What's New Across the Mfg Value Platform This Week (August 9, 2026)

Improvement programs now get marked against what they predicted, and the value analysis moves in alongside your assessment data.

An improvement program can now tell you what capability it should build, and then be marked against what your next assessment actually measured. That is the platform grading itself, on your data rather than on a case study, and it is the thing we have most wanted to be able to do.

It arrived alongside a move. The value analysis now runs on mfgvalueplatform.com, so your maturity data, the programs built from it and the money behind them are one workflow rather than a job split across two sites.

Both of those, and a good deal else, below.

Every program now projects the capability it should build

Your maturity scores have always fed the value math in one direction. A plant already strong in an area has less room to gain there, so the dollars are scaled down accordingly. Nothing ran that relationship the other way, which meant a program could tell you what it was worth and which KPIs it would move, but not where it should leave the assessment scores your teams gave you.

Each program now projects where the areas it addresses should sit on that same 1 to 5 scale, both once every solution is live and adopted and how much of that should have landed already. Underneath sits the detail: each area, where the assessment put it, where it should end up, and what share of it this program's solutions actually deliver. The rest needs work the program does not include, which is worth knowing before you start rather than a year in.

Two numbers behind it cannot be derived from your data, so they are stated on the page rather than buried. A fully adopted program is assumed to reach world class rather than a flawless score, because no plant sustains a perfect five. And practice maturity is assumed to take around three years to fully land after go-live, because capability moves when people change how they work, which is slower than installing anything. Timing is counted per solution rather than per program, so a program whose platform lands in year two shows only the movement its faster work has genuinely produced.

And then the program gets marked against that projection

A program that has been started, and whose plant has completed an assessment since, now gains a section putting the two side by side: what it projected for the areas it targets, and what your own people then measured.

The whole point is that it can come out badly. A projection nobody ever checks is marketing. One that gets checked every quarter is a model.

Most of the difficulty is comparing like for like, so three things are deliberate. A program is graded on how much of it had actually gone live by the day that assessment closed, not on its finished promise, so a program five months into a two-year plan is measured against five months of work. Both sides name their cohort, so it is one specific round against one specific round, never a blend of whoever answered most recently. And every result sits beside what the areas the program does not touch did over the same two assessments, because plants improve anyway, and a program that moves its targets half a point in a company drifting up half a point everywhere has demonstrated nothing.

Where two rounds have closed you get a trajectory rather than a verdict, and it has already told us something. The projection runs ahead of reality at two months and behind it at five, which is what a straight line does against adoption that starts slowly and picks up. That is recorded and left alone rather than tuned, because it is not enough evidence to change an assumption on.

Programs also finally have a list, and it is a review rather than a menu: every program in your organization on one page, showing what was committed to, how it got there, and what actually happened. Once work is underway, "did it land?" is the first thing on the page rather than the last. The figures a program was approved on are now frozen as a record while the live estimate carries on improving, and where the two drift you see both with the reason, because silently restating a number a customer approved is the fastest way to lose their trust in it.

The value analysis now sits with the data it draws on

Baselines, assumptions, scenarios, the audit walk, comparison, the investment plan, programs and reports have moved onto mfgvalueplatform.com, alongside the assessments and committed programs they are built from.

The point of the move is the workflow. Pricing an improvement means reading your maturity results, and grading a program means reading them again a quarter later, so that loop was crossing a site boundary two or three times in a single sitting. Closing an assessment cycle now offers to freeze a baseline in the same place you closed it. Analysing a selection shows the capability projection next to the causal detail behind it, rather than two pages and one domain away. Generating a URS or a value report lands in Reports with everything else.

There are still three sites: Use Cases, Surveys, and the Value Platform. What has gone is the fourth. Old valuemaps.com links forward path for path, so bookmarks and links in past posts still land where they should. Two pages changed name on the way across: Roadmap is now the Investment plan, and Deliverables is now Reports. Sign-in is unchanged, since it was always one account across the properties.

Navigation across the top, and a workspace that shows what needs attention

The platform has had exactly one navigation item since it was built, with everything else reached by going through the workspace and hunting for the right card. That is a reasonable design at four destinations and the wrong one at forty. There are now seven groups across the top: Workspace, Assessment, Opportunities, Company, Analysis, Programs and Deliverables.

Three things you do on a cadence have come up out of company settings. Opening and closing an assessment cycle decides which cohort every maturity read and every trend is taken from. Chasing the people who still owe you an assessment is the recurring job during a collection round. Your plant and division structure is the lens every other page is read through. All three were buried because they are admin-only, and being admin-only had been allowed to imply being a setting.

