Most founders lose 4-6 weeks and £15-40K to the same hiring mistake. They sign the SOW before scope, comms and process are locked. Six weeks in, they find the agency built the wrong thing. We have seen the same pattern across 200+ engagements since 2019 and 47 technical due-diligences we have run on other teams’ code. The cause is not agency skill. It is the 6 decisions you did not lock in writing before the deposit cleared.
The pattern is predictable. Scope stays vague. Comms drift to email. Process stays “we’ll figure it out.” By week 4, you are paying engineering hours for features nobody agreed matter. Two 2025 diligences landed on the same story: £180K spent on a working prototype that missed the actual customer job. Below we break down where the mistake enters the room, and how Empyreal locks the fix in during the first 5 days of the audit.
Vague Scope and Undefined Requirements
The failure starts with a brief that could describe any product. “We want a marketplace for X.” “We need an app that talks to our API.” “It should feel like Stripe, but for Y.” A brief that loose gives 4 engineers 4 different mental models. All 4 will build something. None will build what you actually needed. Warning signs of a scope that will hurt you: no user stories, no acceptance criteria, no named non-goals, no ranked priorities.
If your brief cannot answer, “what would we cut if we lost 2 weeks?”, you do not have a scope. You have a wish list. A wish list becomes a change-request queue by week 3, and that queue becomes the actual invoice by week 8.
We see this most on AI features in 2026. Founders describe “an AI copilot” without defining the data boundary, the decision boundary, or the escalation path when the model is not sure. 4 engineers each interpret that differently. 4 different demos land in front of you at week 6. None is the one you can ship to real customers.
The fix is unglamorous. Write down the top 5 user journeys. Write down what happens when each one fails. Write down what “done” looks like for each. If you cannot draft that in a Google Doc in a morning, book the £8K audit before the build. The 5 days you spend on the audit save 4 weeks on the build.
Communication Gaps That Cost Weeks
The second failure is quieter. Scope was fine on paper. Comms drifts. Slack replaces daily standups after week 2. The weekly demo becomes a screen-share of last week’s slide. The lead engineer takes 3 days to reply because “they’re heads-down.” By the end of month 1, you have no calibrated view of what is landing, what is stuck, and where the risk sits.
Typical pitfalls: no shared source-of-truth doc, no written async standup, no named on-call for questions inside 4 hours. Meetings turn into vibes. Vibes turn into surprises. Surprises turn into rework. Two 2025 audits found the same shape: the client thought week-4 sprint was on track because a standup slide said “on track.” The actual PR queue told a different story that nobody had shown them.
Time-zone drift makes it worse. If your agency is 5.5 hours ahead and their comms window is 8am-11am your time, you get 1 window a day for a real conversation. Miss it, lose the day. Multiply by 40 working days and you have lost a full sprint of context.
The fix is boring and written. Daily async standup in Slack: what shipped, what is blocked, what is next. Weekly 45-minute demo plus a 1-page recap in the same channel by end of day. Named on-call engineer with a 4-hour inbox SLA during working overlap. If your agency will not commit to those 3 in writing, that is your signal.
Process & Cultural Misalignment
The third failure is the one nobody sees coming. The scope was written, the comms were fine, but the process is out of phase with yours. Your team runs 2-week Scrum sprints with story-point estimation. The agency runs a rolling Kanban with no estimation at all. Your team lives in Linear. The agency lives in Jira. Your engineers merge to main behind a feature flag. The agency merges to a long-lived develop branch that gets rebased once a fortnight.
The costs are hidden. Reviews take longer than the writes because the mental models differ. Standups get scheduled around 2 different rhythms. Nobody agrees what “done” means. One team gets to “done” at merged-to-main; the other at deployed-to-staging; the other at signed-off by the founder. When 3 different definitions of done meet the same sprint boundary, the sprint boundary loses.
