Decision analysis software helps you choose between options when the outcome is uncertain. It represents the choice as a decision tree or an influence diagram, attaches probabilities and payoffs to the branches, and works backwards through the structure — a process called rollback, or backward induction — to find the option with the best expected value. Then it does the part that matters more: it tests how sensitive that recommendation is to the assumptions behind it.
One disambiguation before we go further, because it wastes a lot of people's time. A “decision tree” in machine learning is a classifier: it learns rules from data to predict a label. A decision tree in decision analysis is a model of a choice — what you could do, what might then happen, and what each ending is worth. The two share a name and a shape and almost nothing else. This page is about the second kind. If you are looking for classification, none of the tools below are what you want.
The category is smaller and older than the Monte Carlo simulation market beside it, and it is unusually opaque: several products publish no price, one of the best-known can no longer be bought on its own, and the whole field has fewer than fifty public software reviews between it. That is the gap this comparison tries to close.
How we compared. Every feature cell below is backed by the vendor's own material — product pages, help indexes, manuals and issue trackers, all read in August 2026 and linked from each product's section. Where we could not find a feature documented, the table says “not documented” rather than “no”: a red mark is a strong claim, so we only make one when the vendor's own material supports it. ModelChoice is our product; we have held it to the same standard, including the cells where it loses.
Every price below was read on the vendor's own page and is quoted in the currency the vendor lists. Where a vendor publishes no price, the table says so rather than guessing. The detailed 29-criterion comparison follows in the next section.
Sources and retrieval dates are given in each product's section below. Prices change; the figures here are those the vendors published in August 2026.
How to read the table. means the feature is documented by the vendor. amber means it exists with a qualifier that changes the buying decision — the note in the cell says what it is. means the vendor's own material says the feature is absent or not offered. A grey — means we could not document it either way — it is an honest gap, not an accusation.
Notes. “TreePlan family” means TreePlan, SensIt and SimVoi together — three separate add-ins sold as the $179 Decision Analysis Toolkit. In TreeAge's world “utilities” usually means health-state utility values (QALY weights), which is a different concept from the risk-preference utility functions in that row; we could not find the latter documented and mark the cell accordingly. Every cell was checked against vendor material in August 2026 — the sources are linked in each product's section below. If we got a cell wrong, tell us and we will correct it.
ModelChoice is our own product, so treat this section accordingly — the rest of the page is written to be useful whether or not you buy it. It draws decision trees directly on an Excel worksheet, rolls them back as you type, and highlights the optimal path before you ask for it. The model stays an ordinary workbook, which means a colleague, an auditor or a client can open it and follow the logic without buying anything.
A ModelChoice tree on the worksheet, rolled back with the optimal path highlighted.
What it does that the others mostly do not: sixteen analyses run from the one tree. Rollback and the full risk profile; one-way and two-way sensitivity; robustness scoring, which tells you how far an assumption has to move before the recommendation flips and which assumption it is; value of information, so a study can be priced before it is commissioned; multi-criteria analysis with AHP weight elicitation for decisions that are not only about money; and a single-page Decision Brief for the committee, regenerated the moment an input changes. When a chance branch is genuinely continuous — a price, a demand, a duration — it becomes a distribution and the tree runs full Monte Carlo simulation through ModelRisk. It also ships with an MCP server, so an AI assistant can build, audit and analyse the tree for you.
What users say: nothing yet, and we will not pretend otherwise. ModelChoice was released in 2026 and has no third-party review history — the matrix above marks it “none yet” for the same reason it marks others. Judge it on the trial.
Limitations, honestly: it is Windows and Excel only, with no Mac version and no browser version. There is no Markov modelling and no influence-diagram view. Bayesian updating happens inside the EVII analysis rather than as a separate tree-reversal view like PrecisionTree's. The full manual is in-product (in seven languages) rather than on the web — something we intend to fix. And if your decision problem does not live near a spreadsheet — a large influence-diagram model with hundreds of variables, say — a standalone tool like DPL or Analytica is a better fit than an add-in.
