The build-vs-buy math never breaks on ROI — it breaks on wallet identity, and $350 is the worst point on the curve
The question
"At what price does MAC's build-vs-buy math stop resolving toward buy for an IC data engineer expensing it personally vs. putting it on a team card?"
Surfaced as an open follow-up from [[2026-07-23-executable-agentic-info-product-pricing-scan]], which found MAC's decided $350 sits 75% above the highest verified self-serve installable comparable and proposed $200-250 as the evidenced landing zone. This brief tests whether the buyer's funding source changes that answer.
What we already know (from the vault)
- The $350 decision is explicitly built on a threshold claim that has never been evidenced. [[2026-05-14-mac-pricing-intent]] locks three positioning principles, the first of which reads: "At $350 the build-vs-buy comparison resolves toward buy quickly. At $1,500+ it would require deliberation and high-touch sales." Both numbers are founder intuition. Nothing in the vault cites an expense policy, a procurement threshold, or a buyer survey behind either.
- The ROI side of the build-vs-buy math is enormous and is not the binding constraint. The same doc puts a real principal data engineer at $200-300/hour and a comprehensive data-quality matrix at 8-40 hours per non-trivial model — a $1,600-$12,000 build cost. $350 is 3-22% of that. Even $1,500 clears. The founder himself records that "multiple thousands of dollars of value" exists but "we cannot charge that without a higher touch sales motion" — which is a statement about the sales motion, not about the value.
- The verified comparable ceiling is $200 individual / $199 team, and the team SKU is per-seat cheap. [[2026-07-23-executable-agentic-info-product-pricing-scan]] verified CLSkills Hub at $15 / $25 / $30 / $49 (Realtor, regular) / $200 Legal Pack / $199 Team Pack (5 people) — i.e. ~$40/seat for the team product, below the individual flagship. Agensi's installable-skill band is $5-$25 with a median listing under $50/month.
- There is no team, seat, or multi-user SKU anywhere in the MAC corpus. [[2026-05-04-mac-product-shape-decisions]] leaves per-install vs per-engagement open and never raises seats; [[2026-05-14-mac-prelaunch-readiness-checklist]] has no pricing-page tier item. [[2026-05-14-mac-pricing-intent]] explicitly forbids the surfaces a team sale usually needs ("No enterprise sales motion. No 'book a demo.' No 'talk to sales.'"). MAC is currently a one-SKU product aimed at one wallet.
- A single flat mid-price is the weakest ladder shape in the vault's own evidence. [[2026-05-18-low-ticket-dev-tool-launch-friction]] records Passion.io 2026 data: a 3-tier offer ($297/$497/$997) lifted revenue 41% vs a single flat $349 — the closest published number to MAC's exact current price, and it argues against the single-price design rather than the price level. (Caveat: coaching-market data, not developer tooling.)
What the web says
- Developers paying with their own money cluster at ~$20/month or free. The Pragmatic Engineer's April 2026 survey (900+ respondents) finds personal out-of-pocket spend on AI dev tools sits at "~$20/month or free tiers," that more respondents say their employer pays than say they pay themselves, and that only ~5% maintain separate work and personal subscriptions (Pragmatic Engineer). Paywall flagged: the article cuts off after section 4; the archetype breakdowns are subscriber-only and were not retrieved.
- Employer-funded budgets are 5-10x the personal number. Same survey: companies commonly fund max-tier plans at ~$100-200/month per engineer (Claude Code, Cursor, Codex), with budget-constrained shops capping at ~$20/month; one CPTO is cited running $600/month Cursor bills before moving the team to ~$100/month plans. A seed-stage CEO is quoted questioning £25/month per engineer — the low end of the same distribution.