Culture drifts too. Your team writes ADRs for every architecturally consequential decision. The agency writes a JIRA ticket after the fact. Your team pushes back on scope creep in the demo. The agency accepts every founder request without flagging trade-offs. Both patterns are legitimate. They are not compatible on the same repo.
Common misalignment red flags to check for before signing: the agency skips your ceremonies; you are using different project-management tools; you have different definitions of “done”; the agency’s PR reviews are done by a lead you never meet. Every one of those is fixable if you name it in week 1. None is fixable if you discover it in week 6.
How Empyreal Prevents It
Empyreal Infotech was built to close the 3 gaps above at the very start of an engagement. Not with a slide deck. With a fixed £8K, 5-day audit that ends before the build begins. The audit is 2 senior engineers plus Mohit Ramani in your room for a working week. You leave day 5 with a 30-page brief, 6 ADRs on the most consequential technical calls, a 90-day plan with milestones, a risk register, and a single fixed quote for the build. If our quote is too high, you take the brief to another studio. You are not committed to the build. Roughly 20% of 2025 audit weeks did not turn into builds. We are fine with that.
Comms is locked in the same 5 days. Named on-call. Named daily standup. Named weekly demo. Named written recap in the shared channel. Every ADR that gets written during the build lands in your GitHub repo the same day, timestamped and signed. You inherit the reasoning behind every meaningful call, not just the code. If the engineer who wrote the ADR moves on, the next engineer picks up the file and knows why it was decided that way.
Process is aligned before the first commit. We match your ceremonies. If you run 2-week Scrum, so do we. If you run rolling Kanban, so do we. If you use Linear, we use Linear. If you use GitHub, we use GitHub. If you have your own definition of “done,” we adopt it and write it into the SOW. This costs us nothing and saves us the 2-week rediscovery that every misaligned pair goes through in month 2.
The commercial framing keeps the incentive honest. 30-day walk-away, both ways. Milestone billing at 25/25/25/25 so you are never more than a quarter ahead of working software. IP assigns on commit, so your code lives in your GitHub org from day one. No lock-in, no exit fee, no surprise renewals.
The Empyreal Playbook, Three Moves
- Discovery week that ends in a fixed quote. Every engagement starts with a 5-day, £8K audit that closes with a 30-page brief, 6 ADRs, a 90-day plan, and a single fixed price for the build. You can walk away on Monday morning with everything.
- Written comms, named on-call, one channel. Daily async standup, weekly written demo recap, named on-call engineer with a 4-hour SLA during working overlap. All 3 land in one Slack channel so nothing gets lost in the tab-forest.
- Same tools, same ceremonies, same definitions. We adopt your project management, your sprint cadence, and your definition of “done.” No parallel process, no rediscovery in month 2. Your engineers can reach into the repo and know exactly what shape it is in.
Key Takeaways and Best Practices
The number-one hiring mistake is not choosing the wrong agency. It is signing the SOW before the 3 things that decide the outcome are locked in writing. Scope, comms, process. If any of the 3 is soft on day 1, week 6 will cost you 4 weeks and £15-40K. Here is what to do before you sign.
- Define everything up front. Top 5 user journeys, acceptance criteria per journey, named non-goals, ranked priorities. If you cannot write it, book a fixed-price discovery week to write it with you.
- Lock the comms cadence in writing. Daily async standup, weekly demo, named on-call with 4-hour SLA. Put all 3 in the SOW. If the agency will not commit, that is your signal to walk.
- Align processes on day 1. Agree the sprint cadence, the project tool, the definition of “done,” the branching model, and the PR review chain in week 1 of the SOW. Not in month 2 when you have discovered the drift.
- Pick a partner with the guardrails already installed. Look for a fixed-price audit before the build, IP-on-commit, and a 30-day walk-away in both directions. Empyreal built the audit-first shape for exactly this failure mode.
By locking scope, comms and process in the first 5 days, and by choosing an agency that installs the guardrails as standard, founders skip the £15-40K mistake this post opened with. The fix is unglamorous. It is also the single highest-leverage decision you make in the whole engagement.