Making-of: How score4more built a platform for sustainable transformation

In this blog article, we take you into the engine room of score4more and show you from a content and technical perspective how we have designed, developed and implemented the score4more platform since 2022, co-financed and supported by theΒ BIG R&D Program Brandenburg with funding from the European Union.

 

Over 12 months, we developed the platform and technology from scratch in several steps:

‍
Step 0: The platform concept
Step 1: Development of the platform infrastructure and the sustainability profile
Step 2: Development of an automated profile analysis with AI/machine learning
Step 3: Development of an automated scoring system for sustainability performance
Step 4: Development of a benchmarking system with recommendations for action
Next: The future πŸ™‚

 

Step 0: The platform concept

We wanted to create a platform that would help companies to transform sustainably faster and more easily. We didn’t want it to be the 101st ESG reporting solution that supports processes and data management. There are enough of those on the market – we were also skeptical whether companies really wanted to do data management online to meet the new EU ESRS (European Sustainability Reporting Standard) reporting requirements with over 1000 data points. SMEs signaled to us early on that they would start flexibly using Excel and Word to set up reporting processes in the first few years, so we didn’t think a new online solution for reporting made much sense.

 

More important for us was the question of how we could support and accelerate real transformation. In order not to miss the 1.5Β° target and to have an impact on various sustainability crises, we need a platform approach that scales quickly across many companies and connects them. The good news is: thanks to data and technology, it is now conceivable that the 300 million companies worldwide can be connected and empowered for transformation at the same time – Facebook and LinkedIn have already shown the way.

 

Imagine: all companies worldwide have the most important solutions for transformation easily available online. Experts, suppliers and customers network to transform sustainably together on the basis of structured data. If we harness this concentrated innovation and economic power, it would actually still be possible to turn the tide.

 

In other business models, we have observed how structured profiles, transparency and networking have significantly changed industries and markets. One example is the booking.com platform, which has reinvented the rules of the hotel market through structured hotel profiles, hotel scoring and feedback. Quality, prices, bookings and occupancy rates in hotels have developed significantly as a result. Hardly anyone goes to the hotel website anymore, but uses profile portals instead.

 

Sustainability is much more complex. The European Sustainability Reporting Standards (ESRS) alone, the new reporting standard for companies in the EU, are a highly complex compendium with 12 ESRS topics in 4 groups and numerous sub-topics, as well as the 1,000 data points already mentioned. Climate, resources, pollutants, human rights – the range of diverse topics is extensive. Depending on the industry in which a company operates, different topics are decisive, which are material for this industry and then for the individual company (IRO is a magic word here: Impact – Risks – Opportunities). We realized early on that ESRS can be both a curse and a blessing. A curse, because in case of doubt it can severely overburden companies and cost a lot of money for consultants, auditors and software. And a blessing because, for the first time in the EU, there is an opportunity to collect substantial and comprehensive sustainability data from companies in accordance with a standard, which can then be used for the transformation and, in particular, for investors and sustainable finance.

 

While auditors, consultants and the aforementioned 100+ ESG reporting solutions have set about getting companies through the ESRS complexity (outcome still uncertain), we have been preoccupied with the question: How do we manage not to waste any time with the transformation? After all, the worst case scenario would be if companies, especially SMEs, are practically “paralyzed” for the next 3-5 years by ESRS and various other regulations such as EU taxonomy, LKGS, CSDDD, CBAM & Co. and are only occupied with documentation, reporting, ticking boxes and compliance. The focus should be on real change – on practical measures, solutions and innovations.

 

Our platform concept should therefore enable parallelization with the guiding principle: “Reporting and transforming go hand in hand”. While companies collect initial data for reporting, they should be able to use it to engage directly with peers, customers or suppliers and network in order to learn from each other. Just as the German Minister of Economic Affairs Robert Habeck recently said at a conference: “It doesn’t help us to be 100% perfectly audited if we are also 10 years too late with the transformation.” It is better to start step by step, to implement imperfect solutions 80-20 and to learn quickly during the process. That way, we won’t lose any time in this critical phase before 2030 with the closing windows of opportunity in the climate crisis. Companies can use data, even if it is not complete and perfect, to directly add value and get through the transformation faster and easier – the key to this is a digital platform.

 

Step 1: Development of the platform infrastructure and sustainability profile

 

So we got started and built a transformation platform to connect as many companies as possible for sustainability. The first step was to build the infrastructure and basic technical elements of a platform – the tech stack. If the platform is later to be used by thousands or even millions of users worldwide, it needs a scalable and powerful data infrastructure. The choice here fell on a modern serverless cloud-native architecture, which is characterized by the fact that it can be expanded at will. As is now standard on the internet, it is API-capable, meaning it offers interfaces for exchanging data with software partners, companies or banks. We also chose Typescript for the front end and Python for the back end, programming languages that will enable us to connect AI and machine learning models, which are mostly developed in Python, to the platform later on. With this infrastructure, our first steps began to develop user accounts and admin functionalities and get ready for the first functionality: the sustainability profile.

