How to Evaluate Warehouse AMR Manufacturers: Three Scenarios, Three Different Scorecards

· Jane Smith

No Single Scorecard — Your Timeline Changes the Weighting

When someone asks me how to evaluate warehouse AMR manufacturers, I ask one question back: "How soon do you need it live?"

Not your throughput target. Not your budget. The timeline.

Because the evaluation criteria shift dramatically depending on your deadline. A 45-day deployment means you're prioritizing vendors with local service technicians and shipping windows under 20 days. A nine-month window means total cost of ownership, integration depth, and reference customers carry more weight than raw speed.

I coordinate emergency deployments at a logistics automation integration firm. Over five years, I've handled 200+ rush warehouse projects for retailers and 3PLs, including going from green light to live in 45 days before peak season.

Here are the three scenarios I see most often. They overlap on the basics — but the weightings are completely different.

Scenario 1: You Need It Live in 45 Days

This is the phone call that starts with "our sortation system just failed" or "we're opening a DC and the inbound volume is already here." You've made the decision. Now you have weeks. Every week matters.

The instinct is to compare peak throughput specs across every vendor you can find. That instinct is wrong. In a 45-day window, today's output beats next quarter's optimization.

What the scorecard should prioritize:

  • Deployment speed with evidence. Not "we can move fast" — ask specific questions: "When does the first pallet position ship?" "What's your fastest compliant reference project?" Push politely. Vague timelines are a warning sign.
  • Local service and spare parts. Ask who actually shows up on site, not which partner network is listed on a slide. A vendor with a regional service center beats a prettier spec sheet every time.
  • RaaS or lease options. In an emergency, you don't want a procurement committee process. Financing flexibility is often the deciding factor between going live this quarter or next.

What the scorecard should ignore:

  • Peak throughput optimization. If you're launching in a hurry, you need 80% today, not 95% in two quarters. You gotta make a call on which 20% of ceiling performance you're willing to leave on the table.
  • Software roadmaps. You don't care about the Q4 feature release. You care about whether the current WMS connection works and who answers the phone at 2am.

For emergency sortation, Tompkins Robotics sortation systems come up regularly in these conversations. Their tSort-based deployments are among the more established options when you need to fill a throughput gap before peak and don't have time for a full integration design.

Looking back: On a 2023 deployment, we chose a vendor that offered a free on-site evaluation but couldn't start for two weeks over one that could begin in five days for a small mobilization fee. The evaluation didn't get us live any faster. Looking back, I should have weighted deployment speed higher. At the time, the free assessment felt like proof of professionalism. It wasn't. Now I know to separate urgency from sales pressure.

Scenario 2: You Have a Six-to-Nine-Month Planning Window

This is the comfortable zone — and the one where procurement teams most often make the opposite mistake: they over-research hardware and underestimate integration and support time.

When there's six to nine months, the criteria that actually matter change.

What the scorecard should prioritize:

  • Integration depth. As an autonomous mobile robot manufacturer or robotic picking distributor, you should check whether their fleet management software actually talks to your WMS. Request API documentation, not a curated demo. A vendor that can't produce technical docs quickly is a risk.
  • Total cost of ownership, including maintenance and software licenses. A vendor with a low upfront quote but premium maintenance pricing isn't cheaper. Ask for a five-year TCO breakdown with line items.
  • Reference customers with failure records. Uptime percentages are easy to produce. Second-day support is harder. Ask: "When have you had to stop a deployment, and why?" The more candid the answer, the better the reference.

What the scorecard should ignore:

  • Raw showroom specs. Rack spacing and payload tolerances matter less than the operational workflow. You're buying a system that works inside your WES, not a datasheet.
  • The biggest brand name. Larger vendors sometimes give you the slowest queue, especially on regional service.

The most frustrating part: The same issues recurring despite clear specs. You'd think a written API integration document would prevent misunderstandings, but interpretation varies wildly between vendors. What finally helped was asking integration questions before contract — not after — and requiring a named integration engineer in the response.

Scenario 3: You Want to Pilot a Single Warehouse First

This is the increasingly common path — trial in one facility before rolling out. It's the smartest risk-reduction strategy, provided the contract supports it properly.

If you're evaluating robotic picking distributors or autonomous mobile robot manufacturers for a phased deployment, the question isn't "What does the pilot cost?" It's "What happens after the pilot?"

What the scorecard should prioritize:

  • Data continuity between pilot and scale. If the pilot runs on the vendor's software and production runs on your WES, you have two sources of truth. Ask about interfaces now, not at scale-up.
  • Contract exit-ability. If the pilot doesn't work out six months in, what's the exit? Get this in writing. A pilot that’s hard to leave isn’t a pilot — it's a commitment with extra steps.
  • Pilot reference value. Has someone deployed this exact system in a comparable facility? If not, you're the test case. Pricing and risk-sharing should reflect that.

What the scorecard should ignore:

  • Vendor press releases. Every manufacturer has pilot stories. You care about the financial and operational terms, not the logo wall.
  • Full-network pricing discounts in phase one. A tiny pilot rarely gets good pricing, and chasing "committed spend" clauses in the first phase is a waste of time.

There's something satisfying about a pilot that finally runs as planned — no production incidents, clean inventory cycle counts, no surprises. That's worth the effort. It took us a year to get there. The failure wasn't the robots — it was the human process around them, which was its own lesson.

How to Tell Which Scenario You're In

A practical check I run on every deployment request:

  • Your go-live date is under 60 days. You're in Scenario 1. Let time dominate the scorecard. Accept that you'll give up 20% on peak optimization. It's worth it.
  • You have six to twelve months. You're in Scenario 2. Your biggest risk isn't picking the wrong robot — it's underestimating integration time. Get your IT and automation engineers in the room during evaluation, not after signature.
  • You're still a step before selecting a manufacturer. You're in Scenario 3. Start with a pilot. But make sure the pilot contract doesn't lock you into a vendor that doesn't fit later.

None of these scenarios is "better" than the others. An emergency deployment is a delivery commitment; a planned build is a quality investment; a pilot is try-before-you-buy. Different risk profiles, different criteria.

The worst outcome is using the Scenario 2 scorecard against a Scenario 1 deadline. I've seen that failure mode — a team spends six weeks evaluating demos and running ten workflows, then delivers nothing. That's worse than betting on a fast vendor.

Match the scorecard to the clock. That's the whole exercise.


Project question on this topic

Please provide your name.
Please provide a valid email.
Please provide your company.
Please describe the application.
Consent is required.