# Build, buy or partner

> Decide whether to do a job yourself, buy a product or hire a service, using core versus context, five-year total cost of ownership and a weighted matrix.

Source: https://docs.lumina-erp.com/first-and-third-parties/build-buy-or-partner/

**In short.** Build only what makes customers choose you, buy what is standard and well served by products and hire a service when the job needs skills or scale you cannot keep busy. Decide on five-year total cost of ownership including your own staff time, which can be a third or more of the total, rather than on the first invoice.

Every capability your business needs can be done by your own people and systems (first party), bought as a product or handed to an outside service or partner (third party). In short, build what makes customers choose you, buy what is standard and hire a service when the work needs skills or scale you cannot keep busy all year. The steps below cover the decision, with a five-year total cost of ownership comparison, the hidden costs that usually tip it and a weighted decision matrix you can reuse.

:::note[Disclosure]
Lumina ERP is a consultancy, so "hire a partner" is one of the options we sell. We have tried to keep this guide neutral, and in many of the cases it describes the right answer is to build or buy without us or anyone like us.
:::

## The options at a glance

| Option | What it means | Best when | Typical cost shape | Main risk |
|---|---|---|---|---|
| Build (first party) | Your staff write it, run it or do it | The work is core to how you win, or nobody sells a good fit | High internal time, low fees | Key-person dependence, slow delivery |
| Buy a product | License or subscribe to software and run it yourself | The need is common and products are mature | One-time setup, recurring fees, moderate internal time | Fit gaps, price increases, vendor failure |
| Hire a service or partner | An outside business does the work for you | Work is specialized, seasonal or needs scale | Recurring fees, low internal time | Less control, dependence, quality drift |
| Hybrid | Buy a product and pay someone to run or extend it | You want a product but lack the skills to run it | Fees on two contracts | Two parties to coordinate, finger-pointing |

## Before you start

- A one-paragraph description of the job to be done, in business terms, not a product name.
- A rough volume: transactions, users, sites, hours per month.
- A named business owner who will live with the result.
- A realistic loaded hourly rate for the internal staff involved (see step 4).

## Steps

1. Describe the job, not the solution. "Capture 3,000 supplier invoices a month into AP with three-way match" is a job. "Buy an AP automation tool" is a solution. Writing the job first keeps all three options open.

2. Decide whether it is core or context. Geoffrey Moore's core versus context idea (from his book *Dealing with Darwin*) is the most useful single filter. Core work differentiates you: customers choose you because of it. Context work must be done well but earns no preference when it is. For most distributors, pricing strategy, local inventory availability and technical sales are core. Payroll, tax calculation and invoice capture are context. Build or tightly control core work. Buy or outsource context work.

3. Score the options against the criteria. Use the criteria table below. Drop any option that fails a must-have outright.

4. Compare five-year total cost of ownership. Include internal time at a loaded rate as well as invoices. The worked example below shows how.

5. List the hidden costs and risks. Upgrades, key people, switching and data exit. They rarely appear on a quote.

6. Run the decision matrix. Weight the criteria before you score, so the weights cannot be bent to fit a favorite.

7. Pilot or reference check. For buy and partner options, talk to two customers of your size. For build, prototype the hardest part first.

8. Record the decision and a review date. Write down why you chose, what would change your mind and when you will look again (usually at contract renewal or the next ERP upgrade).

## Core or context: examples for a distributor

| Capability | Usually | Why | Exception |
|---|---|---|---|
| Customer-specific pricing logic | Core, build or configure in the ERP | It is how you compete on margin | Almost never. Keep control |
| Sales tax calculation | Context, buy | Rates and rules change constantly across thousands of jurisdictions | Single-state business with few exempt customers |
| EDI with large customers | Context, buy a service | Standards work and partner testing need specialists | Very high volume with in-house X12 skills |
| Warehouse operations | Core for most | Speed and accuracy are what customers notice | A 3PL is better for overflow, a new region or e-commerce spikes |
| AP invoice capture | Context, buy | Common problem, mature products | Very low volume where manual entry is cheaper |
| Payroll | Context, buy or outsource | Compliance-heavy, no competitive value | Rarely |
| Custom reporting | Mixed | Some reports are how you manage the business | Build the few that matter, buy a reporting tool for the rest |
| IT infrastructure | Context, partner or cloud | Hard to staff around the clock at small scale | Large IT team with spare capacity |

