Field notes on Meta advertising, lead qualification, CRM automation, and revenue systems for service businesses. What we tried, what worked, and what the numbers actually showed.
It's always with you.
The most common mistake in service businesses is the framing. You're serving the client. They're the boss. You execute what they ask, hope they're happy, repeat. That's not service. That's order-taking.
Read →And neither are you.
Apple makes the most expensive consumer products in the world. And they're not a product company. Every product they ship exists for one reason: to deliver the best possible experience to the end user. The product is the vehicle. The experience is the business.
Read →And why that's the whole business model.
We turn down more clients than we take on. That's not a capacity problem. It's a deliberate choice. Here is what happens when an agency says yes to everyone, and why we chose the other path.
Read →How a single integration dropped CPL from $3.50 to $1.13.
You run a Meta ad. Someone fills the form. You get a name and number. Your salesperson calls. Maybe they convert. Maybe they don't. Meta never finds out which. That's the gap. It has a fix.
Read →Before a single phone call was made.
Most lead generation problems aren't lead volume problems. They're lead quality problems. A business runs Meta ads, gets 50 enquiries a week, and the sales team spends three days chasing numbers that don't pick up. Here's the qualification layer that fixed it.
Read →DMs picked up in 48 hours. Zero spend.
Open your business Instagram right now. Pretend you've never heard of what you do. Can a new visitor tell, within ten seconds, what your service is, what it costs, and how to get started? If the answer is no, you have an offer visibility problem. No amount of paid advertising will fix it.
Read →The trial isn't a giveaway. It's a closing mechanism.
There's a conventional tension in subscription marketing: free trials attract the wrong people. The logic goes: people who want something free don't intend to pay. This logic is wrong for a specific category of business. Here's why.
Read →Here's the diagnosis, the rebuild, and the $1.28 comeback.
In August 2025, Meta cited 'forms collecting data' as a policy violation on a lead generation ad, on the same platform whose entire lead product exists to collect data. CPL climbed from $2 to $9. This is exactly what we did.
Read →We didn't send a report. We sent a rebuild plan and a promise.
There's a version of agency-client communication most businesses have experienced: monthly reports, benchmarked metrics, a slide that says 'CPL increased due to market conditions,' and a recommended next step that requires more budget. This is not transparency. This is documentation with the hard parts softened.
Read →What to say to a client when the results have been bad.
Every agency relationship eventually reaches a hard moment. Performance drops. The platform changes. A campaign that worked stops working. What happens next, and what the agency says next, determines whether the engagement survives.
Read →And the three-node loop that makes it happen.
Most Meta campaigns get more expensive over time, not less. The algorithm optimises toward what it measures. If it only measures form fills, it finds people who fill forms. If it measures actual subscribers, it finds actual subscribers. The difference in outcome is not marginal. It is the difference between $3.50 and $1.13.
Read →What 'Menu' and 'Trial' actually do inside ManyChat.
Most businesses respond to Instagram comments manually, hours later, after the intent has cooled. A keyword trigger on a Meta ad comment opens an automation instantly, in the same app, at the moment of maximum interest. Here's what happened when we built two of them for a meal delivery subscription.
Read →The $1.28 CPL was not one winning audience. It was fifteen right ones.
The most common mistake in Meta campaign setup is not the creative and not the copy. It is launching against one audience and calling it targeting. Here is what real audience architecture looks like, and what it produced on a campaign relaunch day.
Read →Your pipeline is not a database. It is a diagnostic.
Most businesses have a CRM. Most of them use it as a filing system: a place to store names and numbers. The pipeline stages exist, but they are never read diagnostically. Here is how four stage names across 1,942 leads surfaced revenue that would otherwise have stayed invisible.
Read →60-70% of paying subscribers were vegetarian. The brand had no idea.
Most brands think they know their customer. They have an intuition built from product decisions, early buyers, and the aesthetic they chose for the Instagram grid. Then a lead generation system starts producing real data. The gap between the intuition and the data is where the real customer lives.
Read →Native Meta forms vs. off-platform redirects: what the data showed.
The conventional advice for service businesses is to build a landing page. More copy, more trust signals, more conversion opportunity. For a specific category of offer and audience temperature, this advice loses you leads before you even capture them.
Read →When demand exceeds operational capacity, that is not a problem. That is the goal.
Most businesses measure campaign success in cost per lead, ROAS, or subscriber count. Here is a different measure: the founder called on October 19th to say the kitchen could not keep up with the demand. The ads had to pause. That is not a campaign failure. That is the correct sequence for a product business.
Read →Why a creative bank is insurance, not overhead.
Most businesses build one ad, launch it, and optimise. When it dies, they build another. The cycle is reactive. A creative bank reverses the sequence: you build volume before you need it, so when a campaign stalls or Meta rejects an asset, you are not starting from zero.
Read →Aesthetic and offer are two different things. Most food businesses only have one.
Aesthetic content says: we are real and our product looks good. Offer content says: here is what you get, here is what it costs, here is how to start. Both are necessary. They are not the same thing. Most food and service businesses have built the first and have almost none of the second.
Read →Most businesses discard bad leads. Here is what to do with them instead.
A junk lead feels like a cost: time spent calling a number that doesn't pick up, a postal code outside the delivery zone, an email that bounces. Most businesses mark it and move on. In a system connected to the Conversion API, a junk lead is a signal. And signals have value.
Read →Not a campaign spike. A system compounding.
A campaign that spikes and plateaus is a media buy. A system that compounds is a business asset. The difference shows up in the growth curve: one produces a peak followed by regression to baseline. The other produces a floor that rises each month. Here is what the second one looks like in practice.
Read →ROAS counts the first transaction. A subscription business lives on what happens after.
ROAS is a snapshot. It counts what the ad spend produced in the measurement window. For a subscription business, the measurement window is the wrong frame. The subscriber who joined in July is still paying in November. The subscriber who joined in July referred their colleague in September. ROAS does not count either of those.
Read →And why that made the first six months structurally cheaper than they will ever be again.
Every market reaches a state where the cost of paid acquisition reflects the competition for audience attention. The businesses that enter before that saturation point arrive pay structurally lower CPLs, build cleaner audience models, and establish themselves in the algorithm before any competitor can crowd them out. That window does not stay open.
Read →