Price: €950 per user per year maximum, one edition, everything included. Sixty industry example models ship with it, existing PrecisionTree models import in one click, and there is a fully functional 15-day trial.
Documentation: feature reference · release notes · worked EVPI/EVII example
PrecisionTree is the best-known decision tree add-in for Excel and, for many people, the product that defined the category. It builds trees and influence diagrams on the worksheet, rolls them back, and reports risk profiles, policy suggestions, decision tables, two-way sensitivity, tornado and spider charts, Bayesian revision and utility functions. Its one-click reversal of chance nodes using Bayes' rule — shown below — is a feature most rivals make you build by hand.
Image: Bayes-revised probabilities in PrecisionTree, from Lumivero's tutorial video series. Retrieved August 2026.
The thing to know before you shortlist it: you can no longer buy it on its own. Lumivero's web shop lists it only as a component of the DecisionTools Suite. That means decision trees come bundled with @RISK, StatTools, NeuralTools and Evolver, at €3,680 per year for Professional or €4,790 for Industrial (Lumivero web shop (shop.lumivero.com), retrieved August 2026). If you want the simulation engine too that is good value; if you wanted decision trees, you are buying five products to get one.
On value of information, be precise. PrecisionTree's documentation describes no value-of-information feature: a search of Lumivero's help site — which indexes the product thoroughly, with dedicated pages for sensitivity analysis, policy suggestion and decision tables — returns nothing for “value of information”, “EVPI” or “perfect information”. What Lumivero offers instead is a webinar, Introduction to Decision Analysis Using PrecisionTree: Measuring the Value of Information, which demonstrates how to build the comparison by hand as a second tree. That is a real technique and it works. It is simply not the same as a report you generate from the tree you already have.
What users say. The public record is small — five reviews across G2 and Capterra, rating the suite 4.6/5 and 5/5 respectively — and it repeats two themes:
“One of the best things about this suite is it’s ability to perform complex simulations. It is also incredibly easy to use.” — G2 review, education sector, March 2019
“The price is a little too high and the evaluation period is too little.” — the same review — a complaint that appears, in near-identical words, in all four G2 reviews
“Students loved the software… practical tools that can be used to solve real life problems”, though “setting up the problems for sensitivity analysis is the most difficult part for students.” — Capterra review by an assistant professor, September 2017
Other limitations: Windows only, and the Professional edition caps trees at 1,000 nodes. The suite as a whole is actively maintained — release 8.12 shipped in November 2025 — but the PrecisionTree component's own version history, published inside that same help set, records no release after 7.6.1 in September 2019. And note the documentation format: the current PrecisionTree help ships as a downloadable Windows CHM file rather than browsable web pages — Lumivero's own download page warns that Windows blocks CHM files by default.
Documentation: help portal · CHM manual downloads · tutorial videos · example models
DPL, from Syncopation Software, is the serious standalone decision analysis tool and the one with the most impressive customer list in this comparison — Shell, Chevron, Pfizer, AstraZeneca, Lockheed Martin and the big four consultancies all appear on the vendor's own site. Its signature is the dual view below: the same model as a compact influence diagram and as the decision tree it implies. It computes value of information and value of control in every edition, does Bayesian revision, draws one-way and two-way “rainbow” diagrams that mark where the optimal policy changes, and its tornado analysis runs eleven help topics deep. It can link to spreadsheets — including converting a spreadsheet model to its native language for speed — but the model itself lives outside Excel.
Image: one model, two views — decision tree and influence diagram in DPL. Syncopation Software, syncopation.com. Retrieved August 2026.
Price: $1,495 for a perpetual Professional licence, $1,995 for Enterprise (syncopation.com order page, retrieved August 2026). Two things are easy to miss. Portfolio analysis is a separate product at $8,995 — a six-fold step from the headline. And the licence includes only 60 days of support; after that, maintenance runs from $295 to $1,795 a year depending on edition.
