Artificial intelligence (AI) cannot replace business intelligence (BI), nor is it a panacea for the shortcomings of BI, said Mico Yuk, CEO and co-founder of New York-based BI Brainz. In fact, she told the ITWeb Business Intelligence Summit 2020, more AI big data projects (85%) than BI projects (70%) fail.
Delivering the keynote address at the 15th annual BI summit in Sandton today, Yuk said the BI adoption rate was probably less than 30%, and over $100 billion was wasted on failed BI projects every year.
One of the main reasons for this was that BI professionals try to dazzle their audiences with a deluge of data, when, in reality, ‘gut-analysis rules’.
“No matter how good the data is, people still trust their gut more than numbers”, she said.
She explained that when people don’t like what the numbers are saying – or when they don’t understand or are overwhelmed by the data – they resort to their instincts to come up with an answer or solution they are comfortable with.
“In fact, people generally look for data that confirms their preconceived ideas or notions.”
Emotions, then data
Research has consistently shown that decision-making is largely emotional rather than logical or rational, she noted.
“Unless you tap into people’s emotions first, no amount of data will influence them. In fact, if you give them too much data, you will simply end up with analysis paralysis.”
The best way to overcome this, she continued, is to master the art of storytelling.
In Yuk’s BI world, storytelling is not a warm and fuzzy notion of smart narratives and pretty pictures. Rather, data storytelling is the ability to turn raw information into actionable knowledge.
According to Yuk, a BI dashboard methodology for an effective visual storyboard that appeals to the audience’s emotions first, must address three areas:
- What? – a KPI snapshot;
- So what? – trends (why do I care);
- Now what? – the action that should be taken.
The first step in addressing these is to ask the right questions – and this did not include the three questions most commonly asked by BI practitioners:
- What do you want to see?
- How to you want it to look?
- What do you want to measure?
“Never, ever ask these questions again,” she advised.
The following questions, she said, should be asked and answered in a way that would not confuse the audience with too many KPIs and metrics:
- What does success look/feel like? (Limit to one per storyboard/dashboard)
- KPI snapshot: What important metrics do you need to track to hit your goals (limit to five per goal).
- If your KPI (metric) is not on target, why? (Limit to two per KPI).
- How do you fix it (referring to a good/bad trend? (Limit to one per trend)
BI practitioners should also use language and images that spoke to people’s emotions. “Emotive words are easier to visualise,” she said.
Finally, no BI dashboard could address the needs of every individual in an organisation.
“Creating something for everyone would not please anyone. A dashboard/storyboard should never address more than two job titles, or two lines of business if it is to be meaningful to the people who you are trying to reach,” she said.
* Mico Yuk is presenting a workshop on Learning the art and science of creating actionable data stories on 5 March.