## Criteria to weigh

| Criterion | What to ask | Favors build | Favors buy | Favors partner |
|---|---|---|---|---|
| Fit to your process | How much of the job does it do without change? | Unusual process | Standard process | Standard process, specialist skill |
| Time to value | When does it start paying back? | Rarely fast | Weeks to months | Often fastest |
| Five-year cost | Total including internal time | Low fees, high time | Balanced | High fees, low time |
| Control | Can you change it when you need to? | Full | Limited to configuration | Limited to the contract |
| Skills required | Do you have them, and will you keep them? | You have and will keep them | Admin skills only | You do not have them |
| Upgrade path | What happens at the next ERP release? | You retest | Vendor certifies | Partner handles, at a price |
| Risk and resilience | What if the key person or vendor leaves? | Key-person risk | Vendor risk | Vendor and quality risk |
| Data and security | Where does your data go? | Stays with you | To the vendor | To the provider and its providers |
| Exit | How hard is it to stop? | Easy to stop, hard to replace | Data export and replacement | Transition and knowledge transfer |

## Total cost of ownership

Total cost of ownership (TCO) is everything you spend to get and keep a capability over its life, not only what you pay the vendor. The GAO Cost Estimating and Assessment Guide (GAO-20-195G, March 2020, retrieved 2026-09-28) makes the same point for government programs, asking for an estimate of the full life cycle rather than the purchase price.

**Total cost of ownership:** `One-time cost + sum of annual fees over the years + internal hours per year × loaded hourly rate × years`

Where:

- **One-time cost** is implementation, setup, data migration, training and any hardware.
- **Annual fees** are subscription, license maintenance, hosting or service fees, increased each year by the escalator in the contract.
- **Internal hours** are the time your own staff spend running, supporting, testing and managing the capability.
- **Loaded hourly rate** is what an hour of that staff time costs you, not what the person is paid.

### Getting the loaded rate right

A loaded rate adds benefits, payroll taxes and overhead to wages. The BLS Employer Costs for Employee Compensation release for June 2026 (retrieved 2026-09-28) puts private industry compensation at $46.89 per hour worked, of which wages and salaries were $32.82 (70.0%) and benefits $14.07 (30.0%). So benefits alone add about 43% on top of wages, before office space, equipment, software and management time. Many finance teams use a rule of thumb of 1.3 to 1.5 times wages for benefits plus overhead. It holds for salaried office staff and understates the cost of specialists you would have to hire.

### Worked example: AP invoice capture

These numbers are invented and round. A mid-sized industrial distributor processes about 3,000 supplier invoices a month and compares three ways to automate capture over five years. Internal time is costed at a loaded $85 per hour for all three.

| Input | Build (own scripts and OCR) | Buy (AP automation product) | Partner (outsourced AP service) |
|---|---|---|---|
| One-time cost | $60,000 | $25,000 | $5,000 |
| Year 1 annual fees | $3,000 (hosting) | $12,000 (subscription) | $30,000 (per-invoice fees) |
| Annual increase | 3% | 5% | 3% |
| Internal hours per year | 400 | 120 | 60 |

Year-by-year fees for the buy option, with the 5% escalator:

| Year | Subscription |
|---|---|
| 1 | $12,000.00 |
| 2 | $12,600.00 |
| 3 | $13,230.00 |
| 4 | $13,891.50 |
| 5 | $14,586.08 |
| Total | $66,307.58 |

For the buy option, five-year TCO is $25,000 one-time plus $66,307.58 in subscription fees, which is $91,307.58 paid to the vendor, plus 120 hours × $85 × 5 years = $51,000 of internal staff time. The total is **$142,307.58**, an average of $28,461.52 a year, and internal time is 35.8% of the total.

| Result over 5 years | Build | Buy | Partner |
|---|---|---|---|
| One-time cost | $60,000.00 | $25,000.00 | $5,000.00 |
| Annual fees, all years | $15,927.41 | $66,307.58 | $159,274.07 |
| Internal staff time | $170,000.00 | $51,000.00 | $25,500.00 |
| **Total cost of ownership** | **$245,927.41** | **$142,307.58** | **$189,774.07** |
| Average per year | $49,185.48 | $28,461.52 | $37,954.81 |
| Internal time as share of total | 69.1% | 35.8% | 13.4% |

Two lessons from the invented numbers. First, the build option looks cheapest on invoices (under $76,000 of cash to outsiders) and is the most expensive in total, because 400 hours a year of skilled staff time is real money. Second, even the buy option carries more than a third of its cost as internal time, which never appears on a vendor quote.

Try your own numbers. The calculator opens with the buy option from this example.