- There is no universal documented corporate threshold for "can I just buy this." A direct search for expense-policy approval limits on software returned explicit non-findings: receipt thresholds around $25 are common, subscription approval is delegated to "manager vs director vs VP" by policy, and the aggregated guidance is that "there's no universal standard dollar limit — policies are customized" (Brex, Expensify, Navan). This is a real finding, not a failed search: the threshold the $350 decision assumes is not a documented artifact anyone publishes.
- The GTM literature's converging estimate for where cards stop working is ~$5k-$10k, not hundreds. Multiple 2026 PLG sources put the self-serve ceiling at ~$10K ARR per account ("below which credit cards work and a department head can expense it without procurement approval"), place the PLG/hybrid boundary at ~$5K ACV, and agree that five-figure commitments are what trigger security review, MSA, legal, and multi-stakeholder sign-off (GrowthSpree, GTM Playbook, CloudNuro). Source-quality caveat: these are vendor marketing blogs, not policy research. Treat the convergence as directional, not as a citation of anyone's actual policy.
- The category's canonical dev-education products ship a personal/team fork in the checkout itself. Epic React's buy flow opens with an explicit "Buying for myself or for my team" toggle and supports invoice customization for VAT/company details plus purchasing-power-parity pricing (epicreact.dev). Price did not render on fetch — no dollar figure from this source is claimed here. The structural point stands: the two wallets are treated as two products at the point of sale.
Convergences and contradictions
- Convergence, and it is the load-bearing one: the personal wallet and the team wallet are 5-10x apart, and every source finds the same gap independently. Pragmatic Engineer measures it directly ($20/mo personal vs $100-200/mo employer). CLSkills prices it structurally ($200 individual, $199 for five seats). Epic React forks the checkout on it. The PLG literature puts the next wall — actual procurement — three orders of magnitude above MAC at ~$5k-$10k. Nothing MAC could plausibly charge is anywhere near a procurement trigger.
- Contradiction with the vault's own reasoning: the $1,500 "deliberation" line in [[2026-05-14-mac-pricing-intent]] is attributed to the wrong mechanism. The doc implies $1,500 triggers deliberation because of the amount. The evidence says $1,500 on a team card triggers roughly nothing; $350 out of a personal pocket triggers real deliberation. The variable that moves is not dollars, it is whether the buyer has to ask someone. The founder's intuition about the shape of the curve is right; his placement of the axis is off.
- Contradiction the vault has not surfaced: MAC's second positioning principle argues against the very SKU that its price implies. "Sold INTO the internal data engineer, not OVER them" plus "no enterprise sales motion" is a design for the personal wallet. $350 is a price for the reimbursed wallet. The product is currently priced for one buyer and positioned for the other.
Synthesis for RDCO
The literal question has no single answer, because there are three price regimes and MAC's $350 sits in the seam between the first two. Regime one is the unreimbursed personal wallet: the engineer pays and never files anything. Regime two is the reimbursement regime — personal card then expense report, or a team card the engineer or their lead already holds. Regime three is procurement, and the evidence says it starts around $5k-$10k, which means MAC will never touch it at any price the founder would consider. So "avoid procurement" is not a real design constraint on MAC's pricing, and any argument for $350 that leans on it should be dropped.
Inside those regimes, the build-vs-buy math itself never stops resolving toward buy — the ROI is 5-35x at every price under discussion. What breaks is the buyer's authority to decide alone. My best reasoned estimate of the transition band, built from the $20/month personal anchor, the $200/$199 verified installable ceiling, and the absence of any personal-wallet comparable above $200: an IC data engineer buys unreimbursed and without hesitation at ≤$100; buys unreimbursed with real deliberation from $100 to roughly $200-250; and above roughly $250 flips to assuming the company should pay — which means asking. In the reimbursement regime the ceiling is far higher and much fuzzier: $1,000-$2,500 is where I would expect an expense report on a one-time software purchase to start attracting a conversation, and I want to be explicit that this range is inference from the $100-200/month employer-funded anchor ($1,200-$2,400/yr) and the ~$5k procurement floor, not a documented policy number. Nobody publishes the number this question asks for. A brief that hands over a precise crossover point here would be manufacturing it.
