Processes: process is not sacred|INEMA.CLUB-PRO

Lesson 6 · 20 min

Baseline and target

Every Monday Carlos opens the agent's dashboard and says: "look how it's climbing!" It really is climbing. But the dashboard started counting the day the agent came on board, and it only shows volume. Without the before, any rise looks like a win, and any redesign has no proof. In this lesson you pick three indicators with a denominator and record today's value before you change anything.

Start the lesson
The business owner and the sales manager look at a printed bar chart and a line chart on the laptop; she points to the start of the chart.

In this lesson you will

  1. 1Why volume goes up without proving anything
  2. 2Write an indicator as a fraction, with a period and a source
  3. 3Mark what is observed and what is a hypothesis
  4. 4A target with a deadline and the guardrail for what can't get worse

Objective: define 3 indicators with a numerator, a denominator, and today's value, observed or marked as a hypothesis, before changing the process.

01An indicator in five parts

From the question to the guardrail. Tap each part.

Click each block of the diagram to see what it does.

Indicator rules from the Gestoria method (INEMA): always period, population, denominator, and source; test, pilot, and production kept separate.

02From the dashboard that climbs to the baseline

The dashboard numbers come from the lab (fictional study company). Open the piece and check.

1. The dashboard that climbs

Hora's dashboard shows conversations rising from 702 (week of 09/01) to 901 (week of 10/13), and closed orders from 141 to 168. The collection started the same day the agent came on board. There is no before and no comparison.

Satisfaction comes from a star-rating button that about 6% of customers answer. An average of those who answered doesn't speak for the other 94%.

The math the dashboard doesn't doOrders ÷ conversations: 141 ÷ 702 = 20.1% on 09/01; 168 ÷ 901 = 18.6% on 10/13. Volume went up and conversion went down. "Climbing" doesn't prove the agent helped.

03Templates to copy: indicator sheet and baseline chart

One sheet per indicator; the chart brings the three together, week by week. The tabs show Horizonte.

Indicator sheet
INDICATOR · <short name> · process: <...> · indicator owner: <name>

Business question: <what we want to know>
Numerator: <what is counted>
Denominator: <the universe in which it is counted>
Formula: <numerator> ÷ <denominator>  (or median / p95, for time)
Period: <week / 4 weeks / month> · compare with: <similar period>
Source: <system and field> · who extracts it: <name>

TODAY: <value> · [ ] observed (counted from <date> to <date>, n = <...>)
                [ ] hypothesis (source: <...>) · becomes observed on: <date>
GOAL: <value> by <date> · decided by: <name>
GUARDRAIL: <what cannot get worse> · limit: <value>
DO NOT CONFUSE WITH: <what this number does not show>
Baseline chart (4 weeks)
BASELINE · process: <...> · version in use: <1.0> · from <date> to <date>

| Week | Cases (denominator) | Indicator 1 | Indicator 2 | Indicator 3 | Guardrail | Note (holiday, promotion, out of stock) |
|---|---|---|---|---|---|---|
| <dd/mm> | <n> | <...> | <...> | <...> | <...> | <...> |
| <dd/mm> | <n> | <...> | <...> | <...> | <...> | <...> |
| <dd/mm> | <n> | <...> | <...> | <...> | <...> | <...> |
| <dd/mm> | <n> | <...> | <...> | <...> | <...> | <...> |

Did nothing in the process change during this period? [ ] yes  [ ] no: <what, when>
Signed by: <process owner> on <date>
Prompt · what this indicator hides
You are an indicator analyst. Below are the sheets for the indicators
I chose to measure a process before and after redesigning it.
Do not pick others. For each sheet:

1. Say whether the denominator is clear and whether it can change on its own
   and mislead the comparison.
2. Give a way for the number to improve without the process improving.
3. Say whether today's value is marked as observed or hypothesis and
   whether the source supports the mark.
4. Suggest a guardrail, if the sheet doesn't have one.

Sheets:
"""<paste>"""
Simulation: illustrative example

WhatsApp order → invoice · baseline · Horizonte, fictional study company

IndicatorFractionTodayGoal
Time until the invoicemedian and p95 of (invoice − confirmed order), orders invoiced in the week30 h (hypothesis: 5 cases)24 h or less
Returns from invoicingsent back to customer service ÷ reached invoicing, in the week~9% (hypothesis: stated)3% or less
People minutes per ordersum of people minutes ÷ orders invoiced21 min (hypothesis: 5 cases)under 15

Guardrail: orders with a discount outside the policy = 0 per week. Becomes observed: count in the app and in the ERP for four weeks, before version 2.0. Goal for: 8 weeks after version 2.0. Goals and deadlines: fictional study decision.

04Practice

Fix Horizonte's dashboard; then measure your own process before changing it.

1 · Horizonte's dashboard, corrected

25 min

Open the dashboard in the lab. Calculate orders ÷ conversations for the first and the last week. Then write the three indicator sheets for the process from the order to the invoice, plus the guardrail. Finish with a three-sentence message to Carlos explaining why "goes up" does not answer whether the agent helped.

Done when: both conversions are calculated, the 3 sheets have numerator, denominator, period, source, today's value marked as observed or hypothesis (with a date to become observed), goal and deadline, there is 1 guardrail with a limit, and the message to Carlos has 3 sentences.

2 · Your process baseline

4 weeks

Write the three indicator sheets for your process and start the baseline chart. Don't change anything in the process during the four weeks; if something changes, write it in the notes column.

Done when: the 3 sheets are written with today's value as observed or hypothesis, and the chart has 4 weeks filled in with each week's denominator and the process owner's signature.

Lab piece: indicator dashboard (F11, fictional study company).

05Quick test

5 questions. The explanation appears after your attempt.

Question 1 of 5

The agent's conversations went from 702 to 901 a week. What does that prove?

06Summary

Volume misleadsWithout the before and without a denominator, "goes up" proves nothing.
Full fractionNumerator, denominator, period, and source.
Today, markedObserved over four counted weeks, or a dated hypothesis.
Target and guardrailValue and deadline, without worsening what cannot get worse.

To go further

Laboratório Horizonte: the dashboard with no baseline (fictional study company)Copilot + Agents for Business: indicators that turn into actionAI Management Area: what the manager measures
Previous lessonNext lesson