How to set marketing OKRs for an early-stage startup

Startup marketing targets should be one thing: the company's number for the quarter, split until one marketer can act on a piece this week.

The founders I work with who miss a quarter rarely miss on an ambitious target. They miss on arithmetic.

The marketing number, the sales number and the board's growth number were each set alone. Nobody notices until the end, when the three describe different quarters.

Chain them to one number and a miss shows which link broke.

The numbers get set in separate rooms. Marketing is handed one target, sales another, the board a growth figure on top. All three can pass the planning meeting and still fail to add up to one quarter.

A marketing target is the company number, broken down

The starting point is the one number the company is judged on that quarter. Every marketing target is that number, broken down. That is how Avishai Ben-Tovim, CEO and co-founder of MDI Health, runs the whole company. It is the order I keep coming back to.

OKRs are the company's quarterly goal, written as the numbers that prove you hit it.

The targets, he says, are "everything derived from what we're trying to achieve as a company, that quarter, that year." That is the cascade in one line, and I start every plan there.

From the company number down to deals, leads and channels

From that objective the chain runs down. The company number sets how many deals it needs. The deals set the leads, the prospective customers a campaign brings in.

Each way of reaching them, ads or conferences or networking, gets its own share of the number.

Set every target below the top figure from that one number. Bottom-up metrics rarely add to the goal on their own. It is the same instinct as narrowing to one number the whole company watches, one level up.

The founders who get better work don't hand over the how

Founders who don't come from marketing tend to make one of two moves with a target.

1. They hand marketing a leads goal and step back.
2. They hire strong people, then tell them exactly how to hit it.

Hire strong people, then don't tell them how

Ben-Tovim's mistake to avoid is the second. Bringing in talented people, then telling them what needs to be done, is how you waste them.

His own line for the job is narrow, and I repeat it to founders often: "My goal is just to look at the plan and say if it makes sense or not."

He sets the direction.

He checks the activities add up to the objective. He leaves the activities themselves to the people he hired.

He described a friend who runs a large agency for startups and sees the same split from outside. Founders who understand marketing get better work from the same team than founders who don't.

Does every activity add up to the company number?

When founders show me a plan, I read it for one thing. Would every activity, each hitting its own number, add up to the company number? If not, I am holding good activities that never reach the goal, a traffic chart that traces back to nothing among them.

The price is real. You leave the choice of activities to the people you hired, even when you are sure you would choose differently.

Sales and marketing only align on a number they share

Sales and marketing drift apart the moment each gets its own number to chase. I've watched it from the marketing side: sales says it lacks qualified leads that clear every filter and never show up, and marketing says sales can't close them.

Ben-Tovim's fix is a shared target, and I'd put it in first. "Being a joint target makes things easier," he says.

Give marketing a leads goal alone, he warns, and here is what you get: "they will focus on leads and not on how they're going to be converted."

The conversion he means, turning a lead into a paying customer, is what a leads-only number ignores. My quick check is simple: does marketing work the same number as sales?

That shared number holds at MDI because the team is small. Ben-Tovim treats a bigger team as just needing everyone on the same goals. That is where I'd push back. On a team of five, one person still sees every link, so the cascade holds unwritten. Past that size it doesn't, unless it's written down.

A missed number now points at the link that broke

At the quarterly review, the founder who built the plan this way traces any activity to the company figure in one sentence. A number that came in low lands on one broken link, rather than the whole team. The standoff over who owns the miss gets shorter, because the plan already answers it.