What's fixed, what's left, what's new

I wanted to change this last three months and I change it to something else. Then it resets all of this stuff and you have to start… slightly irritating, right?

Jason Vallier, demonstrating the current product

In the current product, filters must be applied in an implicit order and changing one resets the others. It is the most-cited frustration with the tool.

One refinement worth considering: an applied filter has no visible way to remove it. Adding a clear affordance to each control would finish the pattern.

The current benchmarking screen carries a numbered list of operating instructions above the form — ending at "8. Required – Click 'Retrieve'." Four required dropdowns, two separate Filter buttons that aren't the primary action, two sets of radio buttons, and a Retrieve button that is.

A form that needs instructions printed above it is telling you something. 2.0's filter pattern needs none.

Today a released report sits invisible for roughly four hours while a background job runs between Skylab and MySkydrol — and the person who released it must remember to return and release it a second time.

An engineer responsible for the A320 fleet currently receives every report for every Boeing and Airbus fleet. Profile preferences in 2.0 address this. The customer-facing export is also being brought closer to parity with what technical services can pull internally.

One design question remains on defaults: a new user who sets nothing should probably receive their own fleet rather than everything. Defaults do more work than settings.

The sample kits request history includes a Tracking Number column that is empty on every row — including orders marked Delivered. Scandinavian Airlines System, FedEx Corporate Aviation, PT. Lion Mentari, JSC Aircompany SCAT.

Orders route through technical services to a third-party fulfiller. Customers wanting status email support, who ask the team, who chase the fulfiller. Every one of those emails is a support cost with a known cause — landing on the team whose stated goal is "they're not calling me as much anymore."

Contact details prefill correctly — name, email, company. The entire shipping address does not: address line 1, city, state, postal code and country, all required, on every order.

This is a regulated, repeating workflow. The same operator samples the same fleet from the same facility on a schedule set by Boeing and Airbus. The system has shipped to these addresses before.

The Samples empty state reads "Once we receive your samples, they will show here. Order a sample kit to start sampling." — with Order a sample kit as plain text rather than a link. The one action the screen recommends isn't offered.

Meanwhile disabled Download PDF and Export to CSV buttons sit above it, giving visual prominence to two actions that cannot work on an empty page.

Small, but it's a first-run experience — the moment a new customer forms their impression of the product.

A role without home-page access is redirected to "You're not authorized", whose only action is a Go to Home button — returning to the page that just rejected it.

Role-based access is a stated 2.0 feature, which raises a question worth answering once rather than per-screen: how many roles exist, and does each have a coherent landing destination?

Read from the running build: the primary action colour is #B20077 and the declared critical colour is #B22126. They sit at nearly the same lightness in the same warm family.

In a product where red means the airline is required to act under a Boeing or Airbus limit, the colour used for ordinary buttons and links should not be adjacent to it — and under protanopia or deuteranopia the two converge further.

It doesn't tell me the number of samples it pulled from. It doesn't tell me that this is all the world samples of A320.

Jason Vallier, while demonstrating the charts he uses most

Confirmed visually in the current product: the charts plot fleet against world with no population size, no date range shown on the chart itself, and no indication of confidence.

The audience is systems engineers making airworthiness-adjacent judgments. A comparison against "the world" with no N is, to an engineer, an unsourced claim — the difference between you're above average and you're above the average of 1,240 samples across 89 operators.

In the current benchmarking charts, bar colour encodes fleet versus world — green for your fleet, yellow for the world, red for the limit. But the axis labels beneath those bars read FLEET AIRBUS BLUE, FLEET AIRBUS GREEN, FLEET AIRBUS YELLOW — because blue, green and yellow are the physical hydraulic lines on the aircraft.

So a bar labelled "BLUE" is drawn in green, and a bar labelled "GREEN" is also drawn in green. The colour channel and the words are encoding two different variables, and they happen to use the same vocabulary.

This is the single most confusing thing in the current charts, and it's the kind of collision that is very easy to reproduce accidentally in a rebuild.

If it goes past that, it turns red in our system — the airline is basically required to take action. And those limits are set by the airframe manufacturers, like Boeing and Airbus.

We'll flag it as yellow and just say, hey, keep an eye on this — but the airline is not required by any stretch to take any action.

Jason Vallier

These read as two steps on one severity scale. They aren't. One is a regulatory trigger set by the airframe manufacturer; the other is Eastman offering a courtesy. Collapsing the distinction risks a required action being treated as advisory, and a courtesy flag being treated as an obligation — which erodes trust in the flag.

A secondary benefit: attributing limits to Boeing and Airbus positions Eastman as the trusted interpreter of someone else's rules rather than the author of arbitrary ones.

Benchmarking and sample reports live in separate places, reached by separate navigation and separate filter-building.

The question "is this normal?" arises at exactly one moment — while looking at a number. Requiring the user to leave, navigate elsewhere and rebuild a filter to answer it means most never ask. The most valued feature in the product is unreachable from the moment it is most wanted.

Right now an airline can put a comment into the system, but the comment's only for them. I never see it, unless they send me an email.

Jason Vallier

The product invites input and then discards it. Worse, it looks like communication — so a customer may believe they've raised something with Eastman when nobody will read it. That isn't a missing feature; it's a false affordance, and it costs trust.

It also cuts against the thing identified as the competitive moat: "the thing that delineates us is the personal touch."

I do an annual business review with a lot of the engineers and directors, and one of the things I use is this tool. I'll take a snip picture of this and put it on my slide.

Jason Vallier

The benchmark charts are already a retention instrument — placed in front of airline decision-makers to support a premium price. Nobody has designed them to be presentable: no clean export, no title, no source line, no defensible sample count.

It has about 80% of what I can pull, but not 100%.

Jason Vallier, on the customer-facing export

Location is the notable omission, and at least one customer wants it specifically to detect bad sampling technique by station: "Let's say all the samples coming out of Chicago are terrible. Maybe that's the condition of not pulling the sample correctly."

That's a genuinely novel analysis — using fluid data to audit maintenance process rather than fluid condition. Worth confirming whether the widened 2.0 export includes it.

Aircraft Overdue predicts the next sampling cycle. "It's not something that anybody uses, though, that I know of."

Interesting rather than deletable. The underlying job — when is this aircraft due? — is real and mandated by Boeing and Airbus. If the tool answering it goes unused, the likely causes are discoverability, trust in the data, or that it isn't where the decision gets made. A dead feature usually means a live need aimed badly. Fifteen minutes of conversation would settle it.