← Findings 9 min read

What Happened at Companies Our Size? Five Mid-Market AI Deployments With Measured Outcomes

Brandon Sneider | March 2026

Executive Summary

  • Five American companies with $275M-$655M in revenue deployed AI to specific operational bottlenecks — not open-ended experiments — and produced measurable P&L results within 6-12 months.
  • The two clearest ROI stories are in accounts payable and sales intelligence: Purple Innovation cut invoice processing time by 63% and eliminated the need for an additional hire. Paycor’s sales team achieved a 141% increase in closed deals per seller after deploying AI-powered pipeline management.
  • Every company on this card shares the same pattern: they automated a specific, repetitive workflow where the before-and-after could be measured in hours, dollars, or headcount avoided — not “AI transformation.”
  • The cautionary signal is equally consistent: none of these companies started with a company-wide AI strategy. They started with one problem, one team, and one metric.

The Five Companies

1. Paycor — Sales Intelligence (HCM Software, $655M Revenue, 2,900 Employees)

Paycor provides human capital management software to 40,000+ businesses. Its 54-person client sales team manages approximately 3,000 deals in the pipeline at any given time and closes roughly 1,000 deals per month. Before deploying AI, sales leaders lacked visibility into which deals were winnable and could not forecast accurately across that volume.

What they deployed: Gong’s AI-powered revenue intelligence platform — call summaries, recommended next steps, pipeline prioritization, and an “Ask Anything” natural language interface for deal context.

Measured outcome: 141% increase in closed-won deals per seller on the client sales (upselling) team. Forecasting accuracy improved. Sellers spend less time on administrative deal tracking and more time on high-probability opportunities.

What made it work: The AI does not replace sales reps. It surfaces which deals are winnable and summarizes call context so sellers focus effort where probability is highest. The productivity gain comes from prioritization, not automation.

Source credibility: Medium. Gong case study (2024-2025) — vendor-published with a single headline metric. No dollar amount, no control group. The 141% figure measures upselling deals per seller, not total revenue. Paycor’s acquisition by Paychex for $4.1B in January 2025 provides independent validation that the company’s sales performance trajectory was real.

2. Purple Innovation — Accounts Payable (Comfort Products Manufacturing, $488M Revenue, Public: PRPL)

Purple designs and manufactures mattresses, pillows, and cushions, selling through 60 company stores and over 3,000 retailers. The company processes over 1,100 invoices per month from approximately 1,350 vendors. Before automation, invoices took 8 days to process and only 25% had proper supporting documentation attached.

What they deployed: Stampli’s AP automation platform with OCR for GL coding, machine learning for PO matching (their “Billy AI”), and automated approval routing based on invoice amounts and purchase type.

Measured outcome: Invoice processing time dropped from 8 days to 3 days — a 63% reduction. Documentation compliance went from 25% to 100%. The AI now auto-processes 65% of PO-matched invoices without human intervention. The efficiency gain eliminated the need to hire a third accounts payable staff member, saving approximately $55,000-$75,000 annually in fully loaded headcount cost.

What made it work: Purple automated a structured, repetitive process with clear before-and-after metrics. AP has a known cost per invoice (industry benchmark: $15-$22 manually vs. $2-$4 automated). The AI handles the matching; humans handle exceptions. This is the “boring AI” pattern that produces the most consistent mid-market ROI.

Source credibility: Medium-high. Stampli case study (2025) — vendor-published but with specific, verifiable metrics across four dimensions. Purple is publicly traded (PRPL), so financial context is independently verifiable. Processing metrics align with APQC benchmarks.

3. Jamf — IT Service Automation (Apple Device Management, $627M Revenue, ~2,600 Employees)

Jamf manages over 30 million Apple devices for 75,000+ organizations worldwide. As a fast-growing technology company, its internal IT team was drowning in routine service requests — password resets, software provisioning, distribution list updates, onboarding and offboarding — with software request cycles averaging 8+ days.

What they deployed: Moveworks’ agentic AI assistant (internally named “Caspernicus”), integrated into Slack, ServiceNow, Active Directory, and Okta.

Measured outcome: 70%+ of employees actively use the AI assistant. Software request cycles dropped from 8+ days to minutes. Adoption hit 30% in the first month and has climbed steadily since. Routine IT tasks that previously required human ticket processing now resolve instantly through the AI.

What made it work: Jamf deployed to a single channel (Slack) where employees already worked, eliminating the adoption friction that kills most enterprise AI rollouts. The AI handles structured requests with clear resolution criteria. Complex or ambiguous tickets still route to humans.

