The gap between agencies in this space is not mostly skill. It is whether anyone owns the outcome after launch. Building an automation that works on a Tuesday afternoon in a demo is not difficult. Building one that is still running in six months, after your CRM changed its API and your process changed twice, is a different job. These are the questions that tell you which one you are buying.
The nine questions
- What exactly will exist at the end, written down? If the deliverable cannot be described specifically enough to disagree about, it cannot be delivered.
- Fixed price or hourly? Fixed price puts the estimation risk on the party who understands the work. Hourly puts it on you.
- Who owns and can access what you build? Ask to see the configuration during the project, not after.
- What happens when an integration breaks? Not whether, when. APIs change. Ask who notices and how fast.
- What is the ongoing cost, in a range? Nobody can be exact. Everybody competent can bracket it.
- Can I run this myself if I want to? The answer should be yes, with documentation. Lock-in is a business model, not a technical necessity.
- What have you built that is closest to this? Ask for the one that went badly too. The answer tells you more than the case study.
- What would you talk me out of? An agency that has never declined a project is selling capacity, not judgement.
- Who is actually doing the work? On small engagements, find out whether the person on the call is the person building it.
Answers that should worry you
- "We'll figure out the scope as we go." Sometimes true for research. Never acceptable for a priced build.
- A percentage of the savings. The measurement is controlled by the party being paid.
- Vague about ownership. If they cannot answer who has access, assume you do not.
- Only glowing case studies. Every real practitioner has shipped something that underperformed.
- No mention of maintenance. Automations are not furniture. They need somebody watching them.
Agency, freelancer or build it yourself
| Option | Best when | The risk |
| Build it in-house with a no-code tool | The process is simple, internal, and low-stakes if it fails | It breaks quietly and nobody owns fixing it |
| Freelancer | One well-defined piece, and you can specify it yourself | No continuity, and the knowledge leaves with them |
| Agency | It touches customers or revenue and has to keep working | You pay for overhead you may not need on a small job |
The one thing that predicts the outcome
Ask what they will do in month three. Not what they will build in weeks one to six, which everyone can answer, but what the relationship looks like once the interesting part is done. The agencies whose systems are still working a year later are the ones that had a real answer to that question before you asked it.