The work, in numbers.

Context

The recruitment team had no structured system to track, nurture or convert incoming applicants. Prospects arrived through multiple channels — the website, lead magnets, events — but there was no centralized workflow to move them through recruitment or hand them to admissions. We designed and implemented the full system: intake, qualification, nurturing and progression.

What was broken

No defined recruitment pipeline, inconsistent manual follow-ups, and no standardized journey from first touch to application. Communication, scheduling and engagement tracking were all hand-carried, which caused drop-offs, missed candidates and a hard ceiling on volume.

The constraints

• Multiple lead sources had to funnel into one pipeline
• Eligibility had to be checked before any nurturing began
• Eventbrite and scheduling tools had to drive stage changes directly
• No engineering resources: everything built in HubSpot and the automation stack
• The recruitment team needed real-time visibility throughout

The approach

• Treat recruitment as a structured pipeline, not a pile of ad-hoc conversations
• Nurture on behaviour: what an applicant does decides what they receive next
• Offer two conversion paths — group info session or 1:1 consultation — and let applicants choose
• Wire external tools directly into stage progression instead of copying data by hand
• Alert the team in Slack at milestones rather than asking them to check dashboards

The system, layer by layer

1. Capture and routing. Applications from every channel land in one pipeline, pass an eligibility check, and are routed accordingly.
2. Nurturing engine. Eligible applicants enter personalized email sequences driven by their pipeline stage.
3. Two conversion paths. Path A registers for an info session through Eventbrite; Path B books a 1:1 consultation. Both are tracked automatically.
4. Behaviour-based progression. Registrations, bookings and completed consultations move the applicant forward without anyone touching the record.
5. Team visibility. Slack notifications fire at the milestones the team cares about.
6. Admissions handoff. When an applicant applies, the system moves them to Applied and opens an admissions deal automatically.

What changed

• Manual recruitment workload down roughly 85%
• Hundreds of hours per week returned to the team
• Every applicant nurtured consistently, no one forgotten in a thread
• The funnel absorbed 4,000+ applicants within a few months
• Recruitment-to-admissions handoff runs itself

Before → after

Why it worked

The shift was treating recruitment as a lifecycle system rather than isolated manual tasks. Once progression was triggered by applicant behaviour instead of staff memory, the funnel became predictable — and volume stopped being a staffing problem.

Bring us a problem like this

Context

Clients purchased service hours in Shopify and described what they needed in free text. Staff interpreted every order by hand, posted each shift to the right regional Slack channel, and manually sent confirmations to both sides. We designed and delivered an end-to-end system across Shopify, HubSpot, Zapier, Slack, Gmail, OpenAI, Box and Google Drive in three months.

What was broken

Every step depended on a person: unstructured orders needed interpreting, channel routing was error-prone, shift claims had no conflict prevention (double bookings slipped through), confirmations were drafted by hand, records lived in scattered threads and spreadsheets, and reporting on volume, fill rate or revenue was impossible.

The constraints

• No engineering resources: no-code and low-code plus APIs only
• Client input stayed unstructured — the system had to meet it as-is
• Conflict prevention had to be real-time across all channels
• Sensitive client and attendant details required encrypted handling
• Files had to centralize across Box and Google Drive
• Full delivery inside three months

The approach

• Redesign the whole lifecycle as one connected pipeline instead of patching individual bottlenecks
• Put AI at the edge: parse unstructured orders into standardized postings so clients never change how they write
• Route dynamically from live HubSpot location data
• Make double booking structurally impossible with a first-claim ownership lock
• Automate the communications, then centralize tracking, files and reporting

The system, layer by layer

1. Order capture. Shopify orders and their free-text requests are ingested by Zapier in real time.
2. AI parsing. An OpenAI step interprets dates, times, services and instructions into a standardized shift posting.
3. Enrichment and routing. HubSpot client and location data is appended, and the shift posts automatically to the correct regional Slack channel.
4. Ownership lock. The first valid claim locks the shift; every later claim is rejected. Double booking cannot happen.
5. Confirmations. Encrypted emails go to client and attendant the moment a shift is claimed.
6. Tracking and files. A HubSpot pipeline carries status, ownership and revenue; Box and Google Drive handle documents automatically.
7. Reporting. Dashboards track volume, fill rate, utilization, regional throughput and revenue.

What changed

• Hundreds of hours of coordination and manual communication eliminated
• Double bookings ended by design, not by vigilance
• Cycle time cut from hours to near-instant
• No manual touch from Shopify order to confirmed shift
• First structured data layer for shift operations and revenue
• Higher volume no longer requires more headcount

Before → after

Why it worked

The win came from redesigning the full lifecycle as a connected pipeline instead of optimizing isolated tasks. AI absorbed the messiest input at the edge — unstructured client orders — so everyone upstream kept their habits while everything downstream became automatic.

