warehousing

Related services

The Challenge

The client, a fast-growing SaaS scale-up offering generative AI tools to automate customer interactions, saw strong traction following the launch of a new AI Agent in late 2023. In response, the business increased prices by 15% and raised marketing investment by 50% to boost acquisition through their 4-week product trial.

By mid 2024, performance lagged expectations. Most of the marketing budget was allocated to Paid Search (80%+) and the lack of value-led communication to support price increase sharply reduced trial-to-sign-up conversion. With brand recognition already low due to limited brand-led comms, performance marketing alone was no longer enough to sustain growth.

With low brand equity translating into weak mental availability for the brand, and media investment nearing saturation, the business needed a new way to measure what would actually drive growth.

The Solution

The engagement began with a comprehensive data & martech audit to confirm that data capture, tracking, and infrastructure were fit for advanced marketing analytics. We then implemented a combined measurement framework using marketing mix modelling and geo-targeted incrementality testing to assess whether the current spend mix could support sustainable growth.

Preliminary MMM showed that simply increasing spend in high-intent channels was not enough to offset the price increase without stronger brand support.. Paid Search, SEO, and Performance Max remained strong contributors but had reached saturation. Meanwhile, mid- and upper-funnel channels accounted for less than 6% of spend, were inconsistently used, and were statistically weak within the model.

To overcome these limitations, we ran geo-targeted incrementality tests across underused channels such as Facebook, LinkedIn, and Digital Display, using existing creatives to control costs. The results were fed back into the model to calibrate optimisation and scenario planning, giving the business the confidence to break out of the performance-marketing doom loop and reallocate spend towards a more balanced brand + activation strategy.

The Impact

Incrementality tests were run sequentially to isolate effects cleanly, with each channel phased into the media mix once results were validated.

Over a six-month period, branded impressions increased by 38%, signalling a measurable uplift in brand visibility when benchmarked against key competitors. In parallel, trial-to-sign-up conversion improved by 24%, reversing the post-price-increase decline and restoring confidence in the acquisition funnel.

Paid Search remained stable in absolute spend but declined as a share of total investment. The client now operates a fully integrated multichannel communication plan combining brand-building and performance activity, aligned to both short-term efficiency and long-term growth objectives.

Key Takeaways

• Brand investment was critical to restoring conversion performance following a price increase.

• Marketing Mix Models alone were insufficient given historic lower-funnel bias and required supplementing with testing.

• Geo-targeted Incrementality Tests unlocked scalable mid- and upper-funnel investment decisions.

• Rebalancing spend reduced over-reliance on Paid Search while improving overall acquisition resilience.

Tools and Techniques

• Data & martech audit
• Marketing mix modelling
• Incrementality testing
• Budget optimisation
• SQL - data extraction and transformation
• R - modelling
• BigQuery - data warehousing

Related services

The Challenge

The client, a UK-based financial services provider specialising in insurance products for the over-50s, wanted to diversify their customer acquisition strategy and attract younger prospects within their target market.

Historically reliant on traditional channels such as Direct Mailing and Daytime TV, they struggled to recruit younger customers (aged 50-65) and feared digital channels would not yield a strong return given their core audience, whose average age was 78.

The client was sceptical of digital advertising like Display, Social Media, as well as Prime Time TV (7 pm - 10 pm), viewing them as potential budgetary risks.

The Solution

We implemented a comprehensive analytics suite leveraging marketing mix modelling to quantify the incremental value of the client’s historical media mix.

The customer base was segmented by age (50-65 and 65+), gender (male and female), and acquisition channel (phone and web), resulting in eight distinct customer groups. This segmentation revealed where performance differed most across audience groups. For example, Paid Search proved particularly effective for attracting younger customer cohorts aged 50-65, delivering nearly a third more conversions than for the 65+ segment.

Reassured by these findings, the client was encouraged to explore untapped digital channels.

The Impact

Following the MMM analysis, the client approved budget allocation for four incrementality tests: Prime Time TV (7 pm - 10 pm), Brand Search, Facebook, and Instagram. Each channel was phased into the media mix over time as test results were validated and signed off.

The outcomes were strong: the client saw a 19% reduction in budget wastage, a 22% decrease in customer acquisition costs (CAC), and a 27% improvement in return on ad spend (ROAS).

These results showed that digital channels could reach younger audiences effectively without weakening performance among the client’s core older demographic.

Key Takeaways

• Marketing Mix Modelling - Offered clarity on relative effectiveness of traditional channels and potential of digital channels.

• Segmentation Insights - Revealed that Paid Search disproportionately attracted younger customers, justifying a broader digital push.

• Incrementality Testing - Helped validate new channels before spend was scaled.

• Business Results – Lowered wastage, reduced CAC, and improved ROAS while expanding digital acquisition.

Tools and Techniques

• Marketing mix modelling
• Geo-based incrementality testing
• Customer segmentation
• SQL - data extraction and transformation
• R - modelling
• BigQuery - data warehousing
• Tableau - reporting

Related services

The Challenge

The client, an EU sports equipment start-up, relied on branded search to capture high-intent demand. Organic rankings were strong, and none of the obvious direct competitors appeared to be bidding on its brand terms.

Although brand search looked like having low incremental value, pausing it could hit revenue immediately. The client needed to understand the true incremental value of brand search before making a decision that could damage their bottom line.

The Solution

To measure the true value of brand search, we ran a geo holdout test with spend suppression across three key markets. Regions were split into treatment and control groups, balanced on historic revenue levels and dependence on branded search. In the treatment regions, branded search activity was paused for 6 weeks, while the control regions were left unchanged.

Because this was a holdout design, the goal was to measure the commercial loss caused by removing coverage in matched regions. Outcomes were assessed using sales and revenue deltas through a difference-in-differences approach, allowing us to estimate how much value branded search had really been protecting.

We paired the test with auction monitoring to see who filled the gap once branded coverage was withdrawn. That mattered because pressure in these auctions often comes not just from direct competitors, but also from affiliates, resellers, marketplaces, and smaller niche players.

The Impact

The holdout produced a clear commercial result. Treatment regions saw a -6.8% revenue delta versus control over the test window. Paid brand clicks fell by 88%, organic brand clicks rose by 29%, and overall brand-query sessions were down 21%, showing that paid coverage was protecting demand rather than simply replacing organic traffic.

Auction monitoring revealed that affiliates and smaller competitors were active throughout the test period. As branded coverage was reduced, the client’s top-of-page presence on brand queries fell from 96% to 54%, while non-brand impression share rose from 14% to 37%.

The client retained branded search as a protective channel and used the holdout result to tighten bids and coverage around the highest-leakage terms. Over the following month, incremental ROAS improved from 3.2 to 4.1.

Key Takeaways

• Brand search incrementality is context-dependent – Even with strong organic rank and no obvious direct competitor bidding, branded search can still protect meaningful revenue when affiliates and smaller players are active.

• Competitive pressure is often indirect – Resellers, marketplaces, affiliates, and niche retailers can all siphon demand when paid branded coverage is removed.

• Holdout testing works best with auction monitoring – Pairing the two helps explain the mechanism behind any observed gain or loss and makes the result easier to act on.

Tools and Techniques

• Geo-holdout incrementality testing with spend suppression
• Data & martech audit
• Auction and bidder monitoring
• SQL - data extraction and transformation
• R - modelling
• Snowflake - data warehousing

Related services