There have been predictions for years about the ways artificial intelligence will impact business and everyday life. Perhaps the biggest single focus has been on what jobs will be lost and which ones could be created. There is common agreement that routine and repetitive jobs will be the most vulnerable due to the relative ease of automation, whereas jobs that require creativity, critical thinking, and human interaction are less likely to be impacted.

Yet the creative sector has been rocked by the meteoric rise of generative AI. Users are able to generate images, videos, text, code, and more, at the tap of a button without the need for coding knowledge or creative skills, or expertise. How much of a crowdfunding campaign’s content could be generated by AI? And can AI analyse pitch content to predict crowdfunding campaign success rates?

Preparing crowdfunding campaign content

All crowdfunding campaigns need well-designed documentation that includes a motivational pitch and financial details that support a business plan; plus images and a clear and compelling video. Based on this content, a successful crowdfunding project needs an effective presence on social media to reach a wide audience.

Tools like ChatGPT (which stands for generative pre-trained transformer), Midjourney, Stable Diffusion, Dall-E, and Copilot allow users to generate text, imagery, computer code and other content based solely on natural language prompts. Chat GPT can be briefed to create content to be specifically used on any given social media channels, such as a LinkedIn Update or a Twitter post. AI-generated art models like DALL-E (its name is a blend of the surrealist artist Salvador Dalí and the Pixar robot WALL-E) can create images on demand. YepicAI is a pioneer in text-to-video algorithms and machine-learning models that allow businesses to turn text scripts into professional videos. Today’s content creators seem to have no need for studio facilities.

Such generative AI platforms use neural network architecture to generate human-like text by predicting the next word in a given sequence.  They are trained through machine learning based on large datasets of written text including books, articles and websites, and images. The written word platforms learn the patterns and structures of human language and can generate new text that is similar to the text it was trained on. 

However, there is always scope to enhance AI-generated content by adding unique features about your product, service, or business that make your crowdfunding opportunity worth supporting. ChatGPT cannot access the internet to find these details itself. 

The impact of specific words

Generative AI is unlikely to take account of studies that have taken a scientific approach to examine the impact of specific words and phrases in crowdfunding documentation. Nuno Arroteia is a serial angel investor, and a senior lecturer and researcher in entrepreneurial finance and international entrepreneurship at De Montfort University in the UK. After first studying previous work based on Casual and Effectual wording, he used econometric modelling to show that increasing the effectual core words or phrases by up to 5% can increase the probability of crowdfunding success from between 45% to 65%.

Here is a practical example in the narrative of a campaign that included causal words (underlined).

“To reach our goals we have conducted extensive market and competitive analysis paving an in-depth plan to what will be implemented to maximise our customer base in the long term.”

The following is an example of how the phrase could be tweaked to include more effectual words (underlined), thus potentially benefiting the outcome of the campaign.

“To reach our goals we will collaborate and engage creatively with our existing customers and experiment with different approaches to see which one works best, and growing our customer base as we go.”

Marketing a crowdfunding project

The next stage after creating content is getting it seen by the right sort of people. Finta is an example of a generative AI platform geared specifically around investing. It matches investment opportunities with a database of the investment history of each person in its network, and “speed dates” by sending tailored messages to the most likely backers. Investors thus receive notification of investment opportunities that match their personal key criteria and reflect their legacy portfolio.

Can AI predict crowdfunding success?

Several studies show it is possible to use artificial intelligence to predict the success of a crowdfunding campaign. Machine learning algorithms can analyse data from past crowdfunding campaigns and identify patterns and features that are associated with successful campaigns. This information can then be used to predict the likelihood of success for new campaigns. However, the accuracy of these predictions will depend on the quality and quantity of data used, as well as the complexity of the model used.

Several factors can impact the success or failure of a crowdfunding campaign, including:

  •  The quality and appeal of the project, product, or business being funded.
  • The campaign's marketing and promotion efforts.
  • The extent of “extras,” the added incentives for backers.
  • The campaign's funding goal and deadline – realistic, achievable, worthy?
  • The credibility and track record of the campaign creator or the business management team behind it.
  • The extent of early support through the engagement and involvement of the campaign project’s community.
  • The level of competition in the campaign's category, and extent of standout.

The most important factors will vary depending on the campaign (donations, reward-based, equity) and thus the platform it is on. A well-executed campaign that is appealing to its target audience, has clear and realistic funding goals, and is marketed effectively, will have a better chance of success.

Most attempts to use AI to predict crowdfunding success have focussed on reward-based projects carried out on Kickstarter or Indiegogo. Data on the kinds of variables listed previously can be scraped from hundreds of thousands of projects and analysed using various machine learning techniques such as decision trees, neural networks, and Bayesian networks, to predict the success level of new projects.

Much of the work has been carried out by academics. Notable papers include:

However, equity crowdfunding is a newer form of crowdfunding, and it remains unauthorised in some major economies including India, the world’s third-largest startup ecosystem. Overall, there have been far fewer equity than in reward-based projects, and so there is less data available for training machine learning models.

Additionally, the process of equity crowdfunding is more complex than other forms of crowdfunding, as it involves the sale of securities that are subject to various regulatory requirements. Financial documents included in any pitch are often subject to investors making a specific request and being approved to receive them. Much of the data required for a thorough database to enable machine learning is thus retained by the multiple businesses that traded their equity through crowdfunding, and it is not readily available in the public domain. The businesses that own the data are likely to consider it confidential.

The largest UK equity crowdfunding platforms, Crowdcube and Seedrs, would be remiss if they did not track this data themselves for the campaign they host on their sites, and then use it to give advice to future campaign creators. It would thus represent a point of competitive advantage, and the findings are unlikely to be disclosed and made available to the general public. Nonetheless, broad industry guidelines to improve the likelihood of success have been established.

About us:

CrowdInvest will be a cross-border equity crowdfunding platform offering UK-based sophisticated investors opportunities to invest in impact-driven, high-growth tech startups operating in emerging economies. You can join the waitlist today at to stay up to date with developments on how to be involved.