Bring us a problem like this

Context

The organization ran two flagship programs, each with its own intake application. Applicants with a valid referral code received a 20% discount. Referrals ran through a paid third-party tool that created friction for alumni, applicants and staff alike. We were asked to replace it with something owned in-house.

What was broken

The paid tool compounded problems instead of solving them: a cumbersome multi-page flow for alumni, inaccurate tracking, a weak CRM integration that stored referral data as loose text, manual discount and invoice intervention on every redemption, no real-time visibility for the team — and a subscription bill for all of it.

The constraints

• No engineering resources: HubSpot and Zapier only
• Two programs, each needing its own tracking
• The alumni experience had to be zero-effort
• Discounts and invoicing had to run without touch
• Reporting needed referrer–referee links and per-program volume
• Day-one reliability: alumni trust was on the line

The approach

• Build natively in the stack already paid for, rather than swap one vendor for another
• Anchor the lifecycle to graduation: the moment someone becomes an alum, their referral code exists
• Store everything as structured CRM properties and associations, never loose text
• Capture redemption inside the existing intake forms
• Accept more upfront design complexity in exchange for permanent savings and full data ownership

The system, layer by layer

1. Trigger. A deal moving to Graduated starts the workflow.
2. Code generation. Zapier writes a unique referral code into dedicated HubSpot properties.
3. Distribution. HubSpot emails each graduate their personalized code automatically.
4. Redemption. A referral-code field sits directly in both programs’ intake forms.
5. Discount and invoice. A valid code applies the 20% discount, selects the right invoice template and sends it — untouched by staff.
6. Association. Referrer and referee deals are linked automatically, so attribution is queryable.
7. Visibility and reporting. Slack pings on every redemption; dashboards track volume by program, top referrers and conversion quality.

What changed

• Two successful referrals in the first week validated adoption immediately
• The third-party subscription was retired outright
• Alumni effort dropped from a multi-page flow to receiving a code automatically
• All manual discount and invoice handling removed
• Zero reported breaks after launch
• Referral data became structured and reportable for the first time

Before → after

Why it worked

Moving from a bolted-on tool to native lifecycle automation changed the physics of the process: data became structured, automation became reliable, and the cost line disappeared. The system also left room to grow — a pooled recognition draw for referring alumni was scoped as the natural next step.

Bring us a problem like this

Context

The business lacked structured lifecycle management, reliable attribution and revenue visibility. Every lead got the same treatment, marketing and sales disputed credit for closed deals, and invoicing restarted from scratch in another system after every win. We architected a connected go-to-market system spanning marketing, sales, onboarding and revenue intelligence.

What was broken

No lead qualification model separated high-intent buyers from early researchers. Sales and onboarding workflows were manual and inconsistent. There was no reliable attribution, no structured forecasting, and data sat in silos that made reporting untrustworthy.

The constraints

• Built entirely within HubSpot and the no/low-code ecosystem
• Typeform, PandaDoc, Calendly and QuickBooks had to plug in cleanly
• Both self-serve and enterprise journeys needed support
• High data accuracy across multiple pipelines was non-negotiable
• Architecture had to stay simple enough that the team would actually use it

The approach

• Design one connected operating system instead of isolated workflow fixes
• Make the CRM the single source of truth for every stage of the funnel
• Score leads on behavioural, intent and demographic signals — including negative signals
• Settle the attribution argument with multi-touch data, not opinions
• Connect closed-won directly to invoicing so revenue flows without re-keying

The system, layer by layer

1. Lead scoring. MQL → SQL scoring built on behavioural, intent and demographic signals, so handoffs fire on evidence.
2. Pipeline architecture. Interconnected Sales, Onboarding, Account Growth and Retention pipelines mapped to lifecycle stages.
3. Automated handoffs. Sales-to-onboarding transfer carries deal data and generates stage-based tasks; self-serve and enterprise journeys branch automatically.
4. Documents and scheduling. Typeform prefills intake, PandaDoc generates contracts and approvals, Calendly drives scheduling.
5. Attribution. Multi-touch attribution across campaigns, forms and demo requests shows what actually sources revenue.
6. Quote to cash. Deal stages trigger invoicing in QuickBooks and reconcile payments back, so closed-won flows through to paid.
7. Forecasting. Dashboards project pipeline and revenue from stage probabilities, conversion rates and historical performance.

What changed

• A fully connected GTM system replaced disconnected tools and habits
• Lead qualification and lifecycle stages aligned marketing and sales
• Attribution made marketing ROI a data question, not a debate
• Leadership gained pipeline and revenue forecasts it could plan on
• Manual coordination dropped; reporting became trustworthy
• Closed-won now runs straight through to invoice and payment

Before → after

Why it worked

Connecting lifecycle execution, attribution and forecasting into one system — rather than three tools with three owners — is what made each part reliable. The complexity lives in the architecture so the day-to-day stays simple for the team.

Bring us a problem like this