SPARK
Expert AI · Process Discovery

Project
SPARK

S
Surface
P
Prioritize
A
Automate
R
Reduce
K
Know-how

Project SPARK is Haworth's structured initiative to discover, prioritize, and build automation solutions across HR and operational functions. It starts with listening — going department by department to surface the hidden manual work that consumes member time, creates risk, and prevents our best people from doing their best work.

11
HR Functions InterviewedBenefits, Payroll, Comp, Recruiting, Data Services, OccHealth, Corp Security, and more
100+
Hours Targeted for RecoveryWave 1 alone targets over 100 hours of manual work reclaimed annually
3
Implementation WavesPilot → Scale → Enterprise, each building on the last
48h
Single Process ExampleThe Velocity Report alone consumed 48 hours per year — now automated
🔍
Phase 1
Discover & prioritize
Phase 2
Build & validate
🚀
Phase 3
Deploy & scale
🛠️
Primary Tool
Alteryx · Power Automate · Copilot
👥
Approach
Member-led, AI-enabled

What We're Looking For

Six types of
automatable work.

SPARK discovery interviews consistently surface the same patterns across functions. These are the six opportunity types that appear most frequently — and deliver the highest return when addressed.

📋
Pattern 01
Recurring Report Preparation
Teams manually pull data from multiple systems, reformat in Excel, validate, and distribute — every week or month. This single pattern accounts for the majority of automatable hours discovered across all SPARK interviews.
Found in: Finance (monthly close reports), Operations (production metrics), Sales (pipeline reports), HR (weekly velocity reports)
🔄
Pattern 02
System-to-System Data Bridging
Data lives in two or more systems that don't talk to each other. Members copy, paste, VLOOKUP, or manually match records to create a combined view. Every SPARK interview surfaced this pattern — often multiple times.
Found in: Supply Chain (ERP ↔ WMS), Finance (GL ↔ budgeting tools), Sales Ops (CRM ↔ forecasting platforms)
📁
Pattern 03
File Build & Upload Cycles
Teams build a formatted file — always the same structure — from combined data sources, then manually upload it to a vendor portal or system. The format never changes. The process does nothing that a machine can't do faster and without errors.
Found in: Finance (vendor payment files), Operations (MES batch uploads), Compliance (regulatory submission files)
⚠️
Pattern 04
Single-Owner Process Risk
A critical process — often compliance-related — exists only in one person's head. No documentation. No backup. If that person is unavailable, the process stops or breaks. Automation adds resilience while also creating documentation as a byproduct.
Found in: Every SPARK interview, across every function. Named single owners of compliance steps, audit-critical data, and recurring exception handling.
🔔
Pattern 05
Manual Notifications & Follow-ups
Members send recurring reminder emails, follow up on approvals, or trigger downstream actions manually — work that follows a consistent rule and could be automated with simple workflow triggers.
Found in: Procurement (PO approval follow-ups), Legal (contract renewal reminders), IT (access request status updates)
📊
Pattern 06
Manual Dashboard & Metric Prep
Leaders wait hours or days for data they need to make decisions. Every metric request requires someone to build a custom report from scratch. Automation replaces the build cycle with a self-serve data layer that updates automatically.
Found in: Operations (shift performance dashboards), Finance (executive KPI packs), Marketing (campaign performance summaries)
High-Value Signals
Process takes the same steps every cycle
Output format never changes
Data comes from 2+ systems that don't connect
One person owns it with no documented backup
Risk Signals
Compliance or legal exposure if errors occur
Deadline pressure creates regular error windows
Rework is a normal part of every cycle
Process undocumented — tribal knowledge only
Scale Signals
5+ hours per week spent on repetitive prep
3+ recurring reports produced per month
Team uses Excel as a primary system of record
Shadow systems created due to access frustration

How SPARK Works

A structured path from
discovery to deployment.

SPARK uses a four-step methodology adapted from process improvement best practices. Every step is designed to keep the work grounded in real member experience — not assumptions.

