Most startup marketing plans begin with a channel. Somebody wants a TikTok presence, somebody else wants Google Ads, and an investor mentions that a portfolio company did great things with podcasts. Growth marketing runs the other direction. You work out what a customer is worth and what you can afford to pay for one, and the channel choice falls out of that math.
So this is about the boring part. The numbers a founding team should settle before real money goes out the door, why your dashboards will mislead you a little, and who should hold the targets once things get busy.
Why Picking Channels Too Early Burns Budget
Picking channels before targets feels productive. Campaigns go live fast, the ad platform shows clicks within a day, and everybody has something to say on Monday.
The trouble shows up around month three. Nobody agreed on what a good result looked like, so the team argues about whether, say, 400 leads at $38 each counts as a win. Sales calls the leads junk. Marketing points at the cost per lead, and neither side can settle it because the target never existed.
Software teams know this pattern. Vizologi recently covered how skipping the scoping phase on a software project can nearly double the bill in the example it walks through. Marketing budgets fail in a similar way when nobody defines success up front.
The Four Numbers to Agree On Before You Spend
You don’t need a 40-tab model to start. Four numbers make a practical starting point, as long as the founders, whoever leads marketing and whoever leads sales all accept them:
- Customer lifetime value (CLV). CLV has a few accepted definitions. This article uses the gross margin a customer contributes over the whole relationship, rather than raw revenue.
- Acquisition cost ceiling. The most you’ll pay to win one customer. CLV sets the outer limit, but cash on hand, retention uncertainty and growth goals often pull the number lower.
- Payback window. How many months of margin it takes to earn back that acquisition cost. Startups with short runway often need this one tight.
- The conversion event that counts. A paid account, a demo that sales accepts, a signed contract. One event, written down.
CLV carries a common source of confusion. A short MIT Sloan Management Review piece argues against subtracting acquisition costs from lifetime value before you report it, since acquisition spend belongs in the decision about whether to chase a prospect. Mix the two and a highly profitable customer who cost a lot to win can look identical to a weaker one who came cheap.
And the gap between those two customers is exactly what you want to see. Keep the two numbers side by side instead.
| Metric | Question It Answers | Typical Trap |
| Customer lifetime value | What is a customer worth over time? | Mixing revenue and margin definitions |
| Acquisition cost ceiling | How much can we pay to win one? | Leaving out salaries and software |
| Payback window | How fast does the money come back? | Ignoring cash runway |
| Conversion event | What counts as a win? | Counting form fills that sales never accepts |
Why Proxy Metrics Can Mislead You
Once spending starts, most teams judge channels with last-click attribution, the simplest model and a common default. It’s cheap and easy to read. It also tends to miss what the ad itself caused. Platforms often target people already leaning toward a purchase, and existing demand, organic search and other touchpoints blur the picture further.
Researchers at Northwestern’s Kellogg School studied close to 1,000 randomized experiments on Facebook ads and found that teams can calibrate last-click numbers against experimental results. Their advice for a single advertiser: run a handful of controlled tests, compare them with last-click results and check whether a simple correction factor holds. That factor belongs to your own campaigns, so it needs testing in your context rather than borrowing from someone else’s.
For a startup, a controlled test can be as plain as holding one region or one audience slice out of a campaign for a few weeks and comparing results. It won’t be perfect. It beats taking a platform’s own reporting on faith, which is what most small teams end up doing.
Who Should Own the Numbers
Somebody has to hold the targets when pressure builds. In a lot of startups that job lands on a founder who also runs product and fundraising, and the marketing plan quietly drifts while nobody watches it.
Three setups tend to work, each with tradeoffs:
- A senior in-house hire. Strongest ownership, highest fixed cost, and usually a long search.
- A fractional marketing leader. Senior judgment for part of the week, often paired with outside specialists who run the channels.
- An agency with an executive at the top. Faster to start, though quality depends heavily on who runs your account day to day.
Whichever route you take, ask how the partner plans to prove results before you sign anything. A reasonable test is whether they insist on agreed targets up front, the way northbound growth marketing engagements open with an audit and measure work against goals both sides agree on before it starts.
If a partner wants campaigns live in week one and a conversation about targets later, that tells you a lot about how the reporting will go.
A Sample 30-Day Setup Before Heavy Spending
Here’s one workable sequence that doesn’t need a data team. Teams with messy data or long sales cycles may need more time, and that’s fine:
- Week one: pull the real numbers. Gross margin per customer, churn or repeat rate, and every cost tied to sales and marketing, salaries included.
- Week two: write the targets down. CLV estimate, acquisition cost ceiling, payback window and the conversion event. Get sign-off from whoever owns revenue.
- Week three: fix tracking. Make sure the conversion event fires in analytics and the CRM, and that both systems count it the same way.
- Week four: pick two channels and one test. Choose channels where your buyers already search or gather, set a spend cap, and plan one holdout test so you can check last-click against something sturdier.
Two channels, not six, is a rule of thumb rather than a law. Small budgets spread thin often can’t produce enough data in any single channel to tell you anything useful, so you end up with six inconclusive experiments instead of two readable ones.
Revisit the Numbers Every Quarter
None of these targets stay fixed for long. Pricing changes, churn shifts, a new competitor pushes your cost per click up.
Put a quarterly review of the targets themselves on the calendar, and check campaign numbers weekly or monthly in between. Trim channels that can’t clear the acquisition cost ceiling once they’ve had a fair window to show results. Channels with long sales cycles or brand effects can look weak on short-term last-click reports, so judge those on incremental results where you can.
Most quarters the answer is dull. One channel carries the load, one looks promising, and the rest is noise you can stop paying for.
