GPT-3, Stable Diffusion and Web Analysis Tasks
Whether deep-learning text or text-to-image generators, the current development of AI projects is impressive. However, the quality of the result depends largely on the input prompts or questions. So human intelligence is still required here.
In web analysis, the task or question is also crucial. And data experts are needed to answer these questions on the basis of the data. Artificial intelligence methods will certainly help with the answering of these questions in the future.
Reason enough to take a look at typical questions and tasks in the field of web analysis:
1. Decision support
Should I offer more videos to increase the conversion rate? Is it true that two-minute videos perform better than longer ones?
2. System health indication
Is my conversion rate in the “green zone”? Which KPIs have changed and how?
3. Performance feedback
Have I achieved the goal of increasing the conversion rate?
4. Root cause analysis
Why has the conversion rate dropped? Have many new visitors been approached?
5. Knowledge creation
What factors influence the conversion rate?
6. Vulnerability assessment
Where am I losing the most conversions in the process?
7. Prioritization
Should I focus on the checkout process or the landing pages?
So data analysts will not be replaced by algorithms in the foreseeable future. Rather, the combination of human and artificial intelligence promises even greater added value from the treasure trove of data. Moreover, the level of reporting stands and falls with the data basis, because the principle of “rubbish in, rubbish out” (GIGO for short) also applies to deep learning projects as well as to web analysis. Particularly dangerous are biases in the database, in artificial intelligence for example through discrimination in the data sets, in web analysis through consent bias.