What big-company shipping taught me about startups

Between 2017 and 2020, I shipped eight MVPs inside Accenture and Aurecon. Mining, insurance, government, retail, infrastructure. Some were three-week sprints; some were six-month builds. They were all called MVPs in the same loose way most things in those years were called MVPs.

What I learned shipping inside a hundred-thousand-person company is not what people assume you learn there. The popular assumption is that big-company shipping is about politics. It is not. It is about adoption discipline. The politics question is whether you get the budget; the adoption question is whether anyone uses the thing you built. The startups that I see fail rarely fail on the politics question. They fail on the adoption question, the same way the bad big-company MVPs failed.

The four habits that transferred from that work to my pre-seed engagements were the ones nobody at the small-startup conferences talks about. They sound boring because they are. The boring is the point.

1. The operator hire matters more than the feature

The MVPs that survived inside Accenture all had one thing in common. Not the technology. Not the design. Not the budget. They had a specific person on the customer side, two levels below the sponsor, whose KPI depended on the rollout working.

The MVPs that died had a sponsor, a steering committee, an acceptance process, and no operator. They shipped cleanly into an operational void. Three months later, the spreadsheet workaround was still in use because no one had stayed to make the new thing stick.

This transferred directly. When a founder tells me their product is not getting traction, the question I ask first is not about the product. It is about who on the customer side has their reputation attached to the rollout. If the answer is "our champion loves it," that is a sponsor, not an operator. The product needs an operator, internal or external, before it needs more features.

2. Adoption follow-through is the unsexy moat

The cheapest competitive advantage available to a startup is doing the adoption work nobody else bothers to do. Not the build work. The work that comes after the build, when the customer is live but the workflow is still partially in their head and partially in the old tool.

At Accenture, the project teams that shipped the long-living MVPs were the ones that stayed for the first ninety days post-launch. They sat in the morning ops review. They ran a second training session for the new hire who joined two weeks after launch. They patched the report nobody had asked about until the operator missed it on the third Friday. None of this was glamorous. All of it compounded.

Stripe's documentation. Square's onboarding. Notion's import tools. Each was a multi-year follow-through investment that competitors avoided because the work was unsexy. The startup playbook usually treats follow-through as something the customer success team does after the engineering team has finished. The companies that win treat it as part of the build, owned by the same people who shipped the feature.

3. Decision logs survive reorgs

Big companies reorg constantly. Inside Accenture, the average tenure of a senior sponsor on a multi-year program was about fourteen months. Programs that survived the reorgs had decision logs. Programs that did not had a new sponsor in month fifteen who could not understand why the architecture was the way it was, why the licensing deal looked the way it did, or why two seemingly contradictory tools were both in production.

The new sponsor would either rebuild from first principles, which usually killed the program, or accept the architecture without understanding it, which usually killed it slower. The programs with decision logs survived because the new sponsor could read why each call had been made and either inherit the reasoning or argue against it. Both responses were better than the option without a log.

Startups reorg too, just with smaller tremors. The cofounder who leaves in year two, the head of growth who joins in year three, the board observer who arrives at series A. Each is a reorg of the institutional memory. Decision logs travel through these transitions in a way that the founder's head does not.

4. Kill criteria save quarters, not features

The big-company programs that wasted the most time were the ones that should have been killed at month four and were killed at month fourteen. Not because anyone was lying. Because nobody had agreed in advance what "this is not working" looked like. The signal arrived in month four. The agreement to act on it arrived ten months later, after twelve more all-hands updates that buried the signal under progress reports.

The teams I worked with that ran kill criteria — pre-committed thresholds, dates, and actions — wasted quarters less often. The criteria did not magically tell you the right answer. They told you when to ask the question.

Pre-seed startups lose more time to this than any other single cause. Founders who pre-commit kill criteria for a feature, a channel, or an initiative tend to wind down dead projects at month four. Founders who do not, on average, wind them down at month nine or month eleven. The five-month difference is the entire next bet.

What does not transfer

Three things from big-company shipping that I have to actively un-learn in pre-seed engagements.

• Stakeholder maps. Inside a large org, mapping the political graph is the work that decides who actually ships. At pre-seed, there are no stakeholders; there is the founder and the customer. Mapping is procrastination.
• Phased rollouts. At Accenture, a phased rollout to one site, then five, then fifty, was the right shape. At pre-seed, the phased rollout is usually the founder avoiding the conversation about whether the product is ready for any customer.
• RACI matrices. The R, A, C, and I in a ten-person company are all the founder. The matrix wastes a meeting that should have been a decision.

The companion templates

Three of the resources on the page connect directly to the four habits above. The first-hire role definition matrix is the artefact for the operator hire. The decision-log template is the artefact for the third habit. The kill-criteria template is the fourth. Each is a one-page worksheet that turns a habit into something a small team can run without re-deciding it from scratch every quarter.

Most of what makes big-company shipping work is small-team discipline written down. Most of what makes pre-seed shipping fail is small-team discipline that nobody bothered to write down.