Not a dev agency, but grew my (other kind of) agency from 1 to 25 FTEs in 2020.
- Build a personal brand on LinkedIn and Facebook. Connect and engage with folks in your target demographic
- Create content that adds value to the lives of your target demographic. Figure out if that's non-technical founders looking to build an MVP or engineering managers who need to scale.
- Do this for several years
- You need to ladder up as quick as possible to the highest value projects. This means every project you take on should lead to 2x bigger opportunities.
- Don't take on projects if the case study won't lead to 2x bigger opportunities. It's not worth the short term cash flow.
- Try and only do one thing. If you want to do do machine learning, don't take a website project for short term cashflow. A great outcome wont help you get more machine learning clients.
- Doing one thing will make marketing, sales, operations, and value delivery easier and more consistent. It's really hard to do a lot of things well. You probably won't be able too either.
- When you're calculating cost, don't forget PTO, maternity / paternity leave, cost of recruitment, hiring and retention.
- Document everything. Agencies are labor triage and well documented processes and frameworks enable junior level team members to do senior level work.
Most people I know who have started this, started doing it as side jobs. Those side jobs eventually became enough to be a full job. Followed of course by either hiring more people or going back to a part time job.
It would seem to me you need a sales person to actively find work if you want to grow beyond just you.
For finding customers, the best method I have seen is to teach. Be it at a conference, junior college, etc. People will see you as an expert and when they need something they will hopefully remember you.
Here are several replies that might be useful. We're a tiny boutique consultancy specialized in machine learning. Our clients are exclusively large organizations in diverse sectors and industries. Repeat business is common, as once you consistently deliver, they trust you and this compounds. You'll deal with the same people, things move very fast at some point.
Your experience might be different. You can read the threads where these replies were posted and get more insights.
One way to speed up finding clients is to go meta: do consulting for a consulting company that does something different but complementary/related. This company has clients, and you'll partner to solve problems for their clients. Instead of going after 40 clients, you go after one that has 40 clients.
The link https://news.ycombinator.com/item?id=25760622 addresses the case of NDAs between you (A) and your client (B), and between your client (A) and their client (C), so that you could work on C's problem.
Doing that many projects has lead us to figure out patterns and ramp up on the experience curve, and we then have abstracted several aspects into our "MLOps" platform, https://iko.ai, to be able to deliver faster. This corresponds to the "Abstract" phase in the Twitter thread. You can really do that beforehand (you can see it especially in the machine learning field where some startups with no experience and no prior projects want to improve the tooling and come up with pipelines and what they consider a good way to do things, working on the wrong problems or problems that don't exist. This is dangerous).
The first and the most important advice I can give is to find your thing and focus on it.
For example, say your company is the best in building react native mobile apps for the automotive industry.
Then build your portfolio and ask clients for references. Build a network with the decision-makers in the same industry (LI is your friend) and become known as an expert in the field.
Also, invest in content marketing, but do it genuinely by adding value in your writings. It will payoff.
- Build a personal brand on LinkedIn and Facebook. Connect and engage with folks in your target demographic
- Create content that adds value to the lives of your target demographic. Figure out if that's non-technical founders looking to build an MVP or engineering managers who need to scale.
- Do this for several years
- You need to ladder up as quick as possible to the highest value projects. This means every project you take on should lead to 2x bigger opportunities.
- Don't take on projects if the case study won't lead to 2x bigger opportunities. It's not worth the short term cash flow.
- Try and only do one thing. If you want to do do machine learning, don't take a website project for short term cashflow. A great outcome wont help you get more machine learning clients.
- Doing one thing will make marketing, sales, operations, and value delivery easier and more consistent. It's really hard to do a lot of things well. You probably won't be able too either.
- When you're calculating cost, don't forget PTO, maternity / paternity leave, cost of recruitment, hiring and retention.
- Document everything. Agencies are labor triage and well documented processes and frameworks enable junior level team members to do senior level work.