Your data, your visualisations

Day 5 · 09:30

A visualiz-athon. Eight sector extracts. You pick one that smells like your P&L. You make one chart you would sign, and one you would refuse.

The original Friday-equivalent threw people between public datasets (crime, prizes, lyrics). You get anonymised commercial extracts that span the functions in the room. Work in threes. Forty minutes to a first chart. Then we rotate.

The brief, for every table

1. What decision does this chart change if it is true?
2. What is the unit of observation? (lane-week, employee, campaign-week)
3. What would a bar-of-means hide?
4. What lurking variable would make the title a lie?
5. Where is the file, and what is the extract date?

Pick your P&L

1. Supply chain — OTIF versus lead time

supplier-otif.csv. Four logistics partners, 24 weeks, lanes across Lagos, Kano, Onne, Abuja. Gulf Freight sits on the efficient frontier. Sahel Haulage is late and unreliable. Is that a sourcing decision or a lane decision?

otif <- read_csv("supplier-otif.csv")
 
otif %>%
  ggplot(aes(lead_days, otif_pct, colour = supplier)) +
  geom_point(alpha = 0.7) +
  facet_wrap(~lane) +
  labs(
    title = "Reliability is a lane, not a logo",
    subtitle = "OTIF vs lead time, 24 weeks",
    x = "Lead time (days)", y = "OTIF %"
  )

2. HR — attrition, tenure, engagement

hr-attrition.csv. Role family, band, tenure, engagement, regretted-exit flag. The CHRO question is not ‘is attrition high’. It is where regretted attrition clusters — usually early tenure, specific families, not a company-wide story.

hr <- read_csv("hr-attrition.csv")
 
hr %>%
  ggplot(aes(tenure_years, engagement, colour = factor(attrition_flag))) +
  geom_point(alpha = 0.7) +
  labs(
    title = "Engagement and tenure",
    colour = "Exited"
  )

3. Marketing — ROAS by channel

campaign-roas.csv. TV will lose a ROAS league table to Search. That is not a reason to cut TV. Rank channels against the job (reach, conversion, retention), not against one ratio.

4–8. The rest of the shelf

• breakfast-panel.csv — FMCG brand / channel interaction (Day 2).
• nps-by-channel.csv — banking CX distributions.
• voc-complaints.csv — qualitative, CX or HR.
• macro-correlations.csv — economics / ALCO.
• brand-mentions.csv — comms / strategy.
• sku-descriptions.csv — category / assortment.