What users say. Three G2 reviews, averaging 4.0/5, all from 2024, and they agree with each other:
“Influence Diagrams for visually modelling decision” — G2 review, July 2024, on what works
“High initial cost limits small organization access” — G2 review, June 2024
“Outdated and old looking user interface makes it difficult.” — G2 review, biotechnology, June 2024
A consultancy partner of the vendor puts the case for it more strongly — “industrial grade because it can use trees with hundreds of millions of end nodes” — which, allowing for the source, matches what the feature set suggests: this is the tool for very large, formally structured decision problems.
Limitations: Windows only, no Mac. Markov models are absent — the vendor's own help index has no entry for them. Training is instructor-led only, with no self-paced curriculum. And we could find no free reader or viewer, which means anyone who needs to check your model needs a licence of their own. For a model that will be reviewed by a client, a regulator or an auditor, that is a genuine constraint rather than a detail.
Documentation: DPL 9 user guide (PDF) · help index · all user guides · trial request
Analytica takes a different route: influence diagrams rather than trees, with hierarchical models you can expand and collapse. For large, structured problems with many interconnected variables, that scales far better than a tree, which grows exponentially with every node you add. It handles uncertainty with Monte Carlo simulation throughout, reads and writes Excel through built-in spreadsheet functions, and version 7.0 shipped in January 2026, so it is actively developed. The trade-off is the paradigm itself: decision trees are not native — the vendor's documentation is explicit that Analytica “uses an influence diagram paradigm instead”, and trees must be encoded by hand.
Image: an influence diagram of a rent-or-buy model in Analytica. Lumina Decision Systems, from the Analytica documentation wiki (“Creating Lucid Influence Diagrams”). Retrieved August 2026.
Its best feature is one nobody talks about: the free edition has no time limit and will open and run existing models of any size. The roughly 100-object cap applies to models you create, not models you read. That makes Analytica the easiest tool here to share work in — a reviewer downloads it and opens your model, free.
Price: Free; $1,000 per user per year for Professional; $2,000 for Developer; $4,000 for Optimizer (analytica.com pricing plans, retrieved August 2026). Subscription only — there is no perpetual option listed.
What users say. Analytica has by far the largest review base here — 33 reviews on G2 at 4.8/5 and four on Capterra at 5/5 — and the pattern is unusually consistent: enthusiasm for the modelling paradigm, irritation at the packaging.
“Models of similar complexity require anywhere from 1/4 to 1/2 the amount of time to build.” — Capterra review by a company president, comparing it with spreadsheet modelling
“The user interface is starting to look dated” — G2 review, healthcare risk architect, March 2023 — the same complaint appears in at least six 2023 reviews
“Not available for MACs” — G2 review by an adjunct professor, May 2023 — echoed by a second professor in 2022
“Optimization function requires more expensive version” — G2 review, enterprise research user, May 2023
Limitations: value of information exists, but as a downloadable library you import and call as functions, not a command in the interface; tornado charts are likewise a documented workflow rather than a one-click report. Finding the optimal decision — rather than comparing alternatives you set up yourself — needs the Optimizer edition at $4,000. The influence-diagram paradigm is powerful and takes real learning — this is not a tool you pick up in an afternoon. And your model is not a spreadsheet, so the people who review spreadsheets cannot review it.
Documentation: user guide · tutorial · docs wiki · webinars
TreeAge Pro dominates one field completely: health economics. If you are building a cost-effectiveness model, a Markov model of disease progression, or anything destined for a health technology assessment, TreeAge is the tool your reviewers will expect and it does that job better than anything else here. Markov models, microsimulation, partitioned survival analysis and probabilistic sensitivity analysis are all first-class features with their own video tutorial series. It computes EVPI and EVPPI, which matters enormously in that world, where the question “is another trial worth funding?” is asked in earnest.
Image: a Markov model in TreeAge Pro, with per-state costs, utilities and event probabilities. TreeAge Software, treeage.com. Retrieved August 2026.
Note how it gets to value of information, though: TreeAge derives EVPI from a probabilistic sensitivity analysis run at a decision node, rather than analytically from the tree. That is the right approach for the models it is built for, and a heavier one for a straightforward business decision. Its threshold analysis is a model of clarity — it searches iteratively for the input value where the optimal strategy flips and then explains the flip in words.
Price: the Business edition is $660 per year or $2,000 perpetual; Healthcare is $760 per year or $2,350 perpetual with a $425 annual maintenance line (treeage.com store, retrieved August 2026). A free viewer that never expires lets anyone open your model — a good decision on the vendor's part. It is also one of only two paid tools here that run natively on a Mac (Big Sur or later, Intel and Apple silicon both supported).
What users say. The review-site record is thin — three reviews, averaging 3.5/5 — but TreeAge is unusual in having a peer-reviewed literature about it, because health economists publish comparisons of their own tools:
“Does not require a very steep learning curve, menu driven options and manuals make it easy” — Capterra review, hospital researcher, 2018 — who also reports it “gets hung up at times” and finds Excel integration difficult
“You cannot use it for publications” — G2 review, 2019, on the student licence restrictions
“The efficiency and transparency advantages of programming languages such as MATLAB and R become increasingly valuable when more complex analyses are required.” — Hollman et al., PharmacoEconomics 2017, comparing TreeAge, Excel, R and MATLAB for cost-effectiveness modelling
“A basic problem is the price of the TreeAge software, and the need to renew annually the license… which, in the medium to long term, results in an amount that may seem excessive.” — Bai, del Campo & Keller, Investigación Operacional 2016, reviewing low-cost Markov software options
Limitations: the trial is close to unusable for a real evaluation — 14 days, a maximum of 25 nodes, and you cannot save or print. Academic and non-profit licences are contractually barred from “commercially funded projects of any kind”, which catches more people than you would expect. Outside healthcare the product is a much less obvious choice. And its knowledge base has stale corners — the article telling you where to find tutorials points at a demos page that no longer exists.
Documentation: healthcare help portal · video tutorials · resources hub · knowledge base
Mike Middleton's three Excel add-ins are how a very large proportion of the world first met decision analysis: TreePlan builds and rolls back trees, SensIt does one-way and two-way sensitivity with tornado and spider charts, and SimVoi runs simulation and computes value of information. They are inexpensive, they run on Mac as well as Windows Excel — which in this comparison only TreeAge and the browser-based SilverDecisions can also say — and TreePlan was last updated in May 2025, so this is not abandonware.
Image: a TreePlan decision tree built on an Excel worksheet. TreePlan Software, treeplan.com. Retrieved August 2026.
Its distinguishing habit is worth understanding: TreePlan writes the rollback as plain Excel formulas on the sheet. Nothing is hidden in the add-in — which means the finished model recalculates for anyone who opens the workbook, no licence required, and an auditor can trace every cell. The price of that transparency is that the analyses stay simple: sensitivity comes from Excel data tables via SensIt, and SimVoi's value of information is computed by simulation on spreadsheet models — the vendor's own descriptions never use the terms EVPI or EVII.
Price: $59 each for TreePlan and SensIt, $79 for SimVoi, or $179 for the Decision Analysis Toolkit containing all three (treeplan.com, retrieved August 2026). Note the gap between the headline and the reality: trees and sensitivity and value of information means the $179 toolkit, not the $59 starting price.
What users say. There are no G2 or Capterra reviews to quote — none exist. The record here is a different kind of evidence: the vendor lists sixty textbook editions licensed to bundle the student version — including the standard management-science texts by Anderson & Sweeney, Taylor, Ragsdale and Balakrishnan — and business schools from Kellogg to Tuck distribute it in course materials. If you learned decision trees in an MBA elective, there is a fair chance this is the tool you learned them in.
Limitations: three separate products that do not share a model; no trial, only a 30-day money-back guarantee you claim by email; and student licences may not be used commercially. For teaching, and for anyone whose problem is genuinely small, this is remarkable value.
Documentation: PDF guides and example workbooks · TreePlan overview · support and version history
SilverDecisions is free, open source under LGPL v3, developed at the Warsaw School of Economics, and runs entirely in a browser with nothing to install. It draws trees, rolls them back with six selectable rules, does sensitivity analysis — including tornado diagrams, spider plots and a probabilistic mode that draws values from built-in distributions — plus multi-criteria comparison, exports to PNG, SVG, PDF and JSON, and shares a model by URL in either editable or read-only form. For teaching, for a quick structural sketch, or for anyone who simply needs to see whether a decision tree helps before spending money, it is the obvious first stop.
Image: our worked catalyst-retrofit example, rebuilt in the free SilverDecisions browser app in August 2026 — the same decision we analyse step by step in our value of information article.
It also has real academic credentials: the sensitivity methodology is published in a peer-reviewed paper (Kamiński, Jakubczyk & Szufel, Central European Journal of Operations Research, 2017), and the team published a free 250-page teaching book, Modeling Decisions with SilverDecisions, in 2022.
Limitations: no value of information — that feature request has sat open on the project's issue tracker since December 2016 — and no general Monte Carlo engine: the probabilistic option lives inside the sensitivity module rather than on the tree's payoffs. There are no G2 or Capterra reviews; the visible user base is academic, with published papers crediting it for their tree figures. It is academic software maintained on academic timescales, so support is what the community provides. But it costs nothing and it is genuinely good at what it does.
Documentation: manual (wiki) · free 2022 book (PDF) · issue tracker · launch the app
Because this market is so small, it is worth knowing exactly what independent evidence exists — and how little of it there is.
The INFORMS survey has not run since 2020. The professional society's OR/MS Today Decision Analysis Software Survey — the closest thing this category has to an industry census — last appeared in December 2020, covering 17 packages from 12 vendors, with all data self-reported by the vendors. Of the seven tools on this page, only DPL, SilverDecisions and the TreePlan family took part; Lumivero (then Palisade), Lumina and TreeAge did not appear. No edition has been published since.
The review sites are nearly empty. Add up every G2 and Capterra review for all seven products and you get roughly fifty, and the reviewer population skews heavily academic — professors, students and university researchers rather than industrial decision analysts. Compare that with mainstream analytics software, where a single product can carry thousands of reviews, and the practical conclusion follows: in this category, star ratings cannot carry the weight of a purchase decision. Documentation you can read before buying, and a trial you can actually evaluate in, are worth more than the ratings.
Where independent scrutiny does exist, it lives in journals. Health economists publish comparisons of their own tools: Hollman and colleagues' 2017 PharmacoEconomics comparison found Excel and TreeAge serve education and standard health-technology assessments well but saw “efficiency and transparency advantages” in R and MATLAB for complex analyses, and a 2016 review of Markov software concluded that free alternatives “are good alternatives to the relatively expensive TreeAge software”. SilverDecisions' methodology is itself peer-reviewed. This literature is narrow — it covers healthcare modelling, not business decision analysis — but it is the only truly independent testing this category gets.
We wrote this page to be the resource we wished existed when we did this research: every feature claim traced to vendor material, every price dated, every quote attributed. Where the table above says “not documented”, that is the finding.
GeNIe Modeler (BayesFusion) builds influence diagrams with decision, utility and multi-attribute utility nodes, and computes value of information that rank-orders candidate tests — genuinely strong, and rarely mentioned. The academic version is free, all business versions carry a 30-day trial, and the documentation portal is excellent — but commercial pricing is not published.
Netica (Norsys) finds optimal decisions from influence diagrams and is cheap for the category, with a free size-limited mode. Two cautions: it is Windows-native only (macOS via CrossOver, Linux via Wine), and the vendor's own two pages disagree on its price, so ask before you budget. Its tutorial is solid but the site is 1990s-era frames, and the product is absent from recent academic software surveys.
1000minds is web-based multi-criteria decision analysis using the PAPRIKA method, and is the serious option if your problem is ranking criteria rather than modelling uncertainty. It publishes its price openly, which is welcome, and that price is $25,000 a year for the full suite.
SmartOrg is a hosted platform aimed at R&D and innovation portfolios rather than individual decisions. It publishes no pricing; the route in is a demo request.
Finally, a warning about search results. Look for “decision tree software” and most of what you find will be diagramming tools — they draw a tree beautifully and cannot calculate anything. If a product does not talk about rollback, expected value or sensitivity analysis, it is a drawing package.
Every product on this page is honest about its features. The surprises are in the packaging, and they are worth checking before you shortlist rather than after.
One more thing that is easy to miss across the whole category: the public review record is almost empty. Between them these products have fewer than fifty reviews on the major software review sites, and a striking proportion of the reviewers are academics rather than industrial users. Treat star ratings here with more caution than you would in a larger market.
Four questions settle it faster than any feature list.
And then build the same small model in the two you shortlist, on a trial licence, ideally a decision whose answer you already know. An afternoon spent that way tells you more than any comparison page, including this one.
If you want to see what value of information looks like when it is worked through properly, we published a full example with every figure checkable by hand: what is information worth, EVPI and EVII.
Software that helps you choose between options when the outcome is uncertain. It models the choice as a decision tree or influence diagram, attaches probabilities and payoffs, works backwards to find the best expected value, and then tests how sensitive that recommendation is to its assumptions. It is a different discipline from the decision trees used in machine learning, which classify data rather than compare choices.
From nothing to a great deal. SilverDecisions is free; the TreePlan Decision Analysis Toolkit is $179 one-time; TreeAge Pro Business is $660 a year; Analytica Professional is $1,000 per user per year; ModelChoice is €950; and Lumivero's DecisionTools Suite is €3,680 to €4,790 a year. All retrieved August 2026.
Not according to Lumivero's shop, which lists it only as a component of the DecisionTools Suite. Decision trees therefore come bundled with @RISK, StatTools, NeuralTools and Evolver at €3,680 a year and up.
DPL reports value of information and control in every edition. TreeAge Pro computes EVPI and EVPPI from a probabilistic sensitivity analysis. Analytica offers it through an add-on library you import. SimVoi computes it on spreadsheet models. ModelChoice computes EVPI and EVII directly from the tree as a standard report. PrecisionTree's documentation describes no such feature, and SilverDecisions does not have one.
Three of the seven. TreeAge Pro runs natively on macOS Big Sur or later, on both Intel and Apple silicon. The TreePlan family supports Mac Excel 2019 through 365. SilverDecisions runs in any browser, so the operating system does not matter. ModelChoice, PrecisionTree, DPL and Analytica are Windows-only — a complaint that appears repeatedly in Analytica's own reviews.
DPL publishes its user guides as free PDFs and a complete help index on the web. Analytica maintains a public documentation wiki. TreeAge runs browsable web help portals plus eight video tutorial series. TreePlan's guides are free PDFs. SilverDecisions has a wiki manual and a free 250-page book. PrecisionTree is the outlier: its current help ships only as a downloadable Windows CHM file, a format Windows blocks by default. ModelChoice's manual is currently in-product, in seven languages. Direct links to every one of these are in the product sections above.
SilverDecisions. It runs in a browser, draws and rolls back trees, handles sensitivity and multi-criteria analysis, and shares models by link. It computes no value of information, and simulation appears only as probabilistic sensitivity analysis, but for structuring a decision and seeing whether the approach helps at all, it costs nothing and works well.
You can draw a tree in a spreadsheet and calculate its expected value by hand. What is hard by hand is everything afterwards: rebuilding the rollback when the structure changes, finding which single assumption is holding the recommendation up, working out how far it has to move before the answer flips, and pricing the study that would resolve it. That is the work these tools automate.
It depends entirely on the tool, and it is worth deciding early. Excel-based tools keep the model in an ordinary workbook — TreePlan goes furthest, writing the rollback as plain formulas that recalculate without the add-in. Analytica's free edition runs existing models of any size and TreeAge publishes a viewer that never expires. SilverDecisions shares by URL. For DPL we found no free reader, so a reviewer needs a licence.
Decision trees on the Excel worksheet, rolled back as you type, with sensitivity, robustness, value of information and a one-page Decision Brief from the same tree. Sixty example models included, PrecisionTree models import in one click, and the model stays a workbook anyone can open. Try the fully functional 15-day free trial.