The 90-Day Software Roadmap for a Mid-Size Manufacturer
What to fix in weeks 1–4, 5–8, and 9–12 if you want software that earns its keep by the end of the quarter. Prioritized by measured impact, not vendor pitch.
Short answer: weeks 1–4 go to measurement and plumbing fixes. Weeks 5–8 ship the first two automations that earn their keep. Weeks 9–12 are the first real build—something your team actually uses every day. Anyone promising a finished platform by day 90 is selling future pain.
Weeks 1–4 — measure, don't build
Before anyone writes code, map where hours actually go. A diagnostic ranks friction by measured cost, not by executive hunch. Quote follow-ups, re-keying, scheduling, vendor chasing—these nearly always top the list at mid-size manufacturers. CIRQL calls this diagnose-first, and the output is a ranked list of time sinks with rough dollars per year attached.
- Week 1 — roster the team, hand out a structured interview, collect workflow artifacts (emails, tickets, quote logs).
- Week 2 — rank friction by cost, not volume. Stack-rank the top 10 time sinks.
- Week 3 — clean the two or three data sources the top items depend on (usually your item master, customer table, and open-order table).
- Week 4 — lock the first two automations. Write down what “done” looks like so you can check the work later.
Weeks 5–8 — ship the boring middle
This is where most teams want to start, and this is where vendors want to land. Both are wrong. You need the measurement first so you can prove the ROI afterward. Once you have it, pick automations that remove hours from the same three to five roles you interviewed—not an enterprise-wide “AI platform.”
- Quote follow-ups and status requests (quoting)
- Inventory visibility and low-stock alerts (inventory)
- Re-keying between CRM and ERP
Weeks 9–12 — build the first real thing
By week 9, your team has one workflow they trust and one they still hate. The first real build is the one they still hate. This is the custom tool that replaces the spreadsheet they've been cursing at for two years—not a reimagined ERP.
Good candidates: a vendor-sourcing automation (procurement), an AI-ready catalog layer for buyers running procurement agents (AI-ready infrastructure), or a quoting UI that actually matches how your reps work.
What to measure at the end of 90 days
- Hours per week removed from the top-three roles you interviewed.
- Quote turnaround time, start to “sent.”
- Stale-data errors logged per week (should trend to zero).
- How many times a human had to override the new system (should trend to low single digits).
Next reads: how to assess AI readiness and how we run a retainer.