How to know if your paid ads are driving growth

The right optimization event, the action you tell the platform to chase, depends on your stage, and it will not stay the same for long. Ron Oren, co-founder and CMO of Imagen, started by optimizing toward a subscriber, then toward an earlier action that tends to predict one, and now, at larger scale, toward broad awareness instead. He resets it every few months as the quality of what it brings in shifts.

A paid blackout has to run for weeks, not days, or it measures noise instead of impact. Word of mouth and reputation move over a quarter at the fastest, so a few days with the ads dark tells you nothing about what they carry. Etgar Shpivak, who works with seed and Series A founders on marketing, would not run this test on a company with a few weeks of runway, because the read is only worth the signups it costs.

Growth that isn't coming from paid ads starts with knowing your channel split before anything else. Ron Oren works out what share of Imagen's growth already comes from organic and direct, the signups you never paid for, then looks for the activities that line up with real lifts. He runs the same test city by city, so a rise traces cleanly to one cause instead of getting lost in a national average.

Founders undervalue brand because paid answers by tomorrow and brand does not. Ron Oren sees founders measure results daily, even hourly, chasing the fast number, while brand, the slower work that makes buyers recognize you, needs a quarter or a year to show for it. A figure you can check tonight beats one you cannot read until next quarter. Etgar Shpivak's answer is to pay someone to defend the slow number, because on its own it always loses.

How much to move into brand, the slower work that makes buyers recognize you, depends on whether you can measure it. Ron Oren's rule is not to jump to 40% or 50%: work up to it, or start with a slice small enough to risk. The very large brands at 70% to 80% built that over years, and a company working out what a healthy marketing spend looks like by stage is not there yet.

Before spending heavily on paid, a startup should look for pull it did not buy. Ron Oren argues that ads belong third or fourth in the order: build something worth talking about, get a few influential users genuinely excited, see whether a community forms, and only then put real money behind paid. That early, an unpaid reaction you could not have staged tells you more than any campaign result the platform hands back.

An optimization event is the action you tell the ad platform to chase: a signup, a purchase, or a page view. Meta and Google tune their systems to maximize exactly that action. Because the target shifts every few months, a paid setup that worked in spring may be chasing a different goal by fall.

The target shifts because the company's needs and the platform's algorithm change through the year, and each change points the system at a different action. That means you can't judge a paid setup against a goal that no longer exists. Etgar Shpivak wouldn't renew a budget on performance measured against a target swapped out weeks ago.

Meta and Google's reports are good for cost and cash-flow math, not as proof the ads grew the company. They report a signup and take credit for it, without asking whether that person would have arrived anyway. The one read they can't flatter is the growth that survives when you switch the paid campaigns off for a set window.

Brand marketing pays back over a quarter to a year, far slower than paid ads you can check tonight. That is why it loses almost every budget fight to a fast number. For an early company with under a quarter of runway, betting a large budget on a cycle that slow is a luxury few can afford.

An incrementality test measures the extra sales the advertising truly caused, instead of assuming them from the platform's report. The practical method is geographic: switch ads off in a few regions, sometimes one city, and compare growth against regions that kept running. That way any lift you see is easy to attribute to its source.

A startup should get help once it is spending on paid faster than it can honestly measure the return, which is most early companies. Etgar Shpivak, a marketing consultant who works with seed and Series A founders, sets up the blackout, names the signals worth watching, and reads the result without the platform's built-in optimism. His bio explains how he works with founders on exactly this kind of question.