Skip to content
Black Friday week starts Tuesday, November 24.See the offer

We built the analytics dashboard. It changed nothing.

Editorial portrait artworkKwabena Mensahgbeya teamHR Consultant · membershipGhanaScaling·1mo ago66

Case study

Author-reported results.

Starting situation
A 400 plus person corporate academy in Accra wanted better learning analytics. I pushed for a full dashboard without first agreeing which decision it would support.
What changed
We tracked enrollments, completions and satisfaction scores and presented a clean dashboard to the board.
Result
Satisfaction scores looked good, but line managers saw no change in on-the-job performance. The dashboard went unused after the first quarter.
Limitations
No baseline, no comparison group, no link to a business metric. It does not prove dashboards fail. It proves I did not define use before building.

In brief

Kwabena Mensah opened a case study in Proof & Progress, about wins: I am the one who pushed for the dashboard, so I will say the uncomfortable part first. Last year a corporate academy in Accra with 400 plus employees asked me to help them do learning analytics. 6 replies so far. Nathan Lachance says: I would push back slightly. The board loved it, so the analytics did work for them. The failure was that line managers were never the audience. 5 other replies add further detail below.

I am the one who pushed for the dashboard, so I will say the uncomfortable part first. Last year a corporate academy in Accra with 400 plus employees asked me to help them do learning analytics. I set them up with enrollment, completion and satisfaction data, then a clean dashboard for the board. Six months later the board had nothing but nice screenshots and line managers said performance had not moved. I had no answer because I never tied any metric to a business decision. The academy renewed, but follow-through stayed weak. My mistake was selling better data before agreeing who would act on it and what they would stop doing. Now I start with one decision the L&D lead must make, then measure only what feeds it. The dashboard was the answer to a question nobody asked.

6 replies

Audience: Public. Replies follow this audience.

  • Editorial portrait artwork0
    Nathan Lachancegbeya teamBusiness Coach · self-paced course

    Canada·Established Solo·1mo ago

    I would push back slightly. The board loved it, so the analytics did work for them. The failure was that line managers were never the audience. I separate buyer, user and owner before collecting a single field. Did anyone at the academy have authority to stop a program if the number did not move? If not, you had a reporting project, not a learning analytics one.

  • Editorial portrait artwork0
    Kahu Pohutugbeya teamProductivity Coach · Established Coaches

    New Zealand·Growing Team·1mo ago

    We killed the dashboard after one quarter. Now we track one thing: did the manager see a behaviour change within 30 days. One line per participant. Less data, more follow-up.

  • Editorial portrait artwork0
    Johanna Hoffmanngbeya teamLanguage Tutor · sponsorship & ads

    Germany·Scaling·1mo ago

    You did not have real learning analytics. Completions and satisfaction are activity logs. You need xAPI statements, time-per-slide, pause events. We saw engagement jump once we tracked those.

  • Editorial portrait artwork0
    Larissa Santosgbeya teamHR Consultant · group programme

    Brazil·Scaling·1mo ago

    We tracked time-per-slide and got more events nobody used. Johanna, which decision did that improve? Kwabena, for us the turning point was agreeing with managers what changed meant before the training ran.

  • Editorial portrait artwork0
    Nyokabi Weregbeya teamContent Strategist · paid newsletter

    Kenya·Scaling·1mo ago

    My newsletter's next issue covers this exact measurement trap. I will send you the template.

  • Editorial portrait artwork0
    Elise Brennangbeya teamPublic Speaking Coach · Speakers & Thought Leaders

    United States·Established Solo·1mo ago

    The real problem was the board presentation did not make the cost of doing nothing visible. One slide, one number, one gasp. If nobody in the room feels it, the data will not matter.

Highlights from this conversation

What worked

  • “We tracked time-per-slide and got more events nobody used.” Larissa Santos

What to watch

  • “The academy renewed, but follow-through stayed weak.” Kwabena Mensah
  • “The real problem was the board presentation did not make the cost of doing nothing visible.” Elise Brennan

Explore this topic

This case study sits in Proof & Progress, which holds 938 conversations. Evidence of what works: wins, experiments, lessons learned, build-in-public updates and case breakdowns.

Related discussions

Go deeper

Reference pages that cover the practical side of this topic.

Browse every public discussion

Building a business around what you know?

The Network is free to join. Ask your own questions, answer others, and find people doing the same work.

See the Network