Source credibility: Medium. Moveworks customer story (2025) — vendor-published. Adoption metrics are specific but no cost savings disclosed. Melissa Dunham, Senior Director of IT Support and Experience, is quoted by name, adding accountability to the claims.

4. BambooHR — Employee Support Automation (HR Software, ~$275M Revenue, ~1,800 Employees)

BambooHR provides HR management software to over 30,000 businesses in 190 countries. During a period of hypergrowth — relocating headquarters, onboarding new employees, rolling out new benefits — its lean support team could not keep pace with tier-1 service requests.

What they deployed: Moveworks’ AI platform integrated with BambooHR’s internal knowledge base, plus Agent Studio (low-code automation) for custom workflows connecting to tools like Jamf and NinjaOne.

Measured outcome: 30% reduction in overall support ticket volume. Employees get instant answers to onboarding, IT troubleshooting, and benefits questions instead of waiting for human agents. The lean support team was freed to build an IT business partner model, in-office tech improvements, and new departmental workflows — work that was previously impossible given ticket volume.

What made it work: BambooHR integrated AI with its existing knowledge base first, creating immediate value by deflecting the questions employees were already asking. The 30% reduction came from eliminating repetitive tier-1 queries, not from replacing the support team. The team is the same size — it just works on different problems now.

Source credibility: Medium. Moveworks blog and video (2024-2025) — vendor-published. The 30% figure is specific and attributed to named employees (Kelby Crandall). No dollar savings disclosed. BambooHR’s $2.5B valuation (August 2024, Providence Equity) suggests the company’s operational efficiency trajectory is independently validated.

5. Newman’s Own — Knowledge Work Amplification (Consumer Packaged Goods, ~$600M Gross Sales, 50 Employees)

Newman’s Own competes in the CPG food space against multinationals with thousands of employees. The company donates 100% of after-tax profits to charity, making operational efficiency directly tied to mission impact. With a 50-person team, every hour saved produces measurable leverage.

What they deployed: Microsoft 365 Copilot across marketing, legal, logistics, and research functions.

Measured outcome: Marketing campaigns tripled in monthly volume. Market research that previously required two staff members spending two days now takes one person two hours. Industry news summarization went from half a day to 30 minutes. The Chief Legal Officer describes Copilot as “my junior associate” — handling legal questions and gathering citations that previously consumed hours of attorney time.

What made it work: Newman’s Own is the clearest case of AI amplifying a lean team rather than replacing headcount. CEO David Best: “We have 50 people, and we don’t have the resources of some of the multinational conglomerates we face.” The AI does not replace any of those 50 people. It makes each one more productive in areas where the bottleneck was research, summarization, and first-draft creation — tasks where generative AI’s evidence base is strongest.

Source credibility: Medium-high. Microsoft customer story (2025) and PYMNTS.com coverage with named executives (CEO David Best, CLO Jennifer Millones, CPO Bruce Wallace, Social Media Manager Riley McCarthy). Microsoft is a vendor, but the story is corroborated by independent coverage and includes multiple named sources with specific claims.

Key Data Points

Company Revenue Employees AI Application Headline Result Cost/ROI Data
Paycor $655M 2,900 Sales intelligence (Gong) 141% more deals per seller Acquired for $4.1B (Paychex, Jan 2025)
Purple Innovation $488M ~1,000 est. AP automation (Stampli) 8-day to 3-day processing (63% faster) ~$55K-$75K saved (headcount avoided)
Jamf $627M ~2,600 IT service automation (Moveworks) 70%+ adoption, 8+ days to minutes Not disclosed
BambooHR ~$275M ~1,800 Employee support (Moveworks) 30% ticket reduction Valued at $2.5B (Aug 2024)
Newman’s Own ~$600M gross 50 Knowledge work (M365 Copilot) 3x campaign output, 90% research time saved 100% profits to charity — efficiency = mission

The Pattern: What These Five Companies Did That Most Don’t

Three things separate these five companies from the 95% of AI deployments that produce no measurable P&L impact (MIT NANDA, n=300+, 2025):

They picked one bottleneck, not an “AI strategy.” Paycor did not transform its entire go-to-market motion. It gave 54 sellers a tool that surfaces which deals to focus on. Purple did not reimagine its entire finance function. It automated the PO-matching step in invoice processing. The scope was narrow enough that one team could own the outcome and one metric could prove the value.

They measured before they deployed. Purple knew invoices took 8 days and only 25% had documentation attached. Paycor knew sellers could not manage 3,000 pipeline deals manually. Newman’s Own knew market briefs took two people two days. The before number is what makes the after number credible — and what makes the CFO sign off on expanding the program.

