Reality check: adoption friction in construction is structural

• One of the least digitized sectors - McKinsey has repeatedly ranked engineering & construction among the world’s least digitized industries, citing fragmentation and short, project-based incentives that punish behavior change. 

• Integration gaps force workarounds - Over half of firms still manually transfer data between non-integrated apps; only ~5% report that all their apps integrate. So new systems often add steps instead of removing them. 

• The cost of fragmentation is massive - FMI + PlanGrid found teams spend ~35% of their week on non-optimal tasks (hunting data, fixing mistakes), contributing to an estimated $177B/year in U.S. overspend. 

For example, let's consider portals like Procore that claim to simplify and standardize subs' workflows to share their estimates and that claim to ultimately better bid leveling processes.

“Reimagined” tools vs. how subs actually work

If you’ve built a shiny bid portal and wondered why subcontractors ignore it, you’re not alone.

• Email & spreadsheets still dominate tendering - Autodesk’s BuildingConnected report for the UK & Ireland found email is the most popular tool for inviting and submitting bids—across owners, main contractors, and subs—followed by Google Drive and Excel. Different region, same global behavior: people default to what’s already in their hands. 

• Multi-app reality, low integration - Even as companies use 4+ apps per project, over half are still doing manual double-entry. Without deep integrations, portals become another island—so users revert to Outlook and Excel. 

• Few truly integrated stacks - KPMG’s global survey: only ~16% of E&C executives say their orgs have fully integrated systems and tools. That’s the ocean your product has to swim in. 

Bottom line: Forcing subs into your prescribed workflow (account creation, custom forms, strict templates) often underperforms “messy but native” channels (email threads, CSVs, spreadsheet attachments).

Case study in “forcing the new”: Katerra

Katerra didn’t fail because it lacked world-class tech. It tried to replace entrenched, local, sub-driven workflows with a vertically-integrated, factory model. When market realities (project variability, local codes, financing, supply chain) didn’t bend, the company collapsed into Chapter 11. Analyses call out misalignment with industry workflows—alongside financing shock from partner failures. Different domain than software, same lesson: steamrollers struggle here. 

Even “no-brainer” tech takes a decade

BIM has obvious value—and still took years to tip. The NBS National BIM Report shows UK “aware & using” grew from 13% (2011) to ~73% (2020). That’s a decade to mainstream after government nudges and clear ROI. If BIM needed that ramp, your brand-new app won’t be adopted “because the UI is great.” 

Why Gen-AI makes the adoption trap more tempting

Gen-AI makes it seductively easy to “reimagine the workflow”: new portals, new forms, new behaviors. But the same adoption physics apply. If your AI requires users to leave Outlook, abandon Excel, or learn a new process map, you’ll eat churn.

Meanwhile, the biggest gains still come from eliminating fragmentation and rework—the exact problems caused when tools don’t respect existing behavior. 

What actually gets adopted: a product playbook

1) Start with the inbox and the spreadsheet.

• Auto-ingest bid invites and RFIs from email, parse attachments, and create structured records without making subs log into anything new.

• Treat CSV/XLSX as first-class citizens: import/export flawlessly, preserve their formatting and formulas, and never block copy-paste.

  Why: It matches documented industry behavior (email + Excel), so you remove friction instead of adding it. 

2) Integrate deeply, not broadly.

• Prioritize systems of record (ERPs like Sage/Penta/Vista, PM like Procore, DMS like SharePoint/Drive) and kill double entry.

• Set a bar: if a workflow ends with “retype it elsewhere,” you haven’t shipped it.

  Why: Most firms still stitch together 4+ apps and manually transfer data—fixing that is immediate ROI. 

3) Deliver value without a process change.

• Drop in AI that reads plans/specs, normalizes scope, or levels bids in the background, then outputs to the formats teams already use (PDF packages, XLS bid tabs, Procore submittals).

  Why: You’re augmenting today’s workflow rather than replacing it—crucial in a low-digitization, low-budget sector. 

4) Measure success in eliminated steps, not new ones.

• Track time removed from “non-optimal activities” (searching for info, reconciling conflicts, rework). Tie your claims to hard numbers that echo FMI’s benchmarks.

  Why: That’s where the money is—35% of a typical week. 

5) Earn trust with auditability.

• Every AI action should be explainable: show source docs, markups, and lineage so estimators can spot-check quickly.

  Why: In a risk-averse, multi-party environment, black boxes don’t fly. (And if it adds steps, users will abandon it.)

Put the steamroller away

Construction hasn’t refused technology; it has refused friction. The data says the sector is under-digitized, under-integrated, and allergic to extra keystrokes for marginal gains. Email and Excel aren’t bugs to be eradicated; they’re interfaces you must respect. Tools that adopt the user’s workflow—not the other way around—win the only adoption curve that matters: the one on real jobs, under real deadlines.