The Real Math on Whether to Build or Buy Your Sales Stack.
A back-of-envelope model with its assumptions showing. Buying wins for small teams by a wide margin, building only catches up past roughly 25 to 55 reps, and the number that decides it is the one most teams leave out: maintenance.
Every sales team eventually has the thought: we could just build this. A few engineers, a few months, and we'd own the thing instead of renting it. Sometimes that's right. More often it's a comparison between a price you can see and a cost you can't.
This issue builds the comparison out loud. It is a model, not a case study. The inputs come from a government wage figure and third-party price lists, the structure is mine, and every output below is my arithmetic. Change the assumptions and the answer moves, which is the point.
What Building Actually Costs
Start with the engineer. The US Bureau of Labor Statistics puts median software developer pay at $135,980 (May 2025). Wages are not the whole cost, though. BLS data for June 2026 show benefits at about 30% of total compensation for professional and related workers, so wages are about 69.2%. Dividing the wage by 0.692 gives a loaded cost of roughly $196,500 a year. That's my arithmetic, and it excludes equipment, management time, and recruiting.
Now the assumptions. I'll assume one developer for six months to reach a usable first version, which is about $98,000. Then comes the line people skip. Software needs upkeep: bug fixes, integrations that change under you, security patches, requests from the people using it. One engineering publication calls 15% to 25% of the build cost per year a common rule of thumb. It gives no source, so treat it as an assumption. I've applied it to one developer's loaded annual cost, which is $29,500 to $49,000 a year.
What Buying Costs
CRM list prices move and vary by edition, so I used third-party summaries of Salesforce's published tiers as a rough range: about $100 per user per month for a mid-tier suite and about $175 for the enterprise tier. Those are list prices from secondary sources, and real contracts are negotiated. One advisory vendor reports annual price escalators of 8% to 10%, which I have not verified and left out of the model. Including them would make buying look worse, so the model leans slightly in favor of building on that point.
| Sales reps | Buy at $100 | Buy at $175 | Build (15% upkeep) | Build (25% upkeep) |
|---|---|---|---|---|
| 5 | $30,000 | $52,500 | $245,600 | $343,900 |
| 10 | $60,000 | $105,000 | $245,600 | $343,900 |
| 20 | $120,000 | $210,000 | $245,600 | $343,900 |
| 40 | $240,000 | $420,000 | $245,600 | $343,900 |
Building is a bet that your needs will stay unusual and your engineer will stay put. Buying is a bet that the vendor's defaults are close enough.
Where the Lines Cross
Set the five-year costs equal and solve for team size. At $100 per seat, building breaks even at about 41 reps with light upkeep and 57 with heavy upkeep. At $175 per seat it's about 23 and 33. So for a team of five to ten, buying costs a fraction of building under any assumption in the table. For a team of forty paying enterprise rates, building is competitive and may even win.
Two cautions keep that from becoming a rule. First, the model prices only a CRM-shaped hole. A tailored workflow that no product offers has value that the table can't show. Second, it assumes the build works. A project that overruns, or an engineer who leaves, turns the cheap column into the expensive one. The honest read is that the model tells you when buying is obviously right, and is much less decisive about when building is.
The Questions the Model Can't Answer
Is this a differentiator? If how you sell is part of why customers choose you, owning that process may be worth more than the savings. If it's a contact database, it isn't.
Who's on call? A tool your reps use daily will break at inconvenient times. Someone has to own that, and the same engineer is probably also needed on your product.
Can you start by buying? Most teams can rent a standard tool now, learn what they actually need, and build the unusual part later. Buying first costs a few thousand dollars and gives you a requirements document made of real use.
Where These Numbers Come From
The wage and benefit figures are US government statistics. The prices and the maintenance rule are third-party and unaudited, and the model is mine. It is a thinking aid, not a quote. Figures checked November 2026.
- US Bureau of Labor Statistics: Software developers. Median pay, May 2025.
- US Bureau of Labor Statistics: Employer Costs for Employee Compensation. Benefits share, June 2026.
- Third-party Salesforce price summaries and an engineering blog's 15% to 25% maintenance rule of thumb. Both unsourced; used only as assumptions.
Price Your Own Version of the Table
Replace my inputs with yours: your real quote per seat, your real rep count in two years, and what an engineer actually costs you. Add one honest line for upkeep. If buying wins by more than 2×, stop. If it's close, list the one thing a custom build would do that no product does, and ask whether it's worth a developer's year.
