How to use DecisionMind

Type a sentence like "choose a supplier for our new line" and press the arrow. The consultant answers in the same box, asks what it needs, and opens the right step when your data is in. Nothing you write is a command to the system; the system only runs when you press Run in the last step.

I want to add my data opens the file picker (CSV, Excel, SPSS). I want to enter my data by hand opens the table editor under the box, with a link to templates and sample data. Both lead to the same data check.

From the data check onward the consultant sits in a rail on the right. It answers questions about the step you are on and only names methods the system found for your data; it never invents a number. Suggested questions sit above the input, and you can hide the rail with the edge tab when you need the width.

The card names the file, counts alternatives and criteria and states the data type it detected. The table under it is your data as parsed, the first rows with a link to the rest.

Every column gets a role: a criterion, or the identifier that names the alternatives. Efficiency goals need at least one input and one output column, and the roles menu offers those when the goal calls for them. A column you do not want in the analysis is left out from its role menu.

The small bars at the end of each row draw the spread of that column with its lowest and highest value. A column whose bars are all the same height will not separate the alternatives; the system tells you so in the warnings.

Under the sentence, What the system will do lists the work in order, and What you get lists the outputs. They change as you change the sentence, so you know before you continue what the run will produce.

The sentence stays on top of the next steps as a plan line. Change the goal opens this step again; the steps after it are redone because the method candidates depend on the goal.

Details opens the method's page from the library: authors, year, data type, the worked example and the citation. Select makes that card the primary method; Selected marks the current one.

Under the cards you can add methods that run on the same data alongside the primary one. At the end of the run their rankings appear side by side with an agreement measure, and the report notes where they differ.

Some methods ask for more than weights: thresholds, preference functions, a best and a worst criterion. When the primary method is one of them, this step adds the form the method needs, with the meaning of each field.

Each card draws the weights it would produce, criterion by criterion, before you commit; the numbers are the ones the run will use.

Every part carries what it holds and a Change link that opens its step. The line under the cards shows which parts are set and that the report is still to come. Report language sets the language of the report the run generates; you can generate another language later from the results page.

The run scene names the stages the system passes: reading the data, the value space, the method steps, the quality gates, the report. A run that takes long shows a note instead of a spinner; nothing is faked.

Analysis complete opens the result: the top alternative, a sentence on how far ahead it is and whether the system reads that gap as narrow, the ranking with scores, the method and the weighting used.

When you added methods to compare, a table puts the rankings side by side and an agreement coefficient says how far they concur. Disagreement is a finding, not an error; the report discusses it.

PDF, Word and Excel download from the card. Review the results opens the results page, where every panel of the analysis lives.

The method, the number of alternatives and criteria and the value space, then the three download buttons and the report language. Downloads from here are the same files as in the wizard.

The rail on the right answers questions about this run only and reads only the numbers this run produced: why an alternative won, what the sensitivity panel means, how to report the result.

All, Completed and Draft narrow the list; the search box matches names, methods and file names. Rows are grouped by day.

The sidebar keeps your latest work one click away with its own search; All analyses opens this page. A dot next to a name shows a run still in progress.

One blank decision matrix per data type, in that type's own cell syntax, as Excel or CSV. Fill it in by hand opens the same table on screen instead.

A ready table per data type. Opening one creates a draft and lands on the data check, so you can rehearse the wizard end to end.

Search by name or id, narrow by family and category, pick a method: its page shows authors, year, data type, the definition, a worked example with the formulas, and the assumptions and limits. Start analysis with this method opens the wizard with that method preselected.

The same methods are listed by family on the public Methods page, one page per method, so a colleague without an account can read what you used.

Open Method Lab from the sidebar and pick one of your finished analyses; its data, goal and weights flow in. The lab is part of the plans that include it; the pricing page lists which.

Choose methods from the pools, read each one's fitness and radar, tune the parameters and the validation tests, run them together or separately, and merge their rankings when you need one answer.

Each completed run counts one analysis; drafts do not count until you press Run. The count resets monthly on the paid plans. When it runs out the wizard says so at the Run step and offers the next plan; nothing you already ran is lost.

The free plan lets you run a full analysis and read its results before you subscribe; the pricing page lists its limits on data size and method families.

Account, then Billing: the active plan, the invoices and the payment method. Choose a plan opens the pricing page.