What are the best questions to ask a planning software vendor at a trade show?

Lead with the data model and the implementation model, because both are structural and neither can be fixed by configuration later. Ask whether OTB, assortment, and buy planning share one data model or synchronize as separate modules, then ask who implements the system and who configures it after go-live. If those answers hold up, spend the remaining minutes on workflow fit — your hierarchy, your size curves, your reforecast cadence — and book a full demo for everything else.

How can I tell whether a platform is a single data model or separate modules?

Ask what the OTB view shows, and when, at the moment a buyer revises the assortment. On a single data model the answer is a mechanism: both views read the same records, so the change is simply there. With modules bolted together the answer involves a sync — an integration layer, a batch schedule, an export. Any mention of overnight jobs or an integration that handles it means you will reconcile plan versions yourself, which is the spreadsheet problem restated at platform prices.

What should I ask about implementation in a first vendor conversation?

Three things. Who runs the implementation — the vendor's team, your team, or a third-party integrator. Who configures the system after go-live when your hierarchy, calendar, or channel structure changes. And what a typical change request costs in elapsed time. The staffing answer predicts the shape of the whole project; the after-go-live answer predicts your cost of change for the life of the contract.

How do I evaluate a vendor's AI claims in a short conversation?

Ask what the AI concretely does: the inputs it reads, the outputs it produces, and where those outputs land in the planning workflow. Then ask where the human decides — whether the system proposes and a planner approves, or acts on its own. Vendors with real capability answer with mechanisms and named workflow steps; vendors without it answer with adjectives. A useful follow-up: ask what the AI gets wrong and how a planner sees and overrides it.

What is a red flag in a vendor's answer about references and proof?

A vendor who cannot say which of the capabilities just demonstrated are running in production with paying customers, who deflects on renewal behavior, or who positions the product as the right fit for every retail segment. Strong vendors name limits — segments they serve less well, capabilities that are roadmap rather than shipped product — because software presented as right for everyone has been shaped for no one in particular.

Questions to Ask Planning Software Vendors (at a Conference or Anywhere)

The questions that separate demo-ware from daily-use planning software — data model, implementation model, workflow fit, AI, and proof — with what a good answer sounds like and what's a red flag.

The fifteen-minute filter

Vendor qualification questions are the short list a planning leader uses to decide, inside a single conversation, whether a software vendor belongs in a formal evaluation — the filter that separates demo-ware from software a merchandising team will run a season on. On a conference floor that conversation is fifteen minutes standing up; even a scheduled demo is an hour against a rehearsed script and pre-loaded data. The questions below are built for that constraint. Each is answerable in a couple of minutes, each has a recognizable good answer, and each has a recognizable evasion.

They fall into five families, ordered by how hard the underlying property is to fix later. Architecture is permanent, implementation model is contractual, workflow fit is configurable at a cost, AI claims are testable, and proof is either there or it is not. If a vendor fails the first two families, the rest of the conversation is a courtesy.

This guide owns the question list. The scoring framework around a full evaluation — criteria weighting, the RFP stage, contract red flags — lives in the software selection guide, and the deeper selection framework for apparel specifically is in How to Choose Planning Software.

Data model questions

Architecture is the property no amount of configuration, services, or goodwill can change after signature, which is why it goes first even when time is short.

"Are OTB, assortment, and buy planning one data model, or modules that synchronize?" A good answer names the mechanism: one set of records, read by different views, so a change made in one workflow is visible in the others because there is nothing to transfer. A red flag is the word "integration" applied to the vendor's own product — modules acquired or bolted together and reconciled by a sync layer behave like separate products in daily use, whatever the pitch says.

"A buyer revises the assortment mid-review. What does the OTB view show, and when?" The good answer is boring: the revised receipts are there when the planner looks, because both views read the same plan. The red flag is any answer containing a schedule — overnight jobs, a refresh, a publish step. A sync schedule means that for some part of every day your OTB and your assortment disagree, and someone on your team owns knowing which one is right.

"When two versions of the plan disagree, how does the system reconcile them?" You are listening for versioning as a first-class concept: working versions against an approved plan, with a visible diff and a controlled promotion step. The red flag is reconciliation described as a human process — exports compared side by side, a planner who "checks them against each other." That is the spreadsheet workflow you are trying to retire, rehosted.

Implementation-model questions

The implementation model determines who you depend on, for how long, and what changing your mind costs after go-live. It deserves more scrutiny than most evaluations give it — a demo shows the destination, not the journey. These three questions qualify a vendor on the floor; How to Evaluate the Implementation Model is the deep dive to run before contract.

"Who implements — your team, my team, or a third-party integrator?" A good answer is specific about staffing on both sides and does not involve a third logo. The moment a systems integrator enters the answer, the project has three parties with three sets of incentives, and accountability for the outcome has a gap in it. Ask directly whether the standard implementation involves an integrator, and whether the vendor's own services team or your planners do the configuration.

"After go-live, who configures changes — my planners or your services team?" This is the question that predicts your cost of change for the life of the contract. A good answer is that planning teams change hierarchies, calendars, and workflows themselves in the UI. A red flag is that changes route through the vendor's services organization or a partner — if so, ask what the current queue looks like, in their words, and listen for how comfortable the answer is.

"What does a typical change request cost in elapsed time?" Ask for a recent, concrete example: the last time a customer restructured a hierarchy or added a channel, what happened and how long did it take. A vendor with a self-serve model tells a short story. A vendor with a services-dependent model tells you about scoping.

