Back

Equilia Group's +11% growth with zero additional investment

Category: 2. ANDMED / 2.1 Andmekasutus: kvantitatiivne

Submitter (agency): Analytical Alley

Koostööpartnerid: Equilia Group

Kampaania turud: Latvia

Töö lühikirjeldus (eesti keeles): Equilia on Läti ettevõtte, mille all opereerib 5 krediidibrändi, mis tegutsevad samal turul ning konkureerivad sama eelarve nimel. Analytical Alley modelleeris meedia efektiivsust nii grupi kui ka brändi tasandil. Optimeerimise tulemusel kasvas kvalifitseeritud laenutaotluste arv 11% sama investeeringuga. Hierarhilise süsteemi tootlus oli 30% kõrgem kui tavapärane, portfellipõhine, mudeldamine.

Short description (in English): Equilia Group hired Analytical Alley to optimise media investments across 3 brands competing within the same sector and budget. Sharing a single budget, Analytical Alley modelled media performance at both group and brand levels. The resulting budget reallocation increased total qualified loan applications by 11% at zero extra cost, yielding 30% higher relative gain than standard portfolio modelling.

1. Töö eesmärk (kuni 2000 tm): Equilia’s goal was to generate more qualified loan applications and annual revenue while keeping their media investments the same as during the previous year. All three brands, Sefinance, Moneza and Latkredits had already shown growth trends, and their qualified applications grew by 26% YoY. However, during this period, media investments across all channels increased as well. Analytical Alley was hired to identify the incremental effect from paid media and give clear strategic suggestions about what each of the three brands can and should improve in their media strategies. We created econometric models to isolate the incremental impact of paid media from other business factors. We didn’t take the easy way out; we aimed as far as identifying competitor’s actions, macro factors and other variables relevant to the geography and sector. To make our data modelling applicable in commercial context, we created benchmarks across all Equilia’s brands and marketing channels for optimal investment levels.

2. Sihtrühm, insight ja strateegia (kuni 2000 tm): Although all three Equilia brands target different segments, they all compete for the same local Latvian audience and share one budget. It is well known that the consumer credit category is heavily driven by paid search, and most competitors focus on high-intent channels, mainly bottom-funnel activities. Key Findings Analytical Alley created a hierarchical Marketing Mix Modelling system to give Equilia a clear overview of their marketing investments. Our model showed that marketing activities drove a large share, 41% of Equilia’s qualified loan applications. Incremental paid media impact on business was consistent across brands, ranging from 40 to 48%. However, marketing channel efficiency varied significantly across brands. For example, average radio efficiency was nearly 60% higher for one brand compared to another. In other words, a €1,000 spend on radio would generate 60% more approved applications for Brand A than for Brand B. We also saw a clear gap in efficiency of performance marketing channels across brands. One of Equilia’s brands had a substantially more effective P-Max campaign than the other brands thanks to the synergy with their ATL media investments and existing power of the brand. It was also interesting that our data models found that competitor paid media activities had a clearly distinguishable impact on Equilia Group’s applications. Notably, 76% of this impact came from one competitor's campaigns helping us to pinpoint exactly to which competitor to keep in mind when planning Equilia’s media strategy.

3. Teostus (kuni 2000 tm): We needed to build a hierarchical system with a portfolio model, on channel level, alongside brand-level models to maximise channel efficiency, long-term media effect, and optimal spending levels. We gathered media spend and delivery metrics (daily investments accompanied with TRPs/GRPs, impressions, clicks, etc.), primary business data (daily qualified applications), and control variables (local public and shopping holidays, historical weather data, and macro indicators). All training data relied on actual historical figures. Portfolio Model: We built a unified model analyzing the impact of each channel on total applications (the daily sum of all three brands). Three Independent Brand Models: We created a dedicated model for each brand to evaluate how each channel impacted that specific brand's applications. Optimisation & Delivery: A unified budget allocation with optimal investment levels, seasonal timings and flighting, for each channel within each brand. All to maximise qualified applications for the entire group.

4. Tulemused (kuni 2000 tm): The group model helped us find the most effective channels, highlighting those with the highest contribution, growth opportunity, or lower cost per unit (CPU). However, to determine the exact budget for each channel under each brand, we needed dedicated brand models tracking individual investments and their effect on qualified applications. As a result, we achieved 11% higher paid media return with zero additional budget. The modeling resulted in a new optimal budget for every brand and each marketing channel. We provided detailed recommendations such as “increase Brand A radio investment by 15%” or “reduce Brand B’s YouTube investment by 20%”. As a result, with the exact same media budget, we increased total approved applications by 11%. At the brand level, media returns improved by 10.3% for Brand A, 11.1% for Brand B, and 12.5% for Brand C. Across the group, the cost per approved application dropped by 10%. Using exclusively a portfolio model would have resulted in media contributions growing by 8.4%. Adding brand-level models lifted that growth to 11%, resulting in a 30% higher relative gain from media. Model Accuracy Reliable media optimisation requires high accuracy; our model achieved an extremely high accuracy, 96.5%. To achieve such a high rate, we included a large scope of factors. Alongside media investments activities, we included multiple control variables such as baseline demand, seasonality, weather, holidays, and macroeconomic trends. Reliable media optimisation requires high accuracy; including these variables, brought model accuracy to 96.5%. This means the model predicted monthly approved applications with an error rate below 3.5% over the analysed period.

5. Loomingulisus ja uuenduslikkus (kuni 2000 tm): Marketing Mix Modelling is in most business cases set up typically one dimensional, looking at either an individual brand or at the aggregate group level. In this project, four models were connected into a hierarchical framework. This allowed budget management that guides individual brand planning while maximising profit for the entire group. This setup bridges the gap between marketing and finance teams: the CMO can answer strategic questions like "investing in which brand would yield the highest return?" or "does promoting one of our brands cannibalise another?" This approach also scales easily across markets to maximise group profitability. Equilia is a profitable multi-million group and Analytical Alley’s Marketing Mix Modelling can serve their growing international ambition and is precise enough to plan their marketing investments with great profitability. For example, for a business active in Estonia, Latvia, and Lithuania, it clarifies how much to spend in each country, and which channels perform best in any given market. The Marketing Mix Modelling should not attempt to answer the question, "which channel is the best?", but rather "which channel works best for which brand, at what investment level, and at what time?"

Authors: Timm Rainer Tamsar

Project lead

Ieva Beilina

Pick a role

TULIMUST
2. ANDMED / 2.1 Andmekasutus: kvantitatiivne

This entry has no visuals

This entry has no visuals

Log in to your account Create an account