How an Academic Partner Can Help You Validate Your Startup’s Product

Scientific validation is a crucial stepping stone for startups aiming to build a successful business. By rigorously testing the hypotheses upon which their products are built, tech-oriented founders can mitigate risks, increase their appeal to investors, maintain regulatory compliance, foster customer trust, and enhance their marketing strategies.

However, although it serves as a competitive edge and as a demonstration of the startup's commitment to build quality products, the validation process can encounter several obstacles. This can happen due to data shortages, restricted resources, and lack of expertise. Here is a six-step guide for startups to boost their odds of succeeding at scientifically validating their technology, a necessary precursor of a confident product launch.

1. Define your technology and target audience

Begin with a precise definition of the technology and its proposed function. If your concepts remain nebulous, conduct thorough research on the subject to ascertain your understanding of the market. Accuracy is crucial, and achieving high levels of precision in your results would be advantageous for moving to the next stage.

To compile the most relevant studies to your project, start by focusing on the ones that solve a particular problem, and then dive into the different aspects of topic-related issues. Conducting reliable studies is a tricky proposition, considering that there is hardly a universal solution to any particular problem.  Once you have identified some studies, more due diligence is needed. Be ready to check:

  • Open-sourced code. It allows you to check your idea with less effort, which will save you time. Additionally, the code gives you all the possible details about the potential implementations, something that can be easy to skip on paper. Also, this is a good sign of a good study in general.

  • Citations. If a study is cited often in other studies, there is a higher chance that you will be able to use its ideas for your project.

2. Record your results and share them with the market and investors

Once you have measured your results, you need to share them with your stakeholders and with the market at large. Write a paper that encapsulates the data and results collected, as it will serve as a testament to your research. This process not only provides a tangible record of your work but also lays the foundation for future explorations.

When it comes to investing in a company, it also serves as a form of external validation, which is a critical and highly valuable factor for investors. Funders are very attracted to credibility.

For example, in our case, we wrote a preprint, which is a scholarly paper that can be posted online before it is peer-reviewed, and in this preprint, we discussed the work that has been done on the topic we were researching, and why the world needs it. The preprint is, you could say, the beginning stage of a scientific article. It also included our method, and then we moved on to the experiment, which is the third part of the preprint. Here, we explained how we collected our data, what our initial results were, and whether they validated our hypothesis. After a successful pitch of the preprint to Harvard Medical School with no prior publication, we reached an agreement to collaborate on a joint research project.

3. Write a scholarly or scientific article

In the academic world, the process typically involves publishing an article in a recognized journal and then promoting it at scientific conferences. This exposure often leads other researchers to engage with the community, gain valuable insights, continuously improve the technology and, of course, reference your work in their own research, thereby boosting your h-index, a crucial metric for PhD students, professors, and anyone pursuing an academic research career.

Even if your startup doesn't take off, having published articles under your belt can open doors to better job opportunities. It also acts as a form of insurance. With patents and scientific articles to your name, you have the potential to land attractive roles, like becoming the new head of an engineering group that focuses on innovation and new developments. Who knows where your career path will take you?

In addition, publishing articles adds credibility to your work within the scientific community, and opens up opportunities for recruiting and building the company's HR brand.

4. Find partners to formulate a hypothesis about your technology’s effectiveness

As we delve into the effectiveness of the technology that you are developing, it is important to consider partnering with academic or research institutions to further authenticate the technology and broaden its impact. If this isn't feasible, consider finding another partner to help broaden the study by increasing the data sample.

For example, we first came up with a special version of the Neatsy app that was designed specifically for Harvard Medical School. This was a stripped-down version of the Neatsy application, but it helped researchers at Harvard collect data faster, so they started gathering information about patients and obtaining written consent from them that they were participating in a science experiment.

While negotiating with academic partners, remember that they have their own goals just as you have your own. In some cases, the academic institution’s goal is to get more quality papers published to improve their contribution to science and advance in their careers by improving their h-index, which is calculated based on how many articles they have published and how often these articles get cited. In simpler terms, it is a measure of the quality of an article and an indicator of how famous the author is.

5. Design experiments

Experimental validation helps to mitigate a startup’s risk by confirming the viability of a product before it hits the market. Designing the experiment is the responsibility of both the company and the academic partner. The firm’s engineers are the ones who know how the technology works, and the environment needed for it to flourish. The academic partners know how they can conduct the experiments and what limitations they have on their side.

For instance, our whole experiment had to be approved before the project could begin by the IRB, which stands for the Institutional Review Board. This is a special ethical review committee that every medical school has, in order to guarantee that human rights are being respected in the study.

Before starting with a new experiment that will satisfy both teams, clearly outline the experiment’s goals, rules and limitations for the research activity process. This will help you keep in line with the agreements established with partnering institutions. Good communication with the academic partner in the process of conducting experiments/trials is crucial.

Goals may vary. For example, an experiment might be designed to attain a quality level that will make it possible to move the technology into production at the end of the study stage. To balance scientific rigor with the startup world’s need for speed, you must have time-bound and budgetary constraints. Unfortunately, not every idea is possible to implement and it is important to find the point where you need to stop.

6. Validating the results

When validating the results, keep in mind that the data could still be biased. This means that the data received does not represent what is supposed to represent. For example, all age groups should be represented in the dataset, but if there are only young people, the results are not reliable for elderly people. Usually, the ones who conduct the trials care about this aspect, and will verify the data sets accordingly to prevent these biases from coming out.

There is another type of trial, which collects data for technology development and simultaneous validation. However, this approach generally has a problem of overfitting. This happens when the algorithm becomes good on a particular dataset. There are different machine learning techniques to avoid such overfitting and this is fully the responsibility of engineers. The only thing that the ones who conduct the study can do here is to insist on collecting the independent dataset to test the final model.

To provide incentives to participants, and to increase their enrollment rate, provide opportunities to get vouchers, cash or gifts. This is what we did at Harvard. The study details were published on a student-oriented website, providing them the opportunity to get a voucher to buy sneakers if they came and allowed us to take some pictures. The opportunity went viral, and our study gained tremendous insights as a result.

Once you’ve done this, here’s a final reminder that doesn’t hurt to emphasize. Keep in mind the second step, don’t forget to record all data and observations for your analysis to be accurate.

The post How an Academic Partner Can Help You Validate Your Startup’s Product appeared first on Unite.AI.

文 » A