RetailNorthstar's answers to this family are published design facts: standard onboarding — importing historical data, mapping the planning structure, configuring the workflow — is included in the subscription, the standard implementation involves no third-party integrator, planning teams configure the platform directly in the UI, and most brands are live in weeks. The staged-adoption model is described on How It Works and the commercial terms on Pricing.

Workflow-fit questions

Generic demo data hides the failure mode that matters: the tool that handles a clean four-level hierarchy beautifully and chokes on yours. Every question in this family should be about your business, described precisely.

"Can it hold our hierarchy as-is — division, category, class, style, SKU — plus the attribute layer we plan on?" State your real structure, including the ugly parts: the class that behaves like a division, the collection dimension that cuts across categories. A good answer engages with the specifics and describes how the hierarchy is configured, by whom. A red flag is any suggestion that you adapt your hierarchy to the tool's — that is a re-implementation of your planning process disguised as onboarding.

"Walk me through size curves — at what level do they live, and what happens when the curve differs by channel?" For an apparel brand this is a fast, reliable tell. Size curves that exist as nothing more than a spreading ratio at buy time are a checkbox, not a capability. You want curves as planning objects: maintained at an appropriate level of the hierarchy, overridable by channel, and visible in the buy so a broken size run in the actuals can inform the next curve.

"What does in-season reforecasting look like on a Monday morning?" Describe the real Monday: actuals landed over the weekend, two categories are running hot, one is missing plan badly. A good answer walks the workflow — where the variance surfaces, how the reforecast is made, what happens to open orders and the remaining OTB when it is. A red flag is a reforecast that is effectively a new plan file, because that means in-season management will still live wherever it lives today.

AI questions

Every planning vendor now claims AI. The claims are not equally grounded, and two questions separate them faster than any analyst report.

"What does 'AI-assisted' concretely do in this product?" Demand mechanism, not category: which inputs the model reads, which outputs it produces, and precisely where in the workflow those outputs appear. "It uses AI to optimize your assortment" is an adjective wearing a verb's clothing. "It proposes a size curve from the last four seasons' sell-through at style-color level, and the planner sees it next to the current curve" is a product. Vendors with real capability enjoy this question; vendors without it change the subject to their roadmap.

"Where does the human decide?" The answer you want is a clear line: the system proposes, flags, and drafts; the planner approves, overrides, and owns the number. Be equally wary of both failure modes — autonomy theater, where "AI" turns out to mean a static report, and the opposite, a system that acts on your buy without a visible approval step. A useful follow-up is to ask what the model gets wrong and how a planner sees it — a vendor who cannot name a failure mode has not run the model in production, or is not being straight about it.

Proof questions

The last family is the shortest, and the answers are the hardest to fake.

"Which of the capabilities you just showed me are running in production with paying customers today?" Demos routinely include roadmap. That is legitimate — but you are entitled to know which screens are product and which are prototype, and a good vendor draws the line without being cornered into it.

"What does your renewal behavior look like?" You are not asking for a statistic to audit; you are watching how the vendor handles the question. Comfort with specifics — how long the longest-tenured customers have been live, what tends to drive the departures they do have — is the signal. Deflection to logo counts is the anti-signal.

"Who is this product wrong for?" The most information-dense question on this list. An honest vendor names limits: segments they serve less well, complexity they are not built for, capabilities that are roadmap rather than shipped. A vendor whose product fits every retail segment at every scale is describing no one in particular, and the fit for your business specifically is being asserted, not established. We hold our own comparison pages to this standard — a comparison that names no tradeoffs is an advertisement.

Where you'll use these in 2026–27

The question list works anywhere, but the planning-software conversation clusters around a handful of events.

RetailClub AI Festival — Sep 22–24, 2026, Huntington Beach, CA (retailclub.com). First edition, from the team that founded Shoptalk and Groceryshop, organized around Build, Lead, Sell, and Live with AI. The AI question family will get its heaviest workout here — the event states 150+ speakers, around 80% C-level.
Shoptalk Fall — Sep 29–Oct 1, 2026, Nashville (fall.shoptalk.com). Moved from Chicago; Shoptalk states 3,500+ leaders and 350+ sponsors and exhibitors — small enough that a second conversation with the same vendor is realistic.
RE:PLAN Europe — Oct 13–14, 2026, The Hurlingham Club, London (pi.tv/events/209). Formerly PI Apparel Merchandise Planning; PI bills it as the only event dedicated to merchandise planning in fashion, with 2026 themes including Tech Synergy and Resilient Planning. The workflow-fit questions belong here. PI lists a US edition, RE:PLAN USA, in New York for spring 2027.
NRF Big Show — Jan 10–12, 2027, Javits Center, New York (nrfbigshow.nrf.com). NRF states 41,000+ retail professionals from 100+ countries. At this scale the fifteen-minute filter is the whole game: qualify on the floor, evaluate afterward.
Shoptalk Spring — Mar 22–24, 2027, Las Vegas (spring.shoptalk.com). Shoptalk states 10,000+ attendees and 180+ speakers, with about one in three attendees C-suite.

Whichever floor you are standing on, the discipline is the same: ask the structural questions first, make the workflow questions about your business, and write the answers down — the comparison you run afterward is only as good as the notes you took in the moment.

Related resources

How to Choose Planning Software — the full selection framework these questions feed into
Software Selection Guide — scoring matrix, RFP process, and contract red flags
Build vs. Buy — the question before the vendor questions
Data Readiness for Merchandise Planning — what to have in order before any implementation starts

Ask RetailNorthstar these questions in a live demo — including the ones about limits.

Book a Demo →