First-pass yield gets treated as a single number to chase, when it's really the output of a chain: measurement has to be accurate before a capability study means anything, and a capability study has to be right before process-parameter tuning actually moves the needle. Skip a step and the "improvement" is often just noise settling differently. A structured playbook, worked in order, is what turns a vague yield target into a repeatable gain. FMCG plant managers can walk through this playbook against their own FPY data on iFactory AI to find where their chain is actually weak.
A 3 to 8 Point FPY Gain Starts With Getting the Chain Right
iFactory AI works through measurement, capability studies and process-parameter tuning in order, so an FPY gain is real and repeatable.
Step 1: Measurement Has to Be Right First
What counts as a "pass" must mean the same thing every shift.
Good and reject units counted the same way across lines.
FPY tied to a specific batch, shift and process window.
Step 2: What a Capability Study Actually Tells You
| Output | What It Means | What to Do With It |
|---|---|---|
| Process centered, tight spread | Process capable, running well | Protect current settings, monitor for drift |
| Process off-center, tight spread | Consistent but aimed wrong | Adjust the target setpoint |
| Process centered, wide spread | Aimed right but too variable | Investigate sources of variation |
| Process off-center, wide spread | Both aim and variation are problems | Address setpoint and variation together |
Run a Capability Study on Your Own Process
Bring recent process data to a 30-minute session and see where your line actually sits on this matrix.
Step 3: Parameter Tuning, One Variable at a Time
Identify which process variable correlates most with yield loss.
Change a single parameter, hold others steady.
Confirm FPY moved, and by how much, before the next change.
Set the new standard and watch for drift going forward.
Gut Feel Tuning vs Playbook Tuning
A Composite Scenario: The 5-Point Gain That Didn't Hold
Picture a yield improvement push that adjusts three process parameters at once during a single shift and reports a five-point FPY gain the next morning. By the following week, yield has drifted back to its old level and nobody can say which change actually mattered.
Working the playbook in order, one parameter at a time with a capability study anchoring the baseline, would have confirmed which single change drove the gain, letting the team lock it in instead of losing it within a week.
Where iFactory AI Fits
Validates measurement first
Confirms pass/reject counting is consistent before anything else is trusted.
Runs capability studies
Establishes whether the real issue is aim, variation, or both.
Ranks parameters by correlation
Points tuning efforts at the variable most tied to yield loss.
Confirms the gain holds
Monitors FPY after a change to verify the improvement sticks.
Expert Review
A process engineering perspective, reviewed for accuracy before publication: the "quick win" yield improvement that doesn't survive a week is one of the most common patterns I see, and it's almost always because measurement or capability work got skipped in the rush to tune. Starting with a proper capability study on your own line takes longer upfront and produces a gain that actually holds.
Frequently Asked Questions
What FPY gain is realistic from working through this playbook?
Gains vary significantly by starting point and process, with a 3 to 8 point improvement being a commonly cited range for plants with meaningful room to improve in measurement accuracy or process centering. A plant already well-optimized may see smaller gains, while one with poor measurement discipline could see more. See an estimate based on your own current FPY and process data in a short session.
Why does measurement accuracy matter so much before tuning anything?
If pass and reject counting isn't consistent across shifts or lines, any apparent yield change could just reflect a measurement inconsistency rather than a real process improvement, making it impossible to know whether a tuning change actually worked. Getting measurement right first is what makes every later step trustworthy. Ask for a review of your current measurement consistency across shifts.
How long does a proper capability study take to run?
This depends on your process's cycle time and how much data is needed to get a statistically reliable picture, but it generally requires capturing enough production runs to see the process's natural variation, not just a single good day. Rushing a capability study tends to produce a misleading baseline. Explore a realistic timeline for a capability study on your specific line.
How do we know which process parameter to tune first?
Ranking parameters by how closely they correlate with yield loss in your historical data is more reliable than guessing based on intuition or convenience, since the parameter that's easiest to adjust isn't always the one actually driving the loss. Try a correlation review of your own process parameters against FPY.
How do we make sure an FPY gain actually holds after the initial push?
Continued monitoring after a change is what distinguishes a real, durable gain from a temporary shift that drifts back. Locking in the new parameter as the standard and watching for drift, rather than declaring victory and moving on, is the step that's most often skipped. Join a session on setting up ongoing monitoring after your next tuning change.
Make Your Next Yield Gain One That Actually Holds
iFactory AI works measurement, capability and tuning in order, so an FPY improvement sticks. Book a walkthrough on your own process data.
