When other researchers read a publication, debate about your findings and refer to your article, scientific impact happens naturally.
A scientific publication as such, however, is not impact. What if no one reads it or cites it? Impact requires that others uptake the results. This is also reflected in the evaluation of research and its quality. Traditionally research has been evaluated by peer review, and the number of scientific publications or citations have been primary indicators of scientific impact.
Nowadays, even in peer reviewed scientific evaluation, evaluators increasingly want to see some indication of future application and impact possibilities even if the research is considered basic research, and the larger benefits may take place a decade later. This also reflects the current thinking of universities contributing to the society and economy. Researchers are asked to present numbers to show the potential returns on the tax payers' investment.
Partly as the result of the above, research has become increasingly interdisciplinary, and in order to reach maximum scientific impact, a researcher usually collaborates with not only researchers in their own field or topic, but also researchers from other fields. For example, engineers can develop tools for communication, but in order to understand how humans naturally interact or how we uptake technology needs understanding of psychology, and behavioural and cognitive science.
Other researchers may develop the concept, solution or theory further or in different directions, in which case the impact is wider. They can guide your research in new directions and help you maximise impact by opening up new research topics or even new fields that in turn enable new openings and ideas.
An example of such new scientific fields could be learning technology or human-computer interaction. Neither would be possible without in-depth knowledge of both human behaviour and technology.