_Interactive calculator available on the web page._

## Hidden costs that change the answer

| Hidden cost | Hits hardest | How to estimate |
|---|---|---|
| Internal time to run it | Build, and buy with heavy admin | Ask the people who will do it for hours per week, then multiply by 52 |
| Upgrades and retesting | Build and custom integrations | Hours to retest at each ERP release, times releases in five years |
| Key-person risk | Build | Cost to hire and train a replacement, times the chance they leave in five years |
| Price escalators | Buy and partner | Read the renewal clause. A 7% escalator doubles the fee in about 10 years |
| Switching and exit | Buy and partner | Data export, replacement setup, parallel running, staff retraining |
| Vendor management | Buy and partner | Hours a year for reviews, invoices, disputes and renewals |
| Integration upkeep | All three | Connectors break when either side changes |
| Opportunity cost | Build | What your best people are not doing while they build this |
| Quality drift | Partner | Error rates and rework once the provider's A team moves on |

:::caution[Watch the escalator]
Check whether the annual increase is capped, tied to an index or at the vendor's discretion. An uncapped escalator makes every five-year TCO comparison a guess. Price protection belongs in the contract, not in a sales email.
:::

## The decision matrix

Weight each criterion before scoring, so the weights total 100. Score each option from 1 (poor) to 5 (strong) on each criterion. Multiply and add. The invented scores below are for the AP example.

| Criterion | Weight | Build | Buy | Partner |
|---|---|---|---|---|
| Fit to process | 30 | 5 | 4 | 3 |
| Five-year cost | 25 | 2 | 4 | 3 |
| Risk and resilience | 20 | 2 | 4 | 3 |
| Time to value | 15 | 1 | 4 | 5 |
| Control | 10 | 5 | 3 | 2 |
| **Weighted score (out of 5)** | 100 | **3.05** | **3.90** | **3.20** |

**Weighted score:** `Sum of (weight × score) ÷ 100`

For the buy option: (30 × 4 + 25 × 4 + 20 × 4 + 15 × 4 + 10 × 3) ÷ 100 = 390 ÷ 100 = 3.90. The matrix supports the TCO result here. When the two disagree, look hard at the weights. If control carries 10% but everyone in the room is arguing about control, the weights are wrong.

## Questions to ask

For a build:

- Who will maintain this in three years, and what if that person leaves?
- How many hours will each ERP upgrade cost us to retest it?
- Is this core, or do we enjoy building it?

For a product:

- How many customers of our size and ERP run it, and can we talk to two?
- Which ERP versions is it certified on, and how quickly after a new release?
- What is the renewal escalator, and is it capped?
- How do we export all our data, in what format and at what cost?

For a service or partner:

- Who exactly will do the work, and what happens when they leave?
- What service levels are in the contract, and what are the remedies?
- Which other businesses (fourth parties) will touch our data?
- What does transition back to us, or to another provider, look like?

## Red flags

- A build case that counts no internal hours.
- A buy case that counts only year-one subscription.
- A partner quote with no service levels or exit terms.
- "We will customize it to fit" as the answer to a core fit gap in a product.
- A decision that depends entirely on one enthusiastic employee.
- Weights set after the scores are known.

## Next steps

- [ ] **Write the job in one paragraph.** Business outcome and volume, no product names.
- [ ] **Label it core or context.** If you cannot agree, it is probably context.
- [ ] **Agree the loaded rate with finance.** One rate for the comparison, stated on the page.
- [ ] **Build a five-year TCO for each option.** Use the calculator and keep the inputs.
- [ ] **Weight, then score.** Weights signed off before anyone scores.
- [ ] **Record the decision and a review date.** Tie the review to renewal or the next ERP upgrade.

Once you have decided to buy or partner, [choosing a vendor](/first-and-third-parties/choosing-a-vendor/) covers the selection, and [contracts, SLAs and data rights](/first-and-third-parties/contracts-slas-and-data-rights/) covers what to sign. For ERP extensions specifically, see [ERP add-ons and ISVs](/first-and-third-parties/erp-add-ons-and-isvs/).

## Sources

- [Employer Costs for Employee Compensation, June 2026 (U.S. Bureau of Labor Statistics)](https://www.bls.gov/news.release/ecec.nr0.htm)
- [Cost Estimating and Assessment Guide, GAO-20-195G (U.S. Government Accountability Office, March 2020)](https://www.gao.gov/products/gao-20-195g)

---

Epicor, Prophet 21, P21 and DynaChange are trademarks or registered trademarks of Epicor Software Corporation registered in the United States and other countries. Kinetic is a trademark of Epicor Software Corporation. Lumina ERP is an independent consultancy and is not affiliated with, sponsored by or endorsed by Epicor.