The B2B budget lands.

How a global B2B software company modeled five products across eight countries and found that no single channel plan explained sales in more than one market at a time.

Always-ON measurement, delivered on the MassTer platform.

The Challenge

For a B2B software company selling multiple products into markets with different buyers, different competitors, and different media costs, a single global media plan is the default and rarely the right one. The company approached MASS Analytics to identify what was actually driving sales, calculate the return each channel generated, and get a media budget it could allocate with confidence across products and countries. The scale of the ask was real: five products, eight countries, one shared central budget. This case study focuses on two of the five products, referred to here as Product A and Product B, and what the country-level models revealed once the global view was broken apart.

No visibility below the global blend

Before this project, the company’s marketing team could see total media spend and total sales, by product and in aggregate, but nothing at the level a budget decision actually needs: which channel was earning its keep, in which country, for which product. Paid Search, Email, Promotion, and every other channel were bought against a single global plan, priced and paced the same way from the UK to Japan. Whatever the true country-level picture was, it stayed hidden inside a number that only ever described the average.

Four gaps a single global plan could not close

• →Channel performance was only visible at the global average. A channel judged strong overall could be carrying one market and doing almost nothing in another, with no way to tell them apart.
• →A single blended ROI per channel hid opposite stories. The same channel could be earning its budget in one country and wasting it in the next, and the global number would not say which.
• →Local buying behavior had no route into the plan. A market where promotion carried more weight than paid media, as Japan did for both products in this study, was priced the same as every other market.
• →Scale multiplied the problem rather than averaging it away. Five products and eight countries meant forty distinct combinations, each capable of behaving differently, all planned against one shared assumption.

Each gap sat invisible inside a single global number. MASS Analytics addressed all four by modeling country and product combinations explicitly, summarized below.

The Solution

MASS Analytics built the engagement in three stages rather than one. A proof-of-concept on the client’s own demo data showed what country-and-product-level modeling could reveal before any production model was built. Five products across five countries followed, producing 25 country-product models that identified each market’s real sales drivers and measured media performance on its own terms. The scope then expanded to the full five products across all eight countries: forty models in total, each fitted with its own coefficients rather than one shared global assumption.

Each country-product pair got its own model, not a shared assumption

Every model in the forty used the same core transformation toolkit, applied separately to each country-product combination so a channel’s behavior in one market never diluted its measured behavior in another. AdStock captured the carryover effect of advertising on memory: the fact that a campaign’s influence on a buyer does not end the day the ad stops running. Diminishing Returns modeled the saturation point where an additional dollar in a channel stops returning what the first dollar did. Weighted Sum accounted for the combined effect of channels running at the same time, since Paid Search, Email, and Promotion rarely ran in isolation in any of the eight markets. The three transformations are standard MMM technique; what changed the outcome here was refusing to average them across countries before the coefficients were estimated.

Getting to forty models took a staged data foundation

The proof-of-concept stage used demo data supplied by the client to confirm the modeling approach could scale across products and locations before any real budget or timeline was committed. Production modeling then ran on the client’s actual sales, media, and spend data for five products across five countries, before the client’s own team was brought onto the modeling tool directly and the scope expanded to the remaining three countries and the full five-product set. More granular data at this final stage let the country-product models separate channel performance at a level the earlier stage could not reach.

Results and Impact

The completed set of forty models produced one finding that mattered more than any single number: the same channel performed differently enough by country that a single global media plan could not have been right for more than a fraction of the company’s markets.

Paid Search carried the UK. Email carried Canada.

Paid Search performed well for Product A in every one of the eight countries modeled, but its strength in the United Kingdom stood apart: 18% of Product A’s UK sales traced back to Paid Search, the single strongest channel result recorded anywhere in the study. Product B told a different story in a different market. Email was a stronger driver for Product B in Canada than in any other country, reaching 3.5% of the product’s total sales there. Neither result generalized. Product A’s UK strength in Paid Search did not repeat at the same intensity in the other seven markets, and Product B’s Canadian Email effect did not repeat for Product A or in any other country. A global average across all eight markets would have understated the UK finding, overstated the typical market, and missed the Canada finding entirely.

One shared plan became forty country-product plans

Promotion was the most impactful driver for both Product A and Product B in Japan specifically, a pattern that did not hold in any other market for either product. Findings like this, repeated across the forty country-product combinations, meant the client’s central marketing budget could no longer be allocated the same way in every market. MASS Analytics recommended reallocating the shared budget by country rather than applying one global channel plan, giving the client a basis to decide how much, how often, and when to invest in each channel, market by market.