The actionable consequence is that $350 is the worst available point on the curve, and this is a stronger argument than the parent brief's. [[2026-07-23-executable-agentic-info-product-pricing-scan]] argued $350 is too high because no verified comparable clears above $200. That is a market-ceiling argument and it is contestable — MAC is more verticalized and better outcome-anchored than a Realtor skill pack, and being at the top of an observed curve is defensible. The wallet argument is different and harder to wave away: $350 is above the price an IC can decide alone, and below the price at which you would bother building anything for the person they'd have to ask. It buys the friction of the manager conversation without buying any of the upside of a team sale. It also puts that conversation directly at odds with the "level up, not replace" frame — the engineer now has to explain to their lead why they need purchased expertise, which is precisely the sentence positioning principle 2 exists to avoid.
The fix is a two-SKU split, not a price cut, and the category has already run this experiment. Put the single-seat SKU at the top of the personal-decision band — $199 is the defensible number: it lands in the parent brief's $200-250 evidenced zone, sits at the verified installable ceiling, clears the $15-49 prompt-pack cluster by 4-13x, and critically stays under the threshold where the buyer stops feeling entitled to decide. Then add a self-serve 5-seat team SKU in the $500-$900 range ($100-180/seat) aimed squarely at the reimbursement wallet, where the ROI story is strongest (8-40 hours × $200-300/hour × every model the team owns) and where the observed ceiling is 4-10x higher. This is not an enterprise motion and does not violate the "no book a demo" rule — CLSkills sells a five-person pack through the same checkout as its $15 cheat sheet, and Epic React resolves it with a two-word toggle. It also converts the revenue objection into a non-issue: $350 → $199 is only a 43% cut if the team buyers disappear, and the whole point is that they were never being sold to. Passion.io's 41% lift from laddering over a flat $349 is independent, if imperfect, support for the same move. The single caution worth stating plainly: I am recommending a structure with high confidence and specific numbers with moderate confidence, and the first 100 sales remain the only real test — which is exactly the revisit condition [[2026-05-14-mac-pricing-intent]] already wrote down.
Why this is in the vault
This resolves the last open pricing question standing between MAC and launch: it converts the $350-vs-$200-250 argument from a single-number dispute into a two-SKU design decision, and it directly triggers the "buyer use patterns reveal a clear higher-touch tier candidate" revisit condition in [[2026-05-14-mac-pricing-intent]] before launch rather than after. If accepted, it adds a pricing-page tier item to [[2026-05-14-mac-prelaunch-readiness-checklist]], which currently assumes one price and one buyer.
Open follow-ups
- What is the actual IRS de minimis safe-harbor capitalization threshold ($2,500/invoice without an applicable financial statement, $5,000 with one) and does it function in practice as the line where a software purchase stops being a routine expense? Known from memory, not verified this session — worth one primary-source check against the FAR/IRS text, because it is the only genuinely documented dollar line in this whole area.
- Does any published survey measure one-time (not subscription) out-of-pocket software or education spend by individual engineers? Every number found here is monthly-recurring, and MAC is a one-time purchase — the translation between the two is the weakest joint in this brief.
- What do CLSkills' $200 Legal Pack and $199 Team Pack actually sell in volume, and what is the individual-to-team unit mix? Carried forward unanswered from the parent brief; it is the single data point that would validate or kill the two-SKU recommendation.
- What fraction of developer-education purchases in the $200-600 band are employer-reimbursed vs personally absorbed? Sellers in this category (Epic React, Frontend Masters, ByteByteGo) hold this data via their invoice-customization flows and none of them publish it.
- Do companies with a named annual learning/L&D stipend per engineer represent a distinct, larger wallet than the software budget — and if so, should MAC's copy explicitly name that budget line? This is a fourth regime this brief did not investigate and may be the highest-ceiling one.
