Frequently asked questions

On this page, you can find answers to some commonly asked questions.

IP & data

Who owns the IP of the quantification model?

The IP is fully owned by Upright.

Where does the data come from?

Currently we are using data from three primary sources: open access scientific paper database The Core, open datasets published by i.a. OECD and the English Wikipedia corpus. More about our data approach can be read HERE.

What kind of data do companies report for your model?

None. The model uses no data that comes from companies themselves.


How many companies will there be in your model?

The aim is to cover all companies globally with sufficient internet presence and/or size (revenue USD 500k+). In the free public version (The Playground), we start by publishing a handful of 800 companies and gradually add them.

The model is built in a way where the marginal cost for adding one company is zero. There is no manual analysis required in adding individual companies. A neural network links companies to products and the model automatically calculates the scores. Pursuing our mission requires covering a wide range of companies regardless of ownership structure or reporting requirements.

What does a company have to do to be in your model?


What if something looks weird in my company’s net impact profile?

If something is wrong in your profile (e.g. Leena-Leena has mapped wrong products for you), we hope you let us know about it. We can be reached either by using the contributions pop-ups in the Playground or via email at

Quantification model

How are you using machine learning?

Our quantification model consists of a pretty vast amount of “normal” arithmetics and modeling, and two neural networks. The first one reads and classifies causalities in natural language, especially the scientific papers. The second one maps companies and product/service labels to each other. More about our model can be read HERE.

When is the model going to be ready?

The Upright net impact model will never be ready. It is constantly iterated. Every day new scientific research is published and other data regarding impact of companies put online, new companies are born and new products invented. Companies change strategies, introduce new products and let go of old ones. Their net impact is not static, and neither is the Upright model. Additionally, the Upright net impact model and the neural networks serving it are developed every day.

We update the model multiple times a day, and use the newest version for our own research and client work. For the public Upright Playground, we push a new version of the model every couple of weeks. More about our the status of the model and planned next steps can be read HERE.

How can you compare different impacts to one another?

Our basic principle for putting different impacts into the same unit of measure is defining scores as (teeny tiny) percentages of all of that particular impact caused by all companies globally. Example: Nokia’s greenhouse gas score = all greenhouse gas caused by Nokia / all greenhouse gases caused by all companies globally. After that, the user’s optimization criteria i.e. weights for each of the 19 impact categories are applied. More about our approach to comparing different impacts can be read HERE.

Have you thought about utilizing this data for consumers?

Yes. Our three main user groups who we are doing this for are consumers, employees and investors. You can read more about our users HERE.


I have a project that I think is doing something similar to Upright. Are you interested in collaboration suggestions?

Upright is always interested in mutually learning from others who are driving similar missions. However, due to time restrictions, we are quite strictly focused on our own operations.

Currently we are receiving quite many collaboration suggestions. If you send one, please specify how you’d like to collaborate and why you think this would make sense for both of us.


Is Upright looking for new employees?

Upright is always welcoming applications from hard-working people with relevant skills. Before sending an application, please see what we say about criteria under People.

Couldn’t find the answer you were looking for? Send a question you would like to see answered here to: