Review of Perfect Planner, Planning Software Vendor

By Léon Levinas-Ménard

Last updated: November 4, 2026

Perfect Planner provides material-planning software for manufacturers using MRP. Its Intelliplanning Logic Engine analyzes replenishment information and produces daily, prioritized tasks for planner-buyers. The application brings together part and supplier information, projected shortages, inventory imbalances, and management scorecards. 1 3

Overall supply chain score: 3.4/10

Dimension	Score
Supply chain depth	3.8/10
Decision and optimization substance	3.0/10
Product and architecture integrity	4.0/10
Technical transparency	2.8/10
Vendor seriousness	3.4/10

Perfect Planner automates part of the analysis surrounding an existing MRP process, then directs people toward the resulting work. The public evidence does not establish a decision engine that jointly values uncertain demand, supplier deliveries, production constraints, and competing uses of capital. The marketed accuracy figures concern analytical outputs with an unspecified scoring method; they do not validate future shortage predictions or economic decisions. 2 7

The economic claims remain unsupported by a sustained evaluation of implemented decisions. The product also limits workflow customization while supplying each customer with a tuned proprietary engine, making the vendor’s continuing ability to revise that engine an important part of ownership cost. 5 7

Perfect Planner versus Lokad

Perfect Planner organizes the work generated by MRP: which part needs attention, what changed, and what the planner should investigate. Its detailed screens expose balance projections, notes, part status, and safety-stock adjustments. 3

Lokad’s managed offering combines decision software with Supply Chain Scientists who develop and maintain the customer’s executable models. Their work includes interpreting data, representing uncertainty, valuing choices, and revising the logic when operational results reveal mistakes. The intended daily output is a set of economic decisions, with human effort directed toward improving the system that produces them. Lokad’s service model

Consider two purchased components competing for a limited expedite budget. The component with the earliest projected shortage is not necessarily the one worth expediting. One may interrupt a low-margin order with an acceptable substitute; the other may prevent a costly production stoppage. Supplier reliability, the chance of recovery without intervention, and the value of the affected production all matter.

A task ranking can help a planner examine that choice. To automate it, the software must compare the consequences of the available actions under uncertainty. Perfect Planner’s published material does not disclose how its prioritization combines those probabilities and monetary effects. Lokad’s inventory optimization approach explicitly centers that calculation and the continuing work needed to keep it aligned with operations.

The practical comparison therefore includes both operating labor and maintenance labor. Perfect Planner’s task standardization may reduce manual analysis; Lokad’s Scientists also consume effort while developing and maintaining decision logic. Each approach needs to show that its total cost is justified by the decisions delivered. Introduction to Supply Chain supplies the broader economic perspective behind this assessment.

The decision process: a ranked workload above MRP

The main product page specifies daily analysis of existing MRP data and a six-month forward view. It identifies drivers such as lead-time variance, forecast changes, minimum order quantities, and parameter misalignment. 1 The detailed interface supports reviewing demand and supply balances, investigating causes, adjusting safety stocks, and exporting individual or historical analysis tables. 3

These are concrete operational functions. They can spare a planner repeated joins, calculations, and searches through separate reports. The unresolved question is how the software selects an economically preferable action when several plausible responses compete.

A projected shortage can be addressed by expediting, substituting, transferring, delaying an order, revising production, or accepting the shortage. Those actions have different costs and uncertain consequences. A ranking based on urgency or a weighted impact score does not establish that the most valuable combination has been selected.

The product’s public technical description emphasizes function-specific logic coded against system data. It also claims to correct MRP errors, suppress false positives, and adapt to changing demand. It does not identify the prediction model, the objective function, or a reproducible decision-selection method. 2

Uncertainty needs to enter the decision

Mentioning lead-time variance is not enough to demonstrate probabilistic planning. The calculation needs a distribution of future receipt dates, quantities, and demand, with dependencies where they materially affect the choice. For example, two suppliers using the same port may fail together. Treating their deliveries as independent can make a production plan look safer than it is.

The disclosed product pages do not explain how such futures receive probabilities or how these probabilities affect replenishment and allocation. A six-month balance projection can extend a deterministic plan without quantifying its reliability. The horizon provides visibility into assumptions; it does not establish forecasting accuracy.

The same issue affects safety-stock adjustments. Lowering a buffer can release capital while increasing the expected cost of disruption. A recommendation needs to value both effects and their interaction with production capacity, shared parts, and alternative sourcing. The public evidence does not show that joint calculation.

The accuracy and productivity benchmark

The company’s October 21, 2024 presentation to IISE describes a vendor time study involving 50 part numbers and a planner with more than seven years of experience. The task included shortage analysis, variance calculations, inventory metrics, and locating savings. Its comparison is reproduced below. 7

Measure	Senior planner	Perfect Planner
Analysis time	7.7 hours	Under 1 second
Forward visibility	4 weeks	26 weeks
Shortages identified	11	21
Savings located	$89,000	$154,000
Overall accuracy	88.4%	99.9%

The same presentation compares a $296.15 planner cost, using an $80,000 annual salary, with $1.67 for Perfect Planner. It does not specify the scoring denominator, reference answers, input-preparation work, recurring execution effort, or the derivation of the software cost. 7

What near-perfect accuracy establishes here

The current feature page advertises an insight-accuracy rate above 99% and an engine behaving like an experienced planning specialist. 2 The older experiment does not establish 99.9% demand-forecast accuracy: it evaluates a collection of analytical tasks.

That distinction exposes the real problem with the claim. Correctly computing balances from supplied numbers can approach perfect accuracy while the resulting inventory decisions remain poor. An arithmetic result may be correct under the input assumptions even when the next shipment arrives late, the demand forecast fails, or a proposed saving creates a more expensive shortage.

The reported percentage has no interpretable decision value without a definition of an error, the weights assigned to different errors, and a frozen test on unseen operational outcomes. Counting many easy checks can conceal a small number of costly mistakes. Missing one production-critical shortage can matter more than correctly classifying hundreds of routine records.

The benchmark also changes the amount of analysis completed. Its visibility horizons differ. The result therefore combines faster calculation, broader coverage, and whatever scoring procedure was used; it does not isolate the engine’s predictive performance against an equivalent baseline.

The baseline is manual work

A modern comparison must include an automated implementation of the same analytical task. Batch calculations, joins, and rule checks can be inexpensive without a special planning engine. The experiment does not isolate the proprietary method’s contribution from the general benefit of automating manual analysis.

More importantly, processing time is only one part of operating productivity. Buyers still need to validate recommendations, contact suppliers, negotiate changes, execute orders, handle exceptions, and correct bad assumptions. The time study does not measure that complete workflow over a sustained portfolio.

The distinction between savings located and realized returns is equally consequential. A proposed inventory reduction may be infeasible because of purchasing commitments, production risk, or supplier terms. Its value also depends on what happens after implementation. The benchmark supplies candidates for investigation, not a measured profit improvement.

Economic claims become stronger than their evidence

The dashboards page advertises improvements up to 90% fewer expedites, 300% higher planner productivity, and 40% lower inventory. It also exposes materials and supplier-impact scores, inventory value, and task-completion monitoring. 4

In an April 6, 2026 first-person company profile published by GACC South, Perfect Planner describes the same three improvements as typical. The profile also claims planning accuracy above 99.9%, without publishing customer baselines, an evaluation cohort, or the measurement procedure. 8

“Up to” and “typical” make materially different promises. A maximum can describe one exceptional case; a typical outcome needs a distribution across relevant deployments. Neither source supplies that distribution. The public evidence does not justify converting an upper-bound sales claim into an expectation for an ordinary customer.

The homepage’s named testimonials describe easier analysis and reliable information, but provide no quantified customer return. 1 That leaves the large percentages disconnected from a named implementation with a defined starting point, operating period, and full cost.

Task-completion dashboards create another possible mismatch. Completing more alerts is valuable only if the alerts correspond to economically useful actions. A system can improve the measured throughput of an unnecessarily labor-intensive process. The appropriate target is the cost of obtaining and executing good decisions, including the work that should no longer be generated.

Technology history and roadmap integrity

Perfect Planner’s public leadership material identifies Thomas Beil as founder and chief executive; its August 2024 announcement established the Alpharetta headquarters. 10 11 The relevant technical chronology concerns the changing explanation of the same planning proposition.

Date	Claim and follow-through
August 2023	The founder’s forecasting guide recommends conventional time-series and regression methods, holdout validation, and later adding AI. It presents forecasting advice rather than a released product specification. 13
October 2024	The IISE presentation describes more than 2,000 calculations per SKU per day and publishes the analytical comparison examined above. 7
September 2025	The company’s real-time planning article promotes immediate signals and replenishment based on actual demand. It criticizes slow planning cycles without documenting a product-specific refresh or execution benchmark. 12
April 2026	The GACC profile presents broad inventory, expedite, and productivity reductions as typical outcomes. 8
Current product pages	The homepage counts more than 2,500 data points per SKU, while the accuracy page counts more than 2,500 proprietary algorithms. The published architecture still centers daily analysis and planner work lists. 1 2

Data points, calculations, and algorithms are different units. Increasing or relabeling their count does not demonstrate an improved model. The useful evidence would be a dated explanation of the changed computation and its effect on decisions, operating effort, and economic outcomes.

The real-time argument also needs to be reconciled with daily product processing. Daily execution may be adequate for many procurement decisions; claiming that slower information is inherently inadequate does not establish the value of a faster cycle. What matters is whether new information arrives before a decision can still be changed profitably.

Similarly, observed demand cannot replace a forecast when replenishment takes time. A purchase made today must anticipate consumption before the next receipt. Faster observation narrows some uncertainty but cannot eliminate the future.

The history shows a consistent focus on planner standardization, accompanied by expanding accuracy and benefit claims. The missing follow-through is a public evaluation that connects the claimed improvements to deployed decisions and sustained economics.

Software bill of materials

Component or boundary	Publicly attributable evidence
Microsoft Azure	Hosting platform identified in the FAQ. Specific compute and storage services are not named. 5
Intelliplanning Logic Engine	Proprietary, function-specific planning logic; public materials do not identify a solver, forecasting library, runtime, or database. 2
MRP inputs and analytical exports	The product consumes existing planning information and exports analysis tables. Public material does not specify connector protocols or transactional writeback behavior. 1 3
Engineering ownership	Patrick Wensel’s company biography assigns him architecture, implementation, testing, maintenance, version control, and continuous integration/deployment responsibilities. 9

The published inventory is too thin to assess the system’s operational mass. It leaves unresolved the storage copies, transformation stages, recovery design, and infrastructure needed for the advertised daily calculations. A proprietary algorithm count says nothing about those costs.

The execution path also leaves integration responsibilities unclear. The public interface demonstrates analysis exports, but does not explain how a recommended change becomes an acknowledged transaction in the authoritative ERP, how retries avoid duplicate actions, or how a failed update is reconciled. Those details determine how much work remains with the customer.

The engineering biography identifies responsibility for maintaining the software, but gives no product-specific dependency inventory or inspectable release process. Future ownership costs depend on how efficiently that team can adapt customer logic and integrations without accumulating fragile exceptions.

Configuration maintenance and security

A fixed workflow with customer-specific logic

The FAQ says each organization receives a tuned Intelliplanning engine, but workflow customization is unavailable. Interface customization can incur additional charges. Subscriptions use 12-month terms and monthly per-SKU pricing. The vendor also offers no-commitment processing of a prospective customer’s data before onboarding. 5

That preflight can expose integration and interpretation problems before a contract begins. It does not replace a sustained evaluation of decisions under changing conditions.

The combination of fixed workflow and tailored logic concentrates responsibility with the vendor. If an economically important action does not fit the prescribed process, the customer needs an engine change, a separate process, or a workaround. A lower exception count is not enough if the remaining exceptions become more expensive to resolve.

For holimization, the critical question is how a harmful recommendation leads to a revised objective, constraint, data interpretation, or action. The public materials do not expose customer-controlled versioning, replay tests, approval of numerical changes, or rollback. Adjusting a safety-stock setting is a much narrower intervention.

Total cost consequently includes subscription, data preparation, remaining buyer work, vendor tuning, regression checks, and workarounds around the fixed workflow. A per-SKU charge makes portfolio size a visible billing driver, but does not disclose the service effort required to maintain decision quality.

Data retention and damage containment

The FAQ says raw organizational data is disposed of after 48 hours while derived analytics remain indefinitely. The privacy policy more broadly says collected data remains until a customer requests disposal, subject to its stated exceptions. It does not reconcile this wording with the raw-versus-derived distinction. 5 6

The distinction matters for both exposure and reproducibility. Derived analytical records can still reveal purchasing, inventory, and supplier information. Reconstructing a recommendation also requires the relevant historical inputs and logic version. If those inputs are no longer held in the service, a reproducible investigation needs an explicitly maintained record elsewhere.

The security policy refers readers to an annual SOC 2 report and promises notification within 72 hours after a breach is positively confirmed. It does not publicly explain tenant isolation, administrative privileges, credential handling, tamper-resistant history, or tested restoration. 6

Azure hosting and an audit reference do not establish those safeguards. The security assessment remains weak because the customer cannot inspect how the application limits damage when an account, dependency, or administrator is compromised.

Supply chain score

Supply chain depth

Criterion	Score
Economic framing	4/10
Decision end-state	5/10
Conceptual sharpness on supply chain	5/10
Freedom from obsolete doctrinal centerpieces	3/10
Robustness against KPI theater	2/10

Average: 3.8/10

Economic framing — 4/10. The application addresses shortages, excess stock, and expediting, but does not disclose a common monetary objective for competing responses. Prioritizing costly-looking problems can still miss the most valuable combination of actions. 1 2

Decision end-state — 5/10. Ranked tasks and recommendations influence operational decisions. The documented end-state still requires a planner to turn the analysis into an executed response, and the public integration description does not establish reliable transactional closure. 3

Conceptual sharpness on supply chain — 5/10. The product targets a recognizable problem in manufacturing replenishment. Its explanation becomes weaker where shared capacity, uncertain supply, and the value of competing orders must enter one calculation. 1

Freedom from obsolete doctrinal centerpieces — 3/10. The operating model remains centered on MRP signals, safety-stock intervention, and individual work queues. Organizing those activities more consistently does not establish that the underlying decision process has been economically redesigned. 3

Robustness against KPI theater — 2/10. Undefined analytical accuracy, potential savings, and broad outcome percentages are used to support stronger claims than their published evaluations justify. Task throughput and record-level correctness do not measure the value of resulting decisions. 2 7 8

Decision and optimization substance

Criterion	Score
Probabilistic modeling depth	1/10
Distinctive optimization or ML substance	3/10
Real-world constraint handling	4/10
Decision automation and labor productivity	4/10
Resilience under real operational complexity	3/10

Average: 3.0/10

Probabilistic modeling depth — 1/10. No disclosed method connects calibrated demand and supply distributions to the selected actions. Without that connection, the published risk language cannot establish economically sound decisions under uncertainty. 2

Distinctive optimization or ML substance — 3/10. The engine may encode useful domain rules, but counts of proprietary logic and a comparison with manual work do not isolate a distinctive numerical method. Its decision quality remains difficult to reproduce or benchmark. 2 7

Real-world constraint handling — 4/10. Part balances and parameter-related exceptions expose practical operational issues. The public description does not establish joint treatment of procurement, production, and scarce resources when resolving them. 1 3

Decision automation and labor productivity — 4/10. Automated analysis can reduce clerical work substantially. The measured comparison does not account for the complete recurring workflow or compare against a contemporary automated baseline at equivalent decision quality. 7

Resilience under real operational complexity — 3/10. Root-cause views help diagnose recurrent problems, but customer access to reformulation and controlled testing is not documented. The fixed workflow can turn a newly discovered economic requirement into a vendor change request or an external workaround. 3 5

Product and architecture integrity

Criterion	Score
Architectural coherence	6/10
System-boundary clarity	6/10
Security seriousness	2/10
Software parsimony and ownership cost	4/10
Compatibility with programmatic and agent-assisted operations	2/10

Average: 4.0/10

Architectural coherence — 6/10. The visible functions fit a focused replenishment workflow rather than a collection of unrelated applications. The undisclosed data and execution architecture prevents a stronger assessment of their technical integration. 1 3

System-boundary clarity — 6/10. Dependence on the existing MRP process is explicit. The boundary becomes less clear at execution: public sources do not explain which actions are written back automatically and how inconsistent states are resolved. 1

Security seriousness — 2/10. The public explanation relies on hosting and audit assurances rather than inspectable containment and recovery mechanisms. The inconsistent retention descriptions also leave the customer’s information exposure difficult to determine. 5 6

Software parsimony and ownership cost — 4/10. A narrow product scope can limit implementation work, but the dependency inventory and recurring tuning effort are undisclosed. Remaining operational labor and vendor-mediated changes must be included alongside the subscription. 5 9

Compatibility with programmatic and agent-assisted operations — 2/10. Analytical exports provide a limited integration surface. Public materials do not establish an API for reliable batch execution, inspectable decision logic, or controlled programmatic improvement of the engine. 3

Technical transparency

Criterion	Score
Public technical documentation	2/10
Inspectability without vendor mediation	3/10
Portability and lock-in visibility	4/10
Implementation-method transparency	3/10
Evidence density behind technical claims	2/10

Average: 2.8/10

Public technical documentation — 2/10. Feature descriptions explain what users see, but do not provide the data schema, numerical methods, execution semantics, or architecture needed for technical assessment. The core claims remain specifications of intended benefits. 2 3

Inspectability without vendor mediation — 3/10. Screens and a dated experiment expose some behavior before a sales discussion. Inspecting a recommendation’s full derivation, testing alternatives, and reproducing its validation still require access the public record does not provide. 3 7

Portability and lock-in visibility — 4/10. Exportable analysis helps preserve outputs. It does not provide the customer-specific rules and execution behavior needed to reproduce those outputs independently, leaving a substantial dependency on the proprietary engine. 3 5

Implementation-method transparency — 3/10. The vendor identifies a commercial model, preflight, and customization boundary. It does not disclose how customer-specific logic is specified, tested, upgraded, and accepted over the contract’s life. 5

Evidence density behind technical claims — 2/10. The principal quantitative example is a vendor-authored analytical comparison with an undefined accuracy metric. Public claims about improved business outcomes lack the accompanying evaluation needed to validate their size or typicality. 7 8

Vendor seriousness

Criterion	Score
Technical seriousness of public communication	3/10
Technology roadmap integrity	4/10
Conceptual sharpness	4/10
Economic accountability	3/10
Defensibility in an agentic-software world	3/10

Average: 3.4/10

Technical seriousness of public communication — 3/10. Algorithm counts and near-perfect percentages carry more promotional weight than their disclosed definitions support. The numerical language obscures whether the claimed advance concerns arithmetic, classification, prediction, or decision quality. 2 7

Technology roadmap integrity — 4/10. The product retains a consistent replenishment-workflow focus. Successive publications do not establish how the engine evolved or connect broader benefit claims to measured deployment outcomes. The real-time positioning also lacks a corresponding product-specific evaluation. 7 8 12

Conceptual sharpness — 4/10. Standardizing repetitive analysis addresses a concrete productivity problem. Treating the resulting task list as the center of planning leaves the harder question of which decisions should be computed and executed without routine human review. 1 3

Economic accountability — 3/10. The preflight creates an opportunity to test applicability before purchase. Published return claims still lack bounded customer obligations, full costs, a defined failure condition, and responsibility for remediation when recommended actions disappoint. 5 8

Defensibility in an agentic-software world — 3/10. Domain rules and accumulated operational knowledge may support useful software. A defensible advantage requires evidence that they outperform broadly available automation while reducing total operating work; a faster queue of human tasks is an incomplete answer. 2 7