They automated the repetitive 80%, not the judgment 20%. In every case, the AI handles structured, repetitive tasks: matching POs, summarizing calls, answering FAQ-level support tickets, condensing news articles. In every case, humans handle exceptions, complex decisions, and anything requiring judgment. This is the “boring AI” pattern from the corpus research — and it is the only pattern with consistent evidence of mid-market P&L impact.

What This Means for Your Organization

These five companies are not exceptional. They are normal mid-market companies — $275M to $655M in revenue, 50 to 2,900 employees — that applied AI to problems with measurable baselines and clear success criteria. The results came from discipline in scoping, not sophistication in technology.

The relevant question for your organization is not “should we use AI?” — 91% of mid-market companies already do (RSM, n=966, 2025). The question is whether your first deployment targets a specific workflow with a known cost baseline and a named owner who reports results in 90 days.

If these case studies raised questions about where to start in your organization — which bottleneck to target first, how to measure the baseline, or how to structure the first 90 days — I’d welcome the conversation: brandon@brandonsneider.com

Sources

  1. Gong, “Paycor Customer Story” (2024-2025). Vendor-published case study. 54 reps, ~1,000 monthly deals, 141% increase in deal wins per seller. Medium credibility — single metric, no dollar amounts. https://www.gong.io/customers/case-studies/gongs-revenue-intelligence-fuels-a-141-surge-in-paycors-client-sales-success

  2. Constellation Research, “Paychex acquires Paycor in $4.1B deal” (January 2025). Independent analyst coverage of Paycor’s acquisition. https://www.constellationr.com/blog-news/insights/paychex-acquires-paycor-41-billion-deal

  3. Stampli, “Purple Customer Story” (2025). Vendor-published case study. 1,100+ invoices/month, 1,350 vendors, 8-day to 3-day processing, 63% reduction. Medium-high credibility — specific, multi-dimensional metrics, publicly traded company. https://www.stampli.com/case-studies/purple/

  4. Purple Innovation, “Fourth Quarter and Full Year 2024 Results” (February 2025). SEC filing. $487.88M annual revenue. High credibility — public company financial disclosure. https://investors.purple.com/news/news-details/2025/Purple-Innovation-Reports-Fourth-Quarter-and-Full-Year-2024-Results/default.aspx

  5. Moveworks, “Jamf Customer Story” (2025). Vendor-published case study. 70%+ adoption, 30% first-month adoption, 8+ day to minute-level request resolution. Medium credibility — adoption metrics only, no cost data. https://www.moveworks.com/us/en/customers/jamf-employees-use-moveworks-ai-powered-support

  6. Jamf Holding Corp. (JAMF), “Revenue” (2024-2025). Public company financial data. $627.4M annual revenue, ~2,600 employees. High credibility. https://stockanalysis.com/stocks/jamf/revenue/

  7. Moveworks, “How BambooHR Reduced Ticket Volume by 30%” (2024-2025). Vendor-published case study and video. 30% ticket reduction during hypergrowth. Medium credibility — specific metric, named employee. https://www.moveworks.com/us/en/resources/blog/bamboohr-ai-customer-story

  8. GetLatka, “BambooHR Revenue” (2024). Third-party SaaS metrics tracker. $274M revenue, 26K customers. Medium credibility — aggregated from public and estimated data. https://getlatka.com/companies/bamboohr

  9. Microsoft, “Newman’s Own Customer Story” (2025). Vendor-published case study. 50 employees, tripled campaign output, research time reduced by 90%. Medium-high credibility — multiple named executives, corroborated by independent coverage. https://www.microsoft.com/en/customers/story/23048-newmans-own-microsoft-365-copilot

  10. PYMNTS.com, “Newman’s Own Uses GenAI to Compete With Consumer Products Giants” (2025). Independent trade publication coverage with executive quotes. Medium-high credibility. https://www.pymnts.com/artificial-intelligence-2/2025/newmans-own-uses-genai-to-compete-with-consumer-products-giants/

  11. RSM/Big Village, “Middle Market AI Survey 2025” (n=966, U.S. and Canada, March 2025). 91% of mid-market companies use AI. High credibility — independent professional services firm. https://rsmus.com/insights/services/digital-transformation/rsm-middle-market-ai-survey-2025.html

  12. MIT NANDA, “The GenAI Divide: State of AI in Business 2025” (n=300+ initiatives, 52 interviews, 153 survey responses, 2025). 95% of AI pilots produce no measurable P&L impact. High credibility — independent academic study. https://mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf

Brandon Sneider | brandon@brandonsneider.com March 2026