The workspace page itself used to open on three cards explaining where things live. The menu does that now, so the page lists what needs your attention instead: decisions somebody is waiting on, figures that are no longer true, and work that is available. It is ordered by who is blocked rather than by what is newest, and counts say what they are out of, so you get "3 of your 11 lists have not become initiatives" rather than a bare three.

We no longer recommend what you already own, and running costs are now charged

Three related corrections, and the honest summary is that returns come down.

A solution your survey says is already installed used to make a program look better than it was. The cost was correctly dropped to zero, but the benefit was still credited in full, so something you already owned appeared to deliver its whole value for free. Programs now withhold that value and name the systems responsible. It is credit for work already done, not a silent haircut.

You can also now tell the platform about work you have already committed to: what it puts in place, which site it covers, when it starts, and what is left to spend. The model treats it as ground already covered, so it stops recommending a second system where one is already being bought. It asks what remains to be paid rather than what has been spent, because money already spent is not a future cash flow.

And programs now carry their running costs. Subscription and maintenance used to be invisible for the entire horizon, so a solution that is cheap to buy and expensive to run outranked a capital purchase purely by not being charged for the difference. Every return figure is now net of them, charged from go-live rather than from the day the cheque is signed. Expect some things that looked marginally worthwhile on purchase price alone to stop being worthwhile once the years of running them are counted. That was always the true figure.

Every saved figure now names the version of the math behind it

The value model gets corrected from time to time, and saved reports, plans and target runs keep the figures from the day they were generated. So after a correction some of them are quietly out of date while sitting next to current ones and looking equally authoritative.

Those surfaces now say which version produced them. Anything superseded is marked and refreshes on the next resync, and anything saved before this existed reads as unknown rather than guessing in either direction. On a report you hand to a board the note prints with the document, because that is where the question gets asked and the app is not there to answer it.

This is not hypothetical housekeeping. A stored figure that could not name its own vintage is how one program's headline number drifted to roughly eight times what the current model says, unnoticed. The corrected values are already in place.

You can now walk the whole product without an account

There used to be three separate ExampleCo demos, one per site, each a hand-built imitation of the product that had to be rebuilt every time a real feature shipped. There is now one, it is the actual product with a fictional company's data in it, and every old link still reaches it.

It also walks you through itself. Five steps, each saying what you are looking at and why, opening on the verdict: a program committed to months ago, where the platform predicted the capability gain and a later assessment measured what happened. From there it works backwards through the survey answers those gaps came from, the work that got chosen, what it costs and when it pays back, and the documents you would hand a board. "Explore on my own" turns the guidance off. Everything is read-only, an amber bar names whose numbers you are looking at, and one click leaves.

mfgvalueplatform.com/demo/exampleco

One thing to say clearly, since the demo now leads with a measured result. ExampleCo is a fictional manufacturer we built to demonstrate the method. Its plants, its respondents and its improvement trajectory were written by us on purpose, including a program that plainly did not land, because a demonstration where everything works is not a demonstration. Those assessment answers were authored, not collected. What is not authored is the computation: the projection, the grading, the cohort matching, the control and the privacy floors all run exactly as they will on your data. It shows you the instrument working end to end. It is not evidence that our projections hold up in the field, and when we have a real customer's quarters we will publish those, including the parts we get wrong.

The maturity dashboard opens about four times faster

For a company with a lot of assessment data it took ten to thirteen seconds to open, and now takes about two and a half to three. Most of the wait was work nobody had asked for: the page built all nine of its tabs every time and then showed you one. Each tab now has its own web address as well, so you can send someone straight to the Consensus view or the trend chart.

Smaller things

Lists have a home. Your own saved lists and your company's now open on the platform in one place, with rename, delete and share on the card. It was the last menu item that sent you off the site.
The capability fingerprint reads three ways - against use cases, strategic goals, or KPIs. Reading a 178-use-case list against goals collapses it to 35 columns, which is the point.
Menus on mfgusecases.com open on a click rather than when your pointer passes over them, and behave properly on a touch screen.
Consultants stay in the client's context as they move around, instead of being dropped back into their own company by every menu click.
Fixed: previewing the ranked gaps on the maturity Gaps tab reported "Unauthorized" to anyone without permission to create lists. Ranking and creating were sharing one address and being treated as the same action. Ranking is reading, and it is treated that way now.

Questions, feedback, or want a walkthrough? Reply to this email, or hit the Feedback button on any of the sites. We read everything.

— Tim Stuart, Visual Decisions