‍
The sustainability profile is a “missing link” in sustainability for us. Before booking.com, travelers had to go to the hotel websites, book a hotel by phone or order brochures. Sustainability is no different today. If you want to know something about a company’s sustainability, you can google it, read company website descriptions or read lengthy sustainability reports. Existing databases, whether from rating agencies or reporting platforms, are often too technical, expensive and complicated. Most of the platforms are “closed-shops”, i.e. not accessible to the wider market and SMEs in particular.

‍
AI such as ChatGPT does provide new ways of researching sustainability data. However, the results are not very structured and cannot be compared or interlinked, meaning that although they support qualitative answers to specialist questions, they cannot yet be used directly as a data-supported basis for transformation.

‍
The sustainability profile is one approach to improving the situation. It is a layer between the detailed information on websites and in reports for users and stakeholders who want to find out about the company’s sustainability quickly and in a structured way and analyze and evaluate data in a networked manner. The special feature of sustainability is that the relevant topics are different for each industry. Different topics are important for the automotive industry than for the textile industry. For example, while human rights standards in the supply chain are in demand in both industries, they are far more important in the textile industry than in the automotive industry. There, the drive technology of vehicles is more important in terms of CO2 emissions and pollutants.

‍
The industry-specific sustainability profile therefore summarizes the sustainability performance of companies from the report and website in a simple and structured way. This turns 80-page reports into a 2-page structured summary of the key topics. We wanted to implement this “hack” on the platform in order to dramatically simplify complexity.

 

No sooner said than done: we started with a sustainability framework that takes the 5 transformation fields of climate, resources, nature, value chain and society as well as 12 ESRS topics as a basis and then comprises around 70 criteria that are dealt with in practice by companies in the area of sustainability. From energy efficiency to renewable energies and food waste, we have worked with our partner, the German Sustainability Award (DNP), to build up in-depth knowledge of which measures and topics are being addressed by companies in which sectors. Together with the DNP, we then differentiated the entire economy into 100 sectors (based on the EU NACE industry classification, among others). In doing so, we were also guided by the industry definitions and reporting topic specifications of the EU ESRS for the industries, which are, however, much more coarse-grained and go less deeply into industries, e.g. in the service sector. We then used this toolkit to develop specific sustainability profiles for the 100 sectors, which comprise a selection of the 70 criteria on material topics for a sector.

‍
These profiles were implemented on the platform as the first function. Companies can simply select their sector free of charge and then receive their sector-specific profile, where they can map their sustainability performance in a structured manner and enter the relevant sources, such as their website, sustainability report or declaration on the German Sustainability Code. With the profile, companies can now compare themselves more directly within their sector, as other companies have created their profile in exactly the same structure. The profile is the basis for subsequent networking, data analyses and decisions, which should help to accelerate the transformation. Such a profile is also the basis for the application for the German Sustainability Award. Thanks to the profiles and the platform, thousands of companies could be analyzed and awarded for the competition on a sector-specific basis: a milestone for the DNP and for Germany in terms of making examples of corporate sustainability visible and thus bringing the transformation to the wider economy. We knew that the short profiles with a few criteria per sector were just the beginning, a minimum viable product (MVP). At the same time, we started prototyping a long profile to enable companies to map other ESRS topics in the profile and gradually introduce it.

 

Step 2: Development of an automated profile analysis with AI/machine learning

When launching a platform, the biggest hurdle is motivating users and companies to use it. Platforms only work from a certain critical mass of users and data. It was therefore clear that we had to make an advance effort and create the sustainability profiles of companies ourselves as a first step, which would then make it easier for users to access the platform. To this end, we developed models in cooperation with the Hasso Plattner Institute (HPI) in Potsdam to automatically populate sustainability profiles. The basis was public sustainability data from the companies, which could be analyzed using natural language processing (NLP) models in such a way that a large number of profiles could be filled out. The models are divided into two parts: one model analyzes data for texts regarding measures and solutions for the profile. A second model extracts key figure values from reports for key figures relevant to the profile. The models are integrated into the platform infrastructure and work with both raw data and entire documents as part of pre-processing. These models have helped to significantly accelerate the automation of profile creation.

 

Step 3: Development of an automated scoring system for sustainability performance

 

After a few months, we had 2,000 company profiles from 100 sectors available. In the next step, we wanted to find a way to give the companies an initial assessment of the question “Where are we in the transformation?”.

