Behind The Scenes of Optimization Use Case

The challenge

A media budget is easy to grow and hard to place. A major personal hygiene brand had moved most of its media budget toward digital in recent years, and it ran a product substitution strategy across several regions. So its Marketing team wanted to know what its activity added to sales, which channels worked best, and where its next dollar should go. This case study shows how MASS Analytics found the point where each channel stops paying back, and what moving spend changed.

Marketing spend optimization began with one budget, many questions

The brand’s strategy was to substitute one product for another. Because of that, every metric had to be grouped by that logic, not by standard product categories. Modeling ran at product group level, across several regions, on three years of monthly data. The client also brought a layered list of questions: the total effect of marketing on sales, the most effective channel in total and in detail, the effect of cannibalization within the portfolio, the impact of competitors’ strategy and external influences, and the best split of budget across channels.

Most of these questions look back, but the last one asks for a decision.

Four things marketing spend optimization had to show

• The total effect of marketing. Sales moved for many reasons at once, across regions and product groups. So the team needed marketing’s own share.
• The best channels, in detail. TV and digital each hide many touchpoints, so a channel total cannot say which one works.
• The pressure around the portfolio. Cannibalization between products, competitor moves, and external influences all shifted sales, so each needed its own measure.
• The point where a channel stops paying back. Because spend had leaned toward digital, there was no way to say where more spend stopped earning without a saturation point for each channel.

Each gap points to a modeling choice, so MASS Analytics addressed them with the design summarized in Table 1.

The solution

MASS Analytics worked in an agile flow: prepare the data, fit the model, read results at two levels, then optimize. Automation handled data preparation, and three years of monthly data went into a model built by product group and region. Finally, model output became a budget answer. It is the same approach MASS Analytics uses for marketing spend optimization across a media mix.

Pooled modeling solved thin data

Three years of monthly data gave few points to fit in any single region or product group. To raise the number of fitted points, the team used pooled modeling. As a result, the model draws strength across regions and product groups instead of relying on one thin slice, which makes each estimate steadier.

Two layers of detail

The team read results at two levels. First, it examined digital media as a whole, then broke it down by individual touchpoint. Likewise, it examined TV as a total channel, then by spot length. So a strategic call, such as how much to move out of TV, could be traced to a tactical choice in a specific placement.

Marketing spend optimization from three curves

Optimization needed two measures for each channel: the saturation point, where added spend starts to earn less, and the range in which spend still pays. To get them, the team built three curves. The response curve shows revenue at each budget level. Next, the profit curve subtracts spend from that revenue, and its peak, the max profit point, starts the Optimal Execution Range. Then the cumulative profit curve, a rolling average of the profit curve, crosses the max profit curve at the saturation point, which ends the range. Finally, percentage saturation is the current budget divided by the end of that range.

Results and Impact

The model showed that TV had almost no headroom left, while several digital channels still had room. Therefore, moving spend toward them lifts modeled media revenue 5.5%. Because TV sat at 96% saturation, a 20% TV budget cut costs only 1% of TV revenue.

TV was 96% saturated

The TV response curve showed a current budget of 280k sitting near the flat part of the curve, but the curve alone could not say how flat. The cumulative profit curve could. So MASS Analytics divided the current budget by the end of the Optimal Execution Range, which gave 96% saturation. That confirmed TV was operating close to its ceiling, where each added dollar returns very little. MASS Analytics delivered the range for TV and digital media as a first read on where each channel saturates.

A modeled shift lifts media revenue 5.5%

MASS Analytics then used the curves together to plan a reallocation. Moving spend away from saturated TV and toward channels with stronger marginal returns lifts total media revenue by 5.5%. For example, the plan cuts the TV budget 20% for a 1% revenue change, and raises social media 30% for a 47% revenue gain. In addition, YouTube rises 20% for a 27% gain, Google 25% for a 10% gain, and programmatic display 10% for a 37% gain. However, these are modeled results, not measured outcomes.