01
Step 1 · Discovery
Structured Listening Sessions
We sit with each team and ask them to walk us through their most time-consuming, repetitive, or frustrating work — with no assumptions about what's automatable. The goal is to uncover the "hidden" effort that teams have stopped noticing because it's just the way things get done. Every session produces a raw backlog of candidate processes.
What we document
Process name and frequency (daily, weekly, monthly)
Systems involved and integration gaps
Estimated hours per cycle and annual total
Error frequency and consequence of mistakes
Owner count and documentation status
02
Step 2 · Prioritization
Scoring & Ranking
Each identified opportunity is scored against four criteria: value and impact, data and technology readiness, organizational readiness, and risk and compliance exposure. This ensures we tackle the highest-return, most feasible work first — not just what's loudest or most visible. The output is a prioritized opportunity map.
Scoring criteria
Time savings and error reduction potential
Data quality and system accessibility
Team readiness and change appetite
Compliance or business continuity risk
03
Step 3 · Build
Automation Development
High-priority opportunities move into build — using Alteryx for data automation, Power Automate for workflow triggers, and Copilot for AI-assisted tasks. The SME who owns the process stays involved throughout build and testing. This ensures the solution matches real-world nuance and that the team has confidence before deployment.
Primary tools
Alteryx — data blending, transformation, report automation
Power Automate — workflow triggers, notifications, approvals
Microsoft Copilot — AI-assisted drafting and analysis
Copilot Studio — custom agents and integrations
04
Step 4 · Deploy & Sustain
Rollout & Continuous Improvement
Automation is deployed with full documentation, member training, and a handoff plan. Every solution ships with an owner, a process document, and a feedback loop. After each wave, the AI Transformation team measures outcomes against baseline KPIs and uses what we learn to refine the prioritization model for the next wave.
What we measure
Hours reclaimed vs. baseline estimate
Error rate before and after automation
Member confidence and adoption rate
Business continuity improvement

What SPARK Delivers

Real results from
real discovery.

Project SPARK began in Haworth HR — and these results are from that initial discovery. Every row represents a real process found, prioritized, and addressed. The same methodology is now being applied across all business functions.

Process / Function What Was Found Automation Applied Result
Velocity Report
HR Data Services
4-hour manual preparation process every cycle — data pulled from multiple systems, reformatted, validated by hand Alteryx automation — push-button report generation replacing all manual steps 4 hrs → push button · 48 hrs/yr reclaimed
Pension File Preparation
Benefits
Single owner with sole access to vendor portal. Manual file build every cycle with compliance consequences for errors Automated file assembly and upload process with backup documentation Single-owner risk eliminated · compliance exposure reduced
Step Pay Upload
Compensation
Errors found "almost every cycle" — manual data entry from multiple sources into upload format for ADP submission Automated data pull, format validation, and upload file generation Error rate reduced · cycle time cut significantly
Roundtable Notes Processing
HR Communications
24+ hours to process, summarize, and distribute notes from leadership roundtables AI-assisted summarization and structured output generation 24 hrs → 8 hrs · 66% reduction in processing time
Hearing Test Scheduling
Occupational Health
OSHA compliance requirement tracked manually — year-end pressure created compliance risk annually Automated scheduling triggers and compliance tracking workflow Compliance risk reduced · manual tracking eliminated
"
Every team we talked to had processes they'd stopped questioning — work that just got done because that's how it always got done. SPARK gives those processes a second look. And almost every time, we find something worth fixing.
Project SPARK · Discovery Insight · Haworth AI Transformation

Get Involved

Three ways your
team can participate.

SPARK works because members are at the center of it. The AI Transformation team provides the methodology and builds the solutions — but the insights come from the people doing the work.

🗣️
Option 01
Request a Discovery Session
Schedule a structured listening session with the AI Transformation team. We'll spend 60–90 minutes mapping your team's most time-consuming, repetitive processes and identify your highest-value automation candidates. No technical knowledge required — just your team's experience.
📋
Take the Readiness Assessment
Option 02
Before your discovery session — or on your own — complete the Department AI Readiness Assessment. It takes about 12 minutes, covers seven domains, and generates a custom SWOT analysis and prioritized roadmap specific to your department's situation.
🏆
Option 03
Become an AI Champion
Every successful SPARK build has an internal champion — someone on the team who owns the process, participates in testing, and helps their colleagues adopt the solution. Champions accelerate results and build lasting capability in their department. If that sounds like you, let us know.

Ready to find your SPARK?

The work is already there.
We just need to find it.

Every department has hidden effort — processes that consume member time, create risk, and could be automated. SPARK is how we find them.