William Blair: Manhattan Associates Posts Back-to-Back Record Booking Quarters Amid Market Turmoil — AI Agents Delivering 6–30% Shipment Lift in Production
Manhattan Associates management at the 46th Annual William Blair Growth Stock Conference, June 3, 2026
Manhattan Associates came into the William Blair Growth Stock Conference with a message that cuts against the broader anxiety in enterprise software: supply chain is not discretionary, and customers are not waiting for macro clarity to modernize. CEO Eric Clark, CTO Sanjeev Siotia, and newly-appointed CFO Linda Pinne delivered one of the more substantive management presentations of the conference season, with concrete data points on AI agent adoption, monetization structure, and a notable structural shift in the company's cost base. The headline number that will matter most to investors: Manhattan just closed its best Q4 in history from a sales bookings perspective, immediately followed by its best Q1 ever — during what Clark called "probably the noisiest" period for enterprise software in years, between Liberation Day tariffs, ongoing geopolitical conflict, and the so-called SaaSPocalypse narrative around AI displacing traditional software.
The Reduction in Force Is a Legacy Product Reset, Not a Business Signal
The conference opened with the announcement that landed earlier in the week: a roughly 6% global reduction in force. Clark was direct about the rationale. "We have never as a company de-supported any product, and we're not changing that. But we do have a number of legacy products that we have a shrinking number of customers running because we have more and more of those customers moving to our cloud products." The workforce action is explicitly tied to resizing the cost structure around those legacy on-premise systems, not any softening in core demand. The subtext is that Linda Pinne, stepping into the CFO role, prompted a sharper look at where spending was anchored in a part of the business that is structurally declining. The company's commitment has always been to invest more in sales and marketing without margin deterioration, and this move funds that shift.
A Code Generator Built Before AI Existed Is Now the Company's Most Important Asset
The most underappreciated structural advantage Manhattan possesses may be one that predates the current AI wave by more than a decade. Siotia revealed that twelve years ago, rather than writing traditional enterprise software, Manhattan's engineering team made the decision to build a code generator — and use that to generate the product itself. Today, 75% of Manhattan's code is machine-generated, with approximately 45 million lines of code produced every night from the codebase. "The reason was not efficiency really when we did this thing," Siotia noted. "The core driver behind that was how do we keep up as changes come along." The architectural philosophy — API-first, headless, true micro-services — was designed to survive interface paradigm shifts, from green screens to mobile to whatever came next. What came next turned out to be generative AI, and the architecture is ideally suited for it.
Siotia describes the resulting system as a "deterministic spine" into which probabilistic AI reasoning can be cleanly inserted. This distinction matters enormously in supply chain, where, as he put it, "almost correct is equal to wrong." Shipping 101 units when 100 were ordered can break a downstream process. The company's hybrid approach — keeping rules-based deterministic logic where it must be exact, and deploying LLMs only at specific decision nodes where probabilistic reasoning adds value — is increasingly where even AI-native competitors are landing. "Even people who kind of took that approach now are probably aligning with our point of view," Siotia said.
AI Agents Are Live in Production and Delivering Measurable Shipment Improvements
Manhattan is not in pilot purgatory. The company's agentic AI deployments are in active production use, and the early ROI data is striking. Siotia described the core use case: supply chain operations are fundamentally exception-management businesses, where a professional might spend three to six hours resolving a single issue — often long after the window to act has closed. "If there's a problem with the order not getting allocated, somebody who's chasing that allocation can take 2, 3 hours. And by the time they figure out that the inventory is sitting in the receiving dock, the truck has already kind of left and you missed the shipment." Agents resolving those exceptions in real time are driving shipment completion rate improvements of 6% to 30% for early customers — a range that, even at the low end, represents material financial impact for any operator running at scale.
Clark added an important architectural point on why Manhattan's agents reach production value faster than competitors: "We're never talking to our customers about data lakes or data indexing projects or the latency and the security risk that come with all of that. When we're doing AI on our system, we're doing it using the same APIs that a human user would use." Customers on the cloud platform can activate agents on day one with no data readiness project required. Competitors whose AI layer sits outside the core application face preparation timelines of three months to a year before first production use. The company has also built a dashboard allowing customers to see exactly which agents their teams are using and what value is being extracted — a direct response to enterprise CFO concerns about AI cost-to-value accountability.
Monetization Is a Tiered Subscription Uplift, Not a Usage Tax
Pinne clarified the commercial structure for AI agents, which had been a source of investor ambiguity. Manhattan is pricing agents as a percentage uplift on the base subscription, with tier selection tied to deployment scope — number of sites and number of agents — rather than token consumption. "If they only want to roll out one site or a small number of agents at first, they might be able to select a lower tier from a subscription standpoint. And then as they want to roll that out to more sites or they want to develop more agents, then they could choose to go up from there." The 90-day pilot program is functioning as a demand discovery mechanism, letting customers self-identify the value before locking in a tier. The structure avoids the token-maxing concern that Clark flagged is now top of mind for enterprise buyers.
