A plant manager gives a monitoring deployment about five minutes of real attention in week one, between a shift change and whatever fire is already burning that morning. Whoever’s running the pilot gets one shot to show something that earns a second look. Showing everything the system can do is the wrong instinct, because a plant manager staring at a dozen tiles and three chart types in that five minutes learns nothing except that this is one more piece of software to eventually ignore. Showing the right three screens, in the right order, does the opposite.
The one number that matters more than any screen
Before any screen, there’s a single number worth saying out loud first: what a stop is costing, in dollars, on the specific machine the plant manager already worries about. Not an OEE percentage, not a utilization curve, a dollar figure tied to something they already know is a problem. “The changeover on line 2 ran twenty five minutes yesterday against a fifteen minute standard, and that’s roughly four hundred dollars in lost capacity on that run alone.” That sentence does more in ten seconds than any dashboard does in five minutes, because it’s specific, it’s checkable against what they already suspect, and it’s denominated in the currency a plant manager actually budgets in.
Every screen that follows should exist to support that number, not compete with it for attention. If week one’s presentation leads with a chart and buries the dollar figure in a footnote, it’s backwards.
Screen one: the machine they already worry about
Open with the specific asset that’s already on the plant manager’s mind, not a plant-wide overview. Every floor has one, the machine everyone already knows is a problem, running slow, breaking down more than it should, eating changeovers. Pull up that machine’s actual stop history for the week and let them see it directly: what happened, when, for how long. This is the screen that converts skepticism into attention, because it isn’t a claim about the system’s capability. It’s a mirror held up to something they already believed was true, now with specifics attached.
Resist the urge to open with a plant-wide summary instead. A rolled-up number across every machine is exactly the kind of thing that feels like it needs qualification and invites a debate about methodology before the plant manager has any reason to trust the system enough to have that debate productively. Start narrow, start with the thing they already suspect, and let the specificity do the convincing.
Screen two: the Pareto, ranked by dollars
Once the first machine has earned attention, the second screen should widen the view just enough to show where the rest of the losses sit, ranked by cost, not by frequency. A stop that happens fifty times a week for ninety seconds each and a stop that happens twice for twenty minutes each can cost roughly the same amount, and a Pareto sorted by count alone hides that. Sorted by dollars, the list tells a plant manager exactly where to point attention next, in an order that matches how they’d naturally think about spending time or budget to fix it.
This is also the screen where the pilot gets tested in front of the audience that matters most. A Pareto with a gap on it, a machine the system can’t measure yet showing as a clearly marked gap instead of a guessed value, tells a plant manager the system is being straight with them about its own limits. That’s worth more in week one than a complete-looking chart that’s wrong somewhere nobody’s checked yet, without anyone knowing it.
Screen three: the trend on one number, chosen together
The third screen isn’t a screen full of metrics. It’s a single trend line on the one number the plant manager and whoever’s running the pilot agreed to watch, ideally the same number tied to the machine from screen one. This is where week one turns into a habit instead of a one-time demo. A plant manager who leaves the room with one specific number they’ve agreed to check again next week is far more likely to open the dashboard again on their own than one who was shown everything and asked to pick a favorite.
Choosing that number together matters more than which number it ends up being. A changeover time, a specific machine’s downtime rate, a scrap percentage, whatever’s most tied to a decision the plant manager is already weighing. The point of screen three isn’t comprehensiveness. It’s commitment: one number, one owner, one date to check back.
Say the date out loud, too, not as a formality but because it changes the meeting from a demo into an open loop. “Let’s look at this same number next Tuesday” gives the plant manager a reason to think about the system between now and then, instead of filing week one away as something that happened once and might happen again eventually. A pilot that never sets a next date tends to drift, not because the data stopped being useful, but because nobody scheduled the moment to look at it again.
What not to show yet
Everything else can wait, and showing it in week one usually costs more than it earns. A full OEE breakdown across every machine on the floor, six tiles deep, is the classic overreach: technically impressive, cognitively expensive, and unlikely to survive past the first meeting because nobody’s had time to build trust in any single number inside it yet. Save it for once the floor already believes the system, not as the opening pitch.
Skip the AI-driven insights or predictive features too, however capable they are, in week one specifically. A prediction is only as trustworthy as the track record behind it, and there is no track record yet. Leading with a forecast before the plant has seen a single verified, past-tense number land correctly is asking for trust the system hasn’t earned. Let the first few weeks be about what already happened, stated correctly and specifically, before introducing anything that claims to know what happens next.
Also hold back the full configuration surface, every setting, every threshold, every shift rule available to tweak. A plant manager doesn’t need to see the machinery behind the number in week one. They need to trust the number. Save the tour of what’s configurable for whoever ends up owning the system day to day, usually a supervisor or a process engineer, not the person deciding whether to keep going.
Quick recap
- Lead with one dollar figure tied to a specific stop on a machine the plant manager already worries about, not a percentage
- Screen one is that specific machine’s actual stop history for the week, not a plant-wide overview
- Screen two is a downtime Pareto ranked by dollars, with clearly marked gaps shown instead of filled in guesses
- Screen three is a trend line on one number chosen together, with a date to check back on it
- Hold back the full OEE breakdown, predictive features, and the configuration surface until trust in the basics is already built
- The goal of week one is one number the plant manager agrees to keep watching, not a complete tour of the system