What's New Across the Mfg Value Platform This Week (August 16, 2026)
Subject line: Your own KPI levels now drive the value figures, and the Importance tab has been redrawn.
An improvement used to be describable only as a percentage move. "One percent better first-pass yield" is useful, but it is not a sentence anyone takes into a meeting.
A baseline now records where your own operating numbers actually stood, so that becomes "one point of first-pass yield is worth $90K a year here", computed against your finance lines rather than a generic model. Every value figure an improvement program projects can now be stated in your numbers rather than in generic ones.
The Importance tab has also been redrawn, and it is worth a look. Both below.
Your KPI levels, and what a point of each is worth
Saving a baseline seeds your levels from published industry typicals for your industry, and an optional section on the form lets you replace any of them with your real numbers. Leaving it untouched changes nothing.
It covers all 172 KPIs, not only the ones with a published typical. For a good half of them, including most management, workforce and safety measures, no industry benchmark exists anywhere, so your own figure is the only honest source there will ever be.
The list is ordered by what a move is worth, so the measures that move your money sit at the top. Two limits are stated on the page rather than left to be inferred. It is priced for the conservative, demand-constrained case, so a sold-out plant's volume-driven measures are worth more than shown. And it ranks by what a move is worth, never by what it costs to achieve, because a point of OEE and a day of DSO are not the same amount of work.
Units are shown rather than typed. Every field renders its unit as fixed text beside the input, because 85 and 0.85 are both plausible readings of OEE and differ by a hundredfold in every dollar that follows. Where a published range exists, a value outside it is flagged and still saved. A plant outside the typical range is exactly the plant worth modelling.
Results are marked as results. Profit margin, budget variance and standard cost variance price enormously, because they sit more or less directly on the P&L. They are also things nobody can run an improvement program on, so they are labelled and sorted below the levers rather than crowding the top of a list meant to answer "what should we work on".
The Importance tab, redrawn
The tab now opens on an area chart with two lines, most important first: what the business says matters, and how consistently you do it today. You read whether the second line keeps up with the first, and where it dives away is where capability has fallen behind your own priorities. It is the shape of your operation on one screen, in the order you said things matter.
Every row also shows where your team actually landed, one dot per respondent, with a marker for the row's own score. A tight clump means the team agrees. Spread out means they do not, and that disagreement is often worth more than the average. Where the row's score sits outside the body of those dots, the average is an artifact of averaging rather than something a person reported.
The chart underneath now moves with the table. It used to carry its own plant and assessment pickers, so the two could sit there disagreeing about what they were showing. One set of controls scopes the whole page. It earns its place by showing what a ranked list cannot: among the things you rate equally important, capability still varies by one to two full points, and no ordering can show you that. The difference is between knowing your priorities in order and knowing their shape.
Every column heading now explains itself on hover, including what the gap score is made of and why importance acts as a weight on the shortfall rather than something subtracted from it.
Multi-plant companies can drill by plant
Pick a division and you get everything beneath it, not just work booked directly to it. Narrowing the view genuinely protects fewer people, so more rows fall below the privacy threshold and are listed by name and response count instead, with advice on what would open them up.
Open any department to see what is inside it
Departments, CESMII areas and use-case families all open now. Clicking a row lists the use cases underneath it, so "Quality sits half a point below where it needs to be" becomes a list of the specific capabilities producing that half point.
Rows below the privacy threshold open too, which is where this earns its keep. A department that cannot be scored yet still tells you what is waiting inside it and how many more responses each item needs.
Corrections
A small team's gaps are no longer harder to find than a large one's. If two or three people rated something, it used to have to clear a higher bar to count as a gap than if twenty had. Same answers, higher standard, purely for being a smaller team.
The old rule used two separate tests, on how important something is and on how mature, and both were quietly tightened when fewer people had rated an item. The intention was caution with thin data. The effect was a penalty for being small.
There is now a single test, on the gap score the page already displays, and it eases as respondents accumulate rather than tightening. Getting more of your team to assess opens up more of your own results instead of raising the standard on you. On our first small manufacturer the priority list went from eight items to nineteen. The missing eleven were always there.
The Importance tab and the Gaps page now rank identically. The two pages were ordering the same data differently, because the Importance tab was combining your two ratings in a way the Gaps page did not. They use one number now, and it is the one the Gaps page has always shown.
Sixty industry typicals were showing a hundred times too small. They were stored as fractions where the rest were whole numbers, so a cost of poor quality of 4% displayed as 0.04%. Corrected row by row rather than by a blanket rule, because a handful of measures genuinely do sit below one percent. Baselines saved before the correction are unchanged, because a baseline is a record of what was known when it was taken.
The privacy threshold now counts people who answered rather than people who joined. Someone who joins and assesses nothing adds no answer for anyone to hide behind. Checked against live data before shipping, and no company was affected, so nothing that renders today stops rendering. It closes a gap that would have opened as workspaces grew.
Coming soon
Rows where your team has clearly split into two camps will say so, rather than reporting the average of an argument. And we are working through what a fair privacy threshold looks like for statistics that stand on more data than an average does, which is what currently keeps small departments dark.
— Tim Stuart, Visual Decisions