BlogIlkka Keskiväli 09.11.2022

All about ROMI. ROMI modeling selection guide 

Data-driven leadership Marketing effectiveness Sales growth Marketing optimization Conversion optimization Data Science Sales modelling

ROMI, MMM, sales modeling, econometric modeling. A beloved child has many names. All refer to measuring the effectiveness and efficiency of marketing by means of statistical methods. What kind of options does a marketer have and how to choose the most suitable one for him? Dagmar Drive Director Ilkka Keskiväli compiled a small ROMI modelling selection guide. 

Regular ROMI measurement teaches how to increase marketing revenue 

ROMI “Return on marketing investments” measures the return on marketing investments. ROMI indicates how much euros invested in marketing have generated sales, margins or turnover. It can be used to measure the effects of marketing measures, learn from the results and utilise the information in optimizing future measures. When measured regularly, the effectiveness of marketing can be continuously increased. 

Measurement tools have developed significantly in recent years as better interface connections and new players enter the market. The changing world situation also places new demands on measurement; In addition to more strategic time span measurements, new solutions for measuring marketing effectiveness have emerged. 

Modeling with a partner 

Marketing effectiveness has traditionally been measured using modelling carried out by marketing and research agencies. Agencies have access to the client’s marketing data, which is enriched with the client’s own data and external data sources. This provides a reliable picture of the factors affecting sales. 

In addition to marketing, other factors affecting baseline are carefully differentiated – how much impact the brand, pricing or competitors’ actions have on sales. 

Because customer purchase paths are complex, modeling should not focus solely on sales modeling. In Dagmar´s modelling, we take purchase paths into account – in addition to sales, we also build models for website and search engine traffic. This gives us as realistic a picture of the effectiveness of marketing as possible and allows us to reliably utilise the results to optimize marketing. 

With the help of the scenario tool, the results are agilely put into practice 

Traditional modelling is especially a strategic tool that provides a holistic view of the factors affecting the customer’s sales. Modelling can also be used to make forecasts and scenarios about the impact of future campaigns. For almost 10 years, Dagmar’s customers have been using a modelling scenario tool that makes it agile to optimize future marketing investments. With the help of these optimizations, Dagmar’s customers have improved their marketing efficiency by an average of 30% during the first year of continuous modeling, e.g. in the case of modelling. quarterly. 

Modeling with Open Source tools 

The major media outlets – Google and Meta – now offer their own tools for marketers. Google’s tool is called LightweightMMM and Meta’s tool is Robyn. LightweightMMM is a ready-made python library and Robyn is a ready-made R package. Both help with modeling, but without a deeper technical background, modeling is not easy to implement. 

It is also good to remember that modelling is only the first stage of the whole – the results still need to be made usable. The results must be analysed into conclusions and recommendations for action for future campaigns, which requires a deeper understanding of, for example, the practical limitations of marketing. 

Modeling as a SaaS service 

Today, traditional modelling is also carried out by companies specializing in them, which offer modeling as a SaaS service. Although operators may call modelling automatic, data collection is in fact largely manual. In addition, you have to wait several months for the results – including the modeling period, data collection and modelling work. The results are usually provided through the dashboard tool. 

Like Open Source tools, SaaS services do not contain conclusions and recommendations for action – the most important stage of modeling. 

Dagmar has been increasing the efficiency of its customers’ marketing through modelling for 10 years. Based on our own modelling and the results obtained through the SaaS services of other operators, we dare to claim that if the results of modelling are not put into practice, it will remain a completely useless tool. 

Modeling with fully real-time tools 

Dataputkien automaation avulla myös mallinnus on mahdollista automatisoida ja toteuttaa reaaliaikaisesti. Kaikki data ei kuitenkaan ole vielä automatisoitavissa. Siksi Dagmar kehitti kontribuutiomallinnuksen, jolla selvitetään digitaalisen markkinoinnin tehokkuus konversioiden näkökulmasta. Mallinnuksessa huomioidaan kaikki data, jonka kerääminen voidaan automatisoida. 

Dagmar’s Datacron® service is used to collect digital marketing contacts, investments, conversions, as well as the customer’s own data and data from external data sources in a format that can be used for the contribution algorithm. The contribution algorithm uses new data updated every night to calculate the effectiveness and efficiency of marketing. At the same time, the information is updated in the dashboard tool. Based on the recommendations received from the dashboard, Dagmar’s designers can quickly make changes to actions, the effects of which can be measured even the next day. 

In the future, new data sources will be added to contribution modelling, the most interesting being offline media. In this way, contribution modelling will soon approach traditional modelling – but still as a much faster and more agile tool. In the future, contribution modelling can be used to control automatic media buying extensively, across all media. 


Ilkka Keskiväli

Lead Strategist, Advanced Analytics


Ilkka is the head of the Advanced Analytics unit, whose passion is data-driven management and its enablement and implementation in customer organizations.

More on topic