How long does paid advertising take to work?

The right answer is a sequence of evidence, not a number of days

A campaign launches on Monday. By Friday it has clicks, two enquiries and no revenue. One provider calls that promising. Another says the algorithm needs 90 days. Neither answer tells the owner whether the investment is working.

I separate activity, commercial signal and proof. Activity means the ads reached people. Commercial signal means suitable prospects responded. Proof means those prospects progressed far enough to show that the customer-acquisition economics can work repeatedly.

The distinction matters because every business has a different buying cycle. An ecommerce product may be purchased during the first visit. A legal, consulting or software decision may take several stakeholders and many weeks. If the normal sales cycle is 45 days, a 14-day revenue verdict is structurally premature.

At the same time, “be patient” is not a licence to protect a weak campaign. Broken tracking, irrelevant traffic and an offer that nobody wants should be fixed as soon as the evidence is clear. The business needs a clock for each type of decision.

The Four Clocks of Paid Growth

This is the framework I use to set expectations before spend begins. All four clocks start together, but they do not finish together.

1. The market clock

Existing demand moves faster than demand creation. Google Search can capture a buyer already looking for a solution. Meta, LinkedIn or YouTube may first need to make the problem visible and build confidence. The offer, price, risk and number of decision-makers all change the time required.

2. The signal clock

A small budget in a narrow market may produce only a handful of meaningful outcomes each month. That does not automatically make advertising wrong, but it limits the certainty of the decision. Decide before launch what a fair test can realistically buy. My Revenue-Backed Budget Model shows how to work backwards from customer value instead of choosing a convenient daily amount.

3. The platform clock

Advertising systems adapt as data arrives. Google says a bid strategy can show “Learning” after a new strategy or material setting and composition changes, and that key metrics may fluctuate during that period. Meta similarly describes an initial learning phase and recommends simplifying structure and minimising changes. This explains volatility; it does not prove the offer or economics are sound.

4. The revenue clock

The dashboard may record a form today while the business confirms the opportunity weeks later. Google's conversion-lag guidance notes that delayed conversions can temporarily make cost per acquisition look inflated and return on ad spend look depressed. For owner decisions, the CRM and order system—not the newest platform row—must close the clock.

Use four evidence windows instead of one deadline

The windows below are decision stages, not performance guarantees. A high-volume ecommerce account may move through them quickly. A low-volume B2B service may require a full quarter. The rule is to ask only the question that the available evidence can answer.

Notice what is missing: a universal promise that paid ads “work in 30 days.” Thirty days may be long enough to diagnose a landing-page failure and too short to observe one enterprise sale. I would rather define the evidence standard before launch than move the deadline after the results arrive.

Know when to wait, repair, scale or stop

Wait when the system is healthy but the revenue clock is incomplete

Tracking works. Traffic is relevant. Qualified prospects are entering the expected sales process. The current cohort simply has not had enough time to mature. Keep the campaign stable, monitor leading indicators and schedule the commercial review after the normal decision window.

Repair when one stage is clearly weak

Relevant buyers click but do not enquire. Sales rejects most leads. The landing page promises something different from the ad. The CRM cannot identify the source. These are not patience problems. Repair the earliest weak stage before adding budget. If lead volume looks healthy but pipeline does not, use the CPL-to-Pipeline diagnosis.

