20 reviews from the
teams we shipped for.

We'd been carrying Salesforce for six years and the workarounds had workarounds. The Stack Renew audit was the first time anyone walked our team through which fields and automations actually moved revenue versus which ones existed because someone left in 2022. The replacement shipped in twelve weeks on Next.js and Postgres, parallel-ran for five, and the cutover was uneventful — which, for a system this load-bearing, is the highest praise I can give.

Three years of fighting our own website. Every campaign launch we'd discover another plugin had broken something, and every Friday somebody was applying a security patch instead of working on growth. Michael's team rebuilt the site on Astro with a headless CMS and a plain-English editor — I genuinely write copy by describing what I want now. Mobile LCP went from 4.2s to under a second and our organic traffic is up 38% ninety days in.

Seven figures a year on NetSuite for a database we barely used 30% of. The Stack Renew audit produced something I'd never seen from a consultant: a per-module honesty pass on what we actually used versus what was vestigial. The replacement covered the modules that mattered — GL, inventory with lot tracking, order-to-cash — on a custom stack with ERPNext under the hood for the boring middle. Statutory tax stayed on Avalara, where it belongs. Ten months from audit to running parallel, fourteen to full cutover.

Forty-six Tableau seats and maybe twelve people who used it more than monthly. Stack Renew shipped a custom React + Postgres dashboard for the views we actually checked daily, plus an AI-SQL layer that lets ops ask new questions without filing a ticket. Eight weeks. The seat licenses we retired paid back the build inside a quarter.

We had a Retool sprawl problem — sixteen apps, three abandoned, one person who knew how any of them worked. Stack Renew did the unglamorous work of consolidating the live ones into a single Next.js admin against our existing Postgres, killing the rest, and documenting what was left. Seven weeks. Our on-call rotation got noticeably quieter.

We outgrew our Shopify theme during a launch that should have been our biggest quarter and instead became our worst. Stack Renew came in for a two-week audit and told us — to our faces — that 60% of the apps we were paying for were doing nothing. They rebuilt the store as a Next.js storefront on the Shopify Storefront API, kept what mattered, killed the rest. The honesty was the differentiator.

I was the skeptic. Every vendor for two years had pitched us an "AI lead scoring engine" and they were all dressed-up regression on the same five fields. Stack Renew shipped a gradient-boosted model with eighteen months of our own data, an honest 0.89 ROC-AUC on holdout, and weekly retraining wired into our actual outcomes. Three months in, my reps are working a ranked queue and our close rate on A-grade leads is up almost 40% quarter-over-quarter.

We were three full-time hires deep into a queue that kept growing. The pitch was a Textract + GPT-4 pipeline with human-in-the-loop for low-confidence items, classification accuracy in the high 90s on validation, and an honest forecast of where it would fail. Nine weeks from kickoff to production. Ninety percent of documents now route without a human touch, and the three roles we had on data entry moved to client work they actually wanted to do.

I was tired of vendors pitching me a chatbot that would "deflect 40% of tickets." Stack Renew shipped something more boring and more useful: a copilot in the agent UI that drafts responses, cites our docs, and learns from corrections. They wrote the eval harness first — I watched them argue about it for an afternoon — and the accuracy numbers were honest from day one. Median handle time down 31% inside a quarter.

We tested four "AI contract review" SaaS products. All four sent our drafts to a third-party LLM endpoint in ways our clients would not have approved. Stack Renew built ours on a dedicated VPC with an open-weights model, the document never left our infrastructure, and the structured extraction handles the ten clause types our practice cares about. Privilege boundary intact. That was non-negotiable for us.

Our sales forecast was Excel and intuition, and the variance was making the board nervous. Stack Renew built a Python forecasting service on top of our CRM and billing data — a real ML pipeline with backtesting, lineage, and weekly retraining, not a single regression in a notebook. Forecast accuracy has held within ±6% for three consecutive quarters. The board is no longer nervous.

We were on a Shopify theme that had been "we'll rebuild it next quarter" for two years. Stack Renew shipped a Next.js headless storefront against the Shopify Storefront API, moved search to Algolia, kept Stripe for checkout, and pulled merchandising into Sanity so my team finally got out of the loop on copy changes. LCP under 2 seconds on mobile, conversion up 40%, and the engineering Slack channel is quiet again.

I taught the first cohort on Zoom and a Google Drive folder. The plan was always to graduate to a real platform; the off-the-shelf LMS options either looked like 2014 or charged me per student. Stack Renew shipped a branded React + Node + Mux platform in thirteen weeks with Stripe Billing for cohorts and a quiz engine that does auto-grading and remediation. Year one we onboarded ten thousand students at 86% completion. I run the business from the admin panel.

Twelve warehouses, three sales channels, and a spreadsheet that broke every time someone left a tab open. We oversold a Black Friday SKU two years running. Stack Renew built a React admin with a Node sync service against Shopify, Amazon, and our B2B portal — six weeks, fixed bid. Zero oversells since launch. The reorder triggers handle 80% of SKUs without anyone touching them.

Three tools, weekly CSV exports, no attribution anyone trusted. Stack Renew unified our customer data in Segment + BigQuery, built segmentation and drip execution on Node, and gave marketing a Next.js admin they actually use. Two quarters in, campaign ROI is 5× the baseline and our MQL-to-SQL is up to 32%. The honest part: they were realistic about what AI/ML belonged in this build versus what was just rules.

I review every vendor that touches PHI. Most "HIPAA-compliant" pitches are a pdf and a prayer. Stack Renew showed up with an audit log spec, an encryption posture, and a BAA list before we even signed. The FHIR bridge to our EHR landed clean, the WebRTC implementation has held 99.95% call uptime in production, and our no-show rate dropped from 28% to 11% in the first year. They earned the trust.

Our portfolio managers had been working off five reports that refreshed overnight, which meant risk signals showed up a day late. Stack Renew built an Airflow ingest pipeline into Postgres + Redis, D3 dashboards over WebSockets, sub-second refresh across eleven thousand instruments. Six legacy reports decommissioned. The infrastructure cost is a fraction of what we were paying for the BI seat licenses it replaced.

We needed iOS + Android in production before a contract milestone and our last attempt with a different shop had stalled at "almost". Stack Renew picked React Native for our case — they made a clear argument for why, including the failure modes — and shipped both stores inside the window. Push notifications, offline mode, the device integration we needed. Live for fourteen months now and the crash-free session rate hasn't dropped below 99.6%.

Our AWS bill had grown organically for five years and nobody owned it. Stack Renew did a two-week audit, produced a phased remediation plan, and shipped the first two phases — workload right-sizing, log lifecycle policies, and a Terraform layer for the parts that mattered most. They didn't push Kubernetes where it didn't earn its keep, which is rarer than it should be. Run-rate spend down 34%.

The reason we work with Stack Renew is not the engineering, even though the engineering is excellent. The reason is that Michael will tell us when not to build something. Twice now they've talked us out of a project that I wanted and that they would have happily billed for. That posture is rare. It's the reason they're a multi-year vendor and not just last quarter's contractor.