The most expensive scoping mistake isn’t building slowly — it’s building the wrong size. Try to deliver the whole vision in a proof of concept and you run out of time with nothing working; deliver a broad set of half-features and you prove nothing. The winning move is a single vertical slice: the one workflow, end to end, on real data, that makes the customer believe — plus the discipline to fake or defer everything else. You have a two-week budget and a wishlist. Choose what goes in the POC, and watch whether it will actually prove value in the time you have. This is the scoping discipline from the discovery handbook, made playable.
a timebox + a wishlist · include the vertical slice that proves value · fake/defer the rest · fit the budget
vertical slice
One workflow built end to end on real data — the thing that proves value. The must-have of any POC.
timebox
A fixed deadline for the POC. The best scope-cutter there is: it forces decisions about what to leave out.
must vs nice
Must-have = required to prove value. Nice-to-have = everything else, to fake or defer.
defer
Explicitly moving a request to a visible “later” list, so the customer feels heard without it eating the budget.
scope.js — least you can build that makes them believe
Scope the POC
You have a 10-day budget for the proof of concept. Toggle features in and out. The verdict checks two things: does the scope include the vertical slice that actually proves value, and does it fit the timebox? Aim for the smallest scope that clears both.
0
days used / 10
no
has the slice?
—
verdict
How it works
The simulator is a tiny constrained-selection model, and it captures the core tension of scoping a proof of concept: value versus budget. Each candidate feature has a cost in days and a role. Exactly one is the vertical slice — the core workflow, built end to end on real data — and it is the only feature that actually proves the point of the whole engagement. Everything else is a nice-to-have: a second workflow, a polished UI, auth, edge-case handling, an admin dashboard. Now the deadline bites. The timebox (here, ten days) caps the total cost, and a good scope has to satisfy two conditions at once. First, it must include the vertical slice — a POC without the core proves nothing, no matter how much polish surrounds it, because you’ve built the frame of a painting with no painting in it. Second, it must fit the budget — overload the plan and you don’t get a broad POC, you get an unfinished one, which is worse than a narrow finished one because it demos as broken. The interesting insight the model surfaces is that the best scopes are usually lean, not full: the core slice plus at most one carefully chosen nice-to-have, leaving slack. Slack isn’t waste — it’s the buffer that absorbs the integration surprises and messy data that every real deployment throws at you, and it’s what lets you actually finish. So the optimal play is almost never “use every day”; it’s “prove the value with room to spare, and put everything else on a visible later-list.” That’s why an experienced FDE, handed a long wishlist and a short deadline, spends the first move deciding what not to build — because in a POC, subtraction is the skill.
1
One feature proves the value
Of everything on the wishlist, exactly one — the core workflow, end to end, on real data — is the vertical slice that proves the point. The rest are nice-to-haves.
2
The timebox caps the total
A fixed deadline caps how much you can build. It’s not a constraint to resent — it’s the forcing function that makes you choose.
3
Two conditions to clear
A scope proves value only if it includes the vertical slice AND fits the budget. Skip the core and you prove nothing; overload and you ship nothing.
✓
Lean beats full
The best scopes leave slack — the core plus at most one nice-to-have — because slack absorbs the integration surprises every real deployment brings. In a POC, deciding what NOT to build is the skill.
Must-have
the vertical slice
Skip the core
prove nothing
Overload
ship nothing
Best scope
lean, with slack
The code
# Scoping a POC is constrained selection
budget = 10 # days (the timebox)
features = [
{"name": "core workflow (real data)", "cost": 4, "is_slice": True},
{"name": "second workflow", "cost": 4},
{"name": "polished UI", "cost": 3},
# … more nice-to-haves
]
chosen = pick(features)
cost = sum(f["cost"] for f in chosen)
has_slice = any(f.get("is_slice") for f in chosen)
proves_value = has_slice and cost <= budget # both, or it fails# leaner (more slack) = more buffer for integration surprises
The picture: three scopes against the timebox
Each bar is one scope, at 50 px per day, with the deadline as the line at day 10. The lean scope is the only one that both contains the slice and leaves real slack. The polished scope fits but has nothing to prove. The greedy one has the slice but runs past the deadline, so on demo day the slice may be the part that is unfinished.
Worked example: every scope the lab will pass
The lab has six features (4 + 4 + 3 + 3 + 4 + 3 = 21 days) and a 10-day box, so there are 63 possible non-empty scopes. The rows below cover all 9 that prove value, plus three that fail. “Slack” is the 10 days minus the days used.
Scope
Days
Slice?
Slack
Lab verdict
Core only
4
yes
6
proves it, the sweet spot
Core + UI, or + auth, or + dashboard (3 scopes)
7
yes
3
proves it, warns about slack
Core + second workflow, or + edge cases (2 scopes)
8
yes
2
proves it, warns about slack
Core + any two 3-day items (3 scopes)
10
yes
0
proves it, no buffer at all
Core + second workflow + UI
11
yes
—
over by 1 day
UI + auth + dashboard
9
no
1
no value
The whole wishlist
21
yes
—
over by 11 days
Across all 63 scopes, 9 prove value, 22 fit the box but leave out the slice, and 32 run over. Only one reaches the lab’s sweet spot of four or more days of slack: the core on its own. Now say integration costs an unplanned 3 days, which is easy when access or data is messier than expected. The core alone still finishes with 3 days spare. Core plus a 3-day item finishes on day 10 exactly. Every 8-day and 10-day scope misses the demo. That is the page’s “lean beats full” in numbers.
