I find the workflow everyone works around, and turn it into the one that runs itself.

The Situation

• Each client company experienced their engagement through a patchwork of shared Google Sheets, emailed PDFs, and status updates. Real insights only surfaced in infrequent meetings.
• Every progress update was hand-assembled from data scattered across 8 Monday.com boards.
• No self-serve option existed for clients to see their own data: program insights, attendance, progress, feedback, or schedule.

The Strategy

• Prove the concept with one client's use case first, using Monday.com as the live source of truth so there is no separate sync step.
• Expand that proof of concept into an architecture that works across every engagement type, built from a library of reusable page templates.
• Goal: shift client-facing time away from status updates and toward discussing insights and solving problems, freeing up team bandwidth to build real client relationships.

Also worth knowing

• Credentials never touch the browser, and small-group feedback stays anonymous by design until enough people have responded.
• Scheduling is automated end to end: clients approve proposed session dates, and the system notifies the team and creates the sessions automatically.
• Built a fully anonymized demo version for the sales team, plus complete documentation so anyone on the team can maintain the platform.

• Monday.com GraphQL API with a server-side proxy so API credentials never reach the browser
• Library of reusable page templates assembled per client, one codebase for six engagement types
• Tolerant name matching across training, coaching, intake, and historical feedback boards
• Anonymity gates requiring 3+ responses before NPS and before/after measures display
• 21-check automated verification suite, including a zero real-name sweep of the demo build

The Situation

• Producing each leader's 360 report was fully manual: copying survey data between spreadsheets, relabeling rater categories, merging small groups for anonymity, refreshing charts, and generating the PDF by hand.
• On the intake side, leaders relied on one-off emails to understand the process, chase responses, and check their status.
• At 10 to 20 reports per cohort, this was one of the biggest recurring time sinks in delivery, and every manual step was an error risk on a client-facing document.

The Strategy

• Automate everything the platform allows inside the existing reporting pipeline rather than rebuilding a stack that already worked.
• Flag the few remaining manual steps exactly when they are needed, so nothing depends on memory.
• Make the intake side self-serve so leaders can run their own process without hand-holding.

Also worth knowing

• Built-in anonymity guardrails flag when a page must be dropped, so no small group is ever identifiable.
• A public instructions site gives leaders a self-serve status tracker, plain-language anonymity rules, and a copy-paste survey email.

• Google Apps Script menu inside the source sheet driving the full data-prep sequence
• Rater relabeling, small-group merge logic, and eNPS calculation handled in code
• Existing Sheets-to-Slides merge pipeline retained, with setup documentation for template changes
• Instructions site and status tracker built as a static site on Netlify

The Situation

• Session and coaching feedback lived in a shared spreadsheet plus cohort docs scattered across Google Drive.
• The coaches who delivered our sessions had no way to see their own feedback history without someone compiling it by hand.
• Years of historical feedback was effectively invisible to the people it was about.

The Strategy

• Centralize everything in Monday.com and build a coach-facing dashboard on top of it.
• Iterate directly on named coach feedback rather than guessing at requirements.
• Design for trust: coaches see only their own feedback, and anonymity rules are enforced in code, not by policy.

Also worth knowing

• A one-time migration brought 222 historical feedback docs into the searchable hub, going back to 2021.
• A two-page coach guide and rollout announcement, written from the actual app behavior so every claim is accurate.

• Built as a Monday.com Vibe app with per-coach auto-filtering and admin access controls in code
• AI sentiment labeling pipeline over written feedback responses
• Anonymity thresholds enforced before any group metric displays
• Scripted one-time migration from Google Drive with per-record verification

The Situation

• The entire leader operations workflow, from deal ping to continued coaching, lived in one person's head: mine.
• Fourteen phases across roughly a dozen connected tools and boards, with no map anyone else could follow.
• Any absence or handoff created a bottleneck, with no way for anyone else to pick the work up.

The Strategy

• Capture the real workflow live through narrated walkthroughs instead of writing documentation from memory.
• Treat the documentation as a discovery tool: watching the actual work surfaces the workflows and automations that need correcting.
• Turn what surfaced into a prioritized improvement roadmap, now actively in progress through Q3.

When it misfires

• The skill is only ever as good as the playbook underneath it. When answers came back thin on a few phases, the cause was missing documentation, not a broken skill. I checked those phases against a real project record and found steps nobody had written down.

The guardrail

• It can brainstorm a change and work out what that change would affect elsewhere, but it never acts. It flags the idea to a person. That was deliberate: the team doing the work should own the process and the ideas for improving it, not follow steps handed down.

Also worth knowing

• A gap audit checked the playbook against a real 117-task cohort project plan to catch undocumented steps.
• Documentation targets are operationalized into quarterly goals with an audit cadence and usability testing.

• Narrated screen-share walkthroughs of each phase, structured into a consistent phase template
• Cross-audit against a live 117-task cohort project plan
• Improvement backlog ranked by impact and folded into quarterly goals

The Situation

• Consulting engagements kept reaching the same point: several discovery sessions in, what was left were pointed questions, not another meeting.
• How a business actually runs lives in the owner's head. Most have never had to state any of it out loud.
• Nobody can recite their own policy from memory. Everybody can tell you when a proposed one is wrong.

The Strategy

• Stop asking people to describe their rules. Propose the rules from their own data and let them correct what is wrong.
• Let the disagreements do the work: where the proposal and reality diverge is exactly where the conversation needs to happen.
• Build it so the client can run it themselves, which means the engagement never requires access to their accounts.

When it misfires

• It proposed a response-time rule from actual reply history, slower for prospects and lower-priority contacts, because that is what the data showed. The client overrode it on the spot and wanted same-day for those too.
• The stated standard was stricter than the observed behavior, and that gap is where the real rule lives. Building from data alone would have encoded a lower standard than the one they hold themselves to. The fix is always letting the person correct it.

The guardrail

• No access to a client's inbox or files, by design. They grant the permissions in their own account, run it there, and send back the document they choose to send.
• Read-only regardless: it can see everything and change nothing, and every finding comes back as a proposal to accept or reject.

• Designed to run inside the client's own account, so I never hold their credentials
• Read-only access, granted by the client and switched off by them whenever they want
• Every finding arrives as a draft for a person to approve, never as a change already made
• The output is a document the client reviews and decides whether to share