‍
“Are we still at the beginning, in the middle or already fully transformed towards climate neutrality or closed loops?”.
We have developed a 6-level scoring system for this, which also gave score4more its name. From level 0 = “Inactive” to level 5 = “Leading”, the sustainability performance in the company’s profile is classified for the respective criterion. Scoring criteria are stored that define for each of the six levels which facts must be present for a level to be reached. This scoring feature gives companies simple feedback on their profile, something that has often been lacking until now: many companies have been reporting on sustainability for years, in some cases without having received structured or systematic feedback in the form of scoring, which they can then use to improve in a more targeted manner. We quickly saw the potential of scoring at pilot companies: Sustainability managers can easily and clearly communicate areas of improvement in sustainability to company management. One result is that management and the supervisory board can make more targeted decisions on new measures and initiatives to improve in the respective areas and reach the next level of transformation. Scoring is therefore an important mechanism for driving and accelerating transformation in a targeted manner.

‍
Technically, scoring is based on the company’s profile data. Thanks to AI/ML, scoring can also be automated with a robust AI/ML application strategy: analysis and classification activities that would otherwise be performed by a human analyst are carried out via AI/ML at different prompts. Furthermore, manual scores could previously be used to train a proprietary scoring model in order to carry out automated scorings. Both methods are used for robust scoring, in which the results from the different methods are evaluated in combination and only cases where there are strong deviations in the AI/ML-generated scoring results are checked manually. This procedure is possible because the scoring levels represent a basic positioning of the company in a criterion and are therefore clearly differentiated. The clear demarcation in the level definitions enables AI/ML procedures to be applied robustly.

 

Step 4: Development of benchmarking with recommendations for action

 

Nothing is more powerful than comparison: benchmarking is used by companies in all forms of economic life. Whether cost benchmarks, price benchmarks or service benchmarks. Companies continuously compare themselves to peers and so-called “benchmarks” to find the best in the industry. Benchmarking also has outstanding potential in the transformation to sustainability. Previously, most companies did not know how well their reported sustainability performance was rated or how it compared to their competitors. Until now, only large and listed companies have been able to compare sustainability data via ratings and ESG databases. SMEs, on the other hand, are left out and can hardly afford the expensive ratings and databases. And even the big players can compare data such as CO2 values, but have so far found it difficult to interpret whether and where there are differences in performance (“apples with pears” problem).

‍
With score4more, we are solving several of these problems. All companies can compare themselves with pioneering companies in their sector in a benchmarking process. To this end, we have brought aggregated profile data from the best companies in the industry into the comparison. This allows companies to compare their scoring with the best in the industry and at the same time get a selection of the best practices that other companies are using to make progress in their transformation. The combination of score comparison and best practices provides a company with direct recommendations for action for its own implementation in order to improve. AI/ML is also used in these recommendations for action to automatically summarize selected best practices for benchmarking from the companies’ raw data.

Benchmarking can be used for all companies of any size and in all industries. Recommendations for action in 100 sectors from more than 500 pioneering companies are made available to users online via sector benchmarking. This feature is part of the Pro package, which can be used online with a low-cost subscription. It enables thousands of companies to get started and advance in the transformation in a simple and practical way.

‍
All in all, the score4more development project has managed to create a transformation platform with 2,000 pioneering companies and scoring and benchmarking for the economy as a whole and SMEs in particular, thanks to Big R&D funding, which has created a fundamentally new offering that can be scaled further in Europe and worldwide thanks to data and AI/ML.

 

And then: Future to come πŸ™‚ – the bigger picture

We have the opportunity to accelerate transformation through digital technology, data and AI/ML: the transfer of best practices, knowledge and data is crucial.

 

The good news is that everything that is needed is available: Knowledge, technology, capital and networking globally and in real time. Whether it’s the climate crisis or the circular economy: the solutions for the transformation are already available, others will soon be available and ready for the market thanks to innovation processes.

 

But they are still not visible enough. Companies often reinvent the wheel and only see “the tip of the iceberg” in terms of solutions and good examples. And many problems can be solved more quickly by working together across company boundaries. The magic word is “collaboration”, with customers, suppliers and partners. To solve the climate crisis, we now need to make maximum use of these “superpowers”. In other words, maximum transfer of best practices and solutions between companies worldwide as well as maximum networking and cooperation in order to achieve the transformation to a climate-neutral circular economy with regenerated ecosystems, fair working conditions throughout the value chain and prosperous societies.

 

score4more is a kind of dance floor for this, a platform that networks and enables companies to “dance” transformation together in a variety of ways. Because we are all humans, not robots – software and AI can help us to focus on the essentials – the effective implementation of sustainability. With this in mind: the dancefloor is open, let`s transform together πŸ™‚!