Reading results
What the answer means, where every number came from, and how to tell a real result from a thin one.
A solve gives you one number and a lot of context. This page covers how to read both.
The headline
Every result leads with the objective: Best answer: $6,400 — the value of the best plan the model found, in your currency, alongside solve time and iteration count. NPV-style models show Best NPV with the discount rate in the tile label. When a model operates in one currency and reports in another, the panel states the base currency and the rate used.
The narrative
Underneath sits the result in words — what the plan does, what it's worth, and which limits are binding. The Candle Workshop example reads:
The optimal plan makes 300 Classic and 100 Deluxe candles for $6,400 total contribution ($3,600 Classic + $2,800 Deluxe). BOTH constraints bind simultaneously: all 200 kg of wax (120 kg Classic + 80 kg Deluxe) and all 300 labor hours (150 hr each) are fully consumed.
"Binding" is the useful part: a constraint that's fully consumed is what's stopping a better answer. Those are the levers worth attacking — add wax, add hours, or change the mix.
KPI tiles
Each KPI shows its value plus what it means in your business — "What the workshop clears each
month after wax, wick, and labor spend on the optimal candle mix." Large currency values
abbreviate ($1.24M) with the full figure on hover. Against Base, each tile shows its delta:
= base— unchanged.+$420 vs base— currency delta.+3.2 pp vs base— percentage-point delta.
Click a tile to trace the number to the parts of the model that produced it.
Value-of-more tiles
Some KPIs answer "what would more of this be worth?" — for example "Value of an extra production hour: how much more profit one additional hour of production capacity would unlock — flags whether adding shifts pays off." Read these as the price you should be willing to pay for more capacity, not as a forecast.
When there's no answer
An infeasible scenario means your requirements collide — no plan satisfies all of them. Diagnose explains which requirements conflict, the size of the shortfall, and what relief would unblock it. See Scenarios & what-ifs.
How much to trust it
Two mechanisms keep results honest:
- Verify builds an evidence record — the result, the inputs it used, the checks run, and the trail — exportable for review outside eode. Results start as Estimated; verification re-loads the model, re-solves, and pulls a deterministic per-component breakdown to reach Verified.
- RIA Review (on workflow-run stages) tags each of its findings with provenance — solver, rule, your edit, vs base — and a confidence level: certain, likely, or worth a look.
If RIA flags a caveat about its own model, that's the system telling you where the model is thin. Read the caveat before you act on the number.