Trade-offs: what each scope shape costs
Here Time is the days of the 10-day box the scope uses, and Space is the slack it leaves for surprises.
Case
Time
Space
Why
Whole wishlist
21 days
None: 11 days short
Unfinished on demo day, and an unfinished POC demos as broken.
Horizontal (UI + auth + dashboard)
9 days
1 day
Looks finished and proves nothing. The frame with no painting.
Vertical slice only
4 days
6 days
Proves the value with room to absorb surprises. Looks plain, so say why.
Slice + one nice-to-have
7–8 days
2–3 days
Worth it only if the extra removes the buyer’s biggest objection, such as “but who can log in?”
Slice with faked edges
Slice plus hours
Most of the box
Hard-coded or hand-run parts around a real core flow. Cheap and convincing, if you label what is faked.
The slice is not optional and slack is not waste. Everything else on the list is a bet on what the buyer will ask about, so spend days on it only when you know the answer.
Common mistakes
A slice on sample data
A clean export proves your code works on a clean export. Real data means the customer’s own, with its gaps and odd formats, and getting access can take days. Ask for access on day one, before you build anything.
No agreed success metric
“Proves value” needs a test both sides accept before the build starts, such as “drafts replies to last month’s tickets that agents accept without edits”. Without it, the demo turns into a debate about taste.
Faking without saying so
Faked parts are a fine tool. Hidden ones are not: the customer plans around something that does not exist, and trust goes when they find out. Label every fake and put the real version on the later-list.
Spending slack the moment it appears
Ahead on day 4, the team adds the dashboard. The integration surprise lands on day 8 with no buffer left. Keep slack until the slice is demo-ready, then decide.
Choosing the slice you find interesting
The slice should be the workflow where the buyer feels the most pain, not the one with the best engineering. Check it against what came out of discovery, not your own excitement.
Where this shows up in production and interviews
Paid pilots and POCs
Forward-deployed and solutions teams run timeboxed pilots with written exit criteria. The scope of that pilot decides whether a bigger contract follows, so this lab’s trade-off is often the first real decision of the engagement.
Startup MVPs with design partners
The same rule applies to a first product: one workflow that a real customer uses on real data beats a broad product nobody depends on. The later-list becomes the roadmap.
Internal platform work
A platform team’s first release should serve one real internal user end to end, not offer every feature to no one. Same slice, same slack, same later-list.
Interview angle
FDE and solutions interviews often give a wishlist and a deadline: “they want X, Y and Z in two weeks.” A strong answer names the slice, the success metric, what gets faked or deferred, and how the later-list is shown to the customer.
Quick check
1. You scope a POC with a polished UI, auth, and an admin dashboard — but not the core workflow. What happens?
A POC exists to prove value, and value lives in the core workflow done end to end. Surrounding it with polish, auth, and dashboards while omitting the slice is building the frame with no painting — impressive-looking and completely beside the point.
2. Why is a lean scope (core slice + slack) usually better than a full one that uses the whole budget?
Real deployments always hit surprises — messy data, an auth handshake that fights you, a network that blocks egress. A scope that uses every planned day has no buffer for them and ends up unfinished. Leaving slack is what lets you actually deliver the slice you promised.
3. What is the first thing an experienced FDE does with a long wishlist and a short deadline?
In a POC, subtraction is the skill. The move is to identify the one slice that proves value, cut or fake everything else, and put the deferred items on a visible list so the customer feels heard. A tight, finished slice beats a broad, unfinished plan every time.
FAQ
How do you scope a proof of concept?
Pick the single workflow that best shows value, cut it to the smallest end-to-end slice a user could try, and timebox it. Everything else — extra workflows, polish, auth, edge cases — goes on an explicit "later" list, not into the build. A good POC does one real thing convincingly, not ten in mock-up. The question: what’s the least I can build that makes them believe it’s real and valuable?
What is a vertical slice?
One workflow built all the way through — from real input, through the core transformation, to a valuable output — rather than building every layer horizontally across the feature set. A narrow slice that actually works end to end is far more convincing than a broad set of half-working features, because it proves the whole value chain is real. In a POC, the vertical slice is the must-have; everything else is optional.
Why is a timebox useful for scoping?
A hard deadline is the best scope-cutter there is: it converts an open-ended "what should we build?" into concrete decisions about what to leave out. Without a timebox, scope creeps and the build never finishes; with one, you rank features by value-per-day, keep the essentials, and defer the rest. Announcing a demo date early is a common FDE tactic precisely because it forces the cuts.
How do you handle a customer who wants everything in the POC?
Capture every request on a visible "later" list so they feel heard, but tie POC scope to the discovery success metric and judge each ask against it: does it help prove that value, or is it a separate phase? Then say no to the timing, not the idea: "great idea — to protect the demo date, let’s put it in the next phase." A finished POC that proves one thing beats an unfinished one attempting everything, and shown the trade-off, most customers agree.