Havish M Consulting
Customer Case Study · HR · Succession Planning
Retail × Havish M Consulting

Succession Planning Workbench

One live workbench for every concept: talent KPIs, a 9-box that moves year on year, a succession pipeline for every critical role, and GenAI summaries of the talent notes, in a single HTML file.

ClientRetail FunctionHR ThemeSuccession Planning
Talent review cycle
0weeks
▼ from about 9 weeks
Critical roles with a ready successor
0%
▲ from 42% the year before
Concepts on one view
0
▲ instead of eight spreadsheets
Calibration prep time
−0%
▼ AI summaries of talent notes

Figures indicative; validated metrics available on request.

The Challenge

Succession lived in eight spreadsheets and a lot of free text

Each concept kept its own talent file, with 35 columns from performance history to CEO comments. Pulling it together for the annual talent review took weeks, and the most useful information was in long notes nobody had time to read.

Fragmented data

Every concept had its own version of the template, with different spellings, bands and rating scales.

Notes nobody read

Strengths, development areas and successor notes ran to paragraphs. Calibration meetings ran out of time before they reached them.

Hidden risk

Critical roles without a successor, held by someone at high flight risk, were not visible until someone resigned.

Slow review cycle

Consolidating, cleaning and building slides for the CEO review took about nine weeks every year.

HMC Approach

Clean once, view many ways, summarise with AI

We kept HR's template and rating scales, cleaned the data once, and built the views the CEO, concept heads and HR business partners actually ask for.

Select a step to see what happened.

Discover · Read every column

We mapped the 35-column template with group HR and concept HR heads.

  • Agreed definitions: critical role, HIPO, consistent performer, weak link.
  • Listed the questions the CEO asks in a talent review, and which columns answer them.
  • Found the cleaning rules: band, function and concept names.

Deliverable: KPI dictionary and data-cleaning rules.

Design · Six views, one dataset

Each view answers one kind of question.

  • Concept view and KPI dashboard for "where is our risk?".
  • Employee list and 9-box for "who are our people?".
  • Succession pipeline and full data view for "who is next?" and "show me the detail".

Deliverable: Wireframes signed off by Group HR.

Build · A single offline file

Built with Claude as one HTML file: no server, no install.

  • Filters by concept, role level and year, carried across views.
  • Click any KPI to see the people behind the number.
  • Optional GenAI summaries using HR's own API key and prompt.

Deliverable: Succession workbench and refresh guide.

Adopt · Run the talent review from it

Calibration and the CEO review ran live from the workbench.

  • HRBPs set status, priority and remarks during calibration.
  • AI summaries replaced reading paragraphs aloud.
  • Actions agreed in the room, exported straight after.

Deliverable: Calibrated talent decisions for all concepts.

Scale · From annual to always-on

The same file is refreshed each quarter.

  • Quarterly refresh of ratings, attrition risk and readiness.
  • Year-on-year 9-box movement tracked automatically.
  • Next: link development plans and retention actions to outcomes.

Deliverable: Quarterly talent pulse.

The Solution

The workbench. Click around.

Six tabs on one dataset of 35 columns. Filter by concept, click any KPI to see the people behind it, switch the 9-box year to watch people move, open anyone's profile, and use ⚙ AI settings to add your own OpenAI key and prompt for summaries.

LG
Talent & Succession WorkbenchIllustrative data ·

All employees, ratings and notes are fictional and generated for this case study. Faces are generated illustrations. Concept names are used for context only.

Impact Dashboard

Deeper benches, faster reviews

With risk visible by concept and notes summarised in seconds, the review spent its time on decisions, and more critical roles ended the year with a ready successor.

Critical roles with a ready successor

% of critical roles with a Ready Now or Ready Soon successor, by concept

Where the review time went

Talent review cycle by stage, weeks

Data collection & cleaningCalibrationSuccession mappingCEO review pack

Figures indicative; validated metrics available on request.

Before vs After

How a talent review runs

DimensionBeforeAfter
DataEight concept spreadsheets, different formatsOne cleaned dataset, 35 columns, every concept
Risk viewBuilt by hand for the CEO deckLive KPIs; click any number to see the people
9-boxStatic slide, one yearInteractive grid with year-on-year movement
Talent notesParagraphs read aloud in calibrationAI summaries with HR's own prompt
SuccessionNames in a notes columnPipeline by readiness, gaps highlighted
Cycle timeAbout 9 weeksAbout 3 weeks
Tools & Tech Used

Light, private, offline-first

Claude data cleaning & workbench build OpenAI API optional summaries with HR's own key Excel / HRMS extract source data D3.js 9-box, charts & drill-downs Single HTML file no server, no install
Key Takeaways

What transfers to any HR team

01

Define the KPIs before the dashboard

Agreeing what "critical", "HIPO" and "weak link" mean did more for the review than any chart.

02

Every number should open to names

Leaders trust a KPI when one click shows the people behind it.

03

Let AI read the notes

The richest talent data is free text. Summaries in HR's own words put it back into the conversation.