Cloud Revenue Set to Surpass Services in Q4; Margin Expansion to Accelerate
Clark flagged a structural inflection point that investors should mark on their calendars: "In Q4 of this year, our cloud revenue will surpass services revenue. And once it does, that gap is just going to get bigger and bigger." Cloud margins exceed services margins, so the revenue mix shift has a direct read-through to operating margin expansion beyond the current trajectory. The company is growing services revenue in absolute terms this year, which Clark interprets as a signal of higher project velocity — more go-lives, more deployments — rather than services business health per se. The key leading indicator Pinne introduced at the start of the year is ramped ARR, which looks at contracted revenue in four years' time; that metric grew 23% year-over-year at the end of 2025, offering structural revenue visibility that RPO alone does not capture.
Fixed-Price Conversions and AI-Assisted Migration Are Removing the Primary Barrier to Cloud Adoption
One of the more operationally specific disclosures from Clark concerned the mechanics of on-premise to cloud conversion. The historical barrier for Manhattan's legacy customer base has been the perception — often accurate, based on decades of painful ERP upgrades — that migration is a large, expensive, multi-year risk. Manhattan is dismantling that perception with fixed-timeline, fixed-price conversions. The company already knows exactly what each legacy customer is running, knows the target cloud configuration, and is using AI-assisted tools to auto-configure the migration. Clark cited a 40% reduction in the number of extensions required, and the extensions that do remain are being written twice as fast. "We're committing to fixed timeline and fixed price conversions," he said — a commercial commitment that substantially de-risks the decision for the remaining on-premise base.
The dedicated conversion sales team, stood up within the past year, is now showing up in pipeline metrics. Cross-sell and upsell volume at time of renewal is growing significantly, with the pipeline reflecting deals not just for current quarters but for Q3 and Q4 renewals. The renewal team engages customers two quarters before contract expiration specifically to identify and position the full product expansion opportunity, not merely to secure the base renewal.
Multi-Product Attach Rates Are Rising Structurally, Not Cyclically
Manhattan's historical business was built on warehouse management, but the platform's integrated architecture is creating durable multi-product pull. Clark noted that roughly half of new logo customers buying warehouse management today are also purchasing transportation management at the same time — compared to essentially zero when transportation cloud was first introduced. The mechanics are compelling: because Manhattan built a unified micro-services platform rather than integrating separate applications, WMS and TMS share core data objects. A shipment is a single record accessible to both modules. Siotia illustrated the business case: "When you can do the planning and knew exactly what you have from an inventory perspective, so you can plan accordingly, and chances of you shipping a truck at the right capacity levels are much higher." That efficiency gain, he argued, can justify the combined WMS and TMS investment on transportation cost savings alone.
Supply chain planning, launched in the cloud just over a year ago, is already appearing as a common add-on across the customer base. Order management and point of sale continue to move together. The natural renewal cycle for the warehouse cloud cohort — first introduced five years ago, meaning those customers are now entering their first renewal windows — represents a structured multi-product expansion opportunity that will play out over the next several years.
Partner Ecosystem Shift Creates Net New Pipeline From Competitor Customer Bases
A year ago, Manhattan restructured its partner model, moving from a dynamic where partners largely followed Manhattan into deals chasing services revenue, to one where partners are explicitly expected to originate pipeline. The early indicators are positive. Clark described one partner running a six-city European tour targeting prospects who are current customers of Manhattan's competitors approaching renewal, with the explicit goal of educating them on Manhattan Active and funneling them to the EMEA Exchange user conference as active purchasing prospects. "We're seeing new deals that we wouldn't have otherwise seen," Clark said. Given that partners typically have deeper relationships with Tier 2 accounts — precisely the segment Manhattan is trying to penetrate more aggressively — the ecosystem shift has direct implications for addressable market expansion.
Tier 2 Market Expansion Is the Next Growth Vector, With More Specifics Coming in H2
Manhattan has historically defined its target market as Tier 1 and Tier 2, which together represent approximately 80% of total supply chain software spend. But Clark acknowledged that the lower end of Tier 2 has not always been fully accessible — not because Manhattan lacks capability, but because the brand carries an association with complexity and scale that creates self-selection bias among smaller prospects. The speed and cost reductions from fixed-price migrations and AI-assisted deployment are intended to lower total cost of ownership sufficiently to make Manhattan viable further down the Tier 2 spectrum. Clark indicated that the second half of 2026 will bring more formal announcements about how the company plans to expand its addressable market in that direction, suggesting a structured go-to-market initiative rather than an organic drift.
Siotia offered the most forward-looking framing of the session, articulating how he expects enterprise software to be evaluated five years from now: "It will be measured in terms of its IQ — the autonomous part and how much it can automate and bring intelligence — and its EQ, which is really the contextual intelligence and software adapting to users versus users adapting to software." That framing, whether or not investors adopt the terminology, points to a company that has spent a decade building the infrastructure for a market that is only now forming around it.