No. Dataro does not include personally identifiable information such as donor contact details in our predictive models, nor do we combine donor personal data with data from other clients (i.e. we do not match donors in your database with their donations to other organisations).
Dataro's predictive models are built using de-identified fundraising data from several hundred organisations. This 'data pool' is not a separate repository of data. It is more accurately thought of as a modelling concept, whereby data from multiple sources (i.e. multiple charities, in this context) is leveraged in order to train models which are more powerful and more robust. This concept is called ‘foundation modelling’. Whenever any modelling takes place, the data is first converted from its raw format (i.e. the format we receive it in from you), to numerical values. This process is known as ‘featurisation’ and serves to de-identify the raw data. You can find out more about Dataro's model features here.
This approach is especially important when modelling rare events (i.e. major donations), helping an organisation 'cold-start' a program (i.e. providing convert to RG scores for an organisation that has never run this program previously) or providing predictions to a smaller organisation. In all of these cases, a single organisation will not have enough positive instances of the desired event occurring to train a robust model, necessitating the foundation modelling approach.
When would Dataro match a donor record across data sets?
We only match donor records across your own datasets when your instruct us to do so - e.g. to combine your CRM data with your email data and create a 'single view' of your donor.
Can my organization's fundraising data be used in this de-identified data set to train models?
Yes, unless you opt-out during contracting.