- Does the "ask my manager" friction show up as cart abandonment at a measurable price point? MAC's own launch could A/B $199 vs $350 on the single-seat SKU and read the answer directly, which would make RDCO the first public source on it.
Related
- [[2026-07-23-executable-agentic-info-product-pricing-scan]] — parent brief; this question was its final open follow-up
- [[2026-05-14-mac-pricing-intent]] — the $350 decision and the three positioning principles this brief tests
- [[2026-05-18-low-ticket-dev-tool-launch-friction]] — refund environment, tool-vs-course framing, Passion.io tier-ladder evidence
- [[2026-06-20-bookstores-for-agents-2026-live-comps]] — marketplace take-rates and the free-catalog competitive floor
- [[2026-05-04-mac-product-shape-decisions]] — per-install vs per-engagement question left open; no seat model
- [[2026-05-14-mac-prelaunch-readiness-checklist]] — P0 list that would need a pricing-tier item if the split is accepted
- [[2026-05-14-four-tier-buy-build-stack-soloproneur-tam-filter]] — tier-2 soloproneur-TAM frame the $350 was fitted to
- [[2026-05-23-agent-deployer-competitor-pricing-scan]] — adjacent pricing-model survey for the agent-deployer category
Sources
- Vault: [[2026-05-14-mac-pricing-intent]] (
~/rdco-vault/01-projects/mac/2026-05-14-mac-pricing-intent.md) - Vault: [[2026-07-23-executable-agentic-info-product-pricing-scan]] (
~/rdco-vault/06-reference/research/2026-07-23-executable-agentic-info-product-pricing-scan.md) - Vault: [[2026-05-18-low-ticket-dev-tool-launch-friction]] (
~/rdco-vault/06-reference/research/2026-05-18-low-ticket-dev-tool-launch-friction.md) - Vault: [[2026-06-20-bookstores-for-agents-2026-live-comps]] (
~/rdco-vault/06-reference/research/2026-06-20-bookstores-for-agents-2026-live-comps.md) - Vault: [[2026-05-04-mac-product-shape-decisions]] (
~/rdco-vault/01-projects/mac/2026-05-04-mac-product-shape-decisions.md) - Vault: [[2026-05-14-mac-prelaunch-readiness-checklist]] (
~/rdco-vault/01-projects/mac/2026-05-14-mac-prelaunch-readiness-checklist.md) - Vault: [[2026-05-14-four-tier-buy-build-stack-soloproneur-tam-filter]] (
~/rdco-vault/06-reference/concepts/2026-05-14-four-tier-buy-build-stack-soloproneur-tam-filter.md) - Vault: [[2026-05-23-agent-deployer-competitor-pricing-scan]] (
~/rdco-vault/06-reference/research/2026-05-23-agent-deployer-competitor-pricing-scan.md) - Web: The impact of AI on software engineers in 2026, Part 1 — The Pragmatic Engineer (900+ responses, April 2026; partially paywalled)
- Web: How to write an expense policy employees will follow — Brex
- Web: Corporate Card Policy Tips & Best Practices for 2026 — Expensify
- Web: Corporate Credit Card Policy: Guidelines and Best Practices — Navan
- Web: PLG for B2B SaaS 2026: Hybrid Self-Serve + Sales-Led — GrowthSpree (vendor marketing; directional only)
- Web: Bottom-Up vs. Top-Down SaaS Adoption — GTM Playbook (vendor marketing; directional only)
- Web: SaaS Procurement Has Changed: What Modern Buying Committees Look Like — CloudNuro (vendor marketing; directional only)
- Web: Epic React (personal/team checkout fork + invoice customization confirmed; prices did not render, no figure claimed)
- Web: What to Know About Pricing Developer Tools — Heavybit — fetched; contains no dollar thresholds or price points, entirely conceptual. Logged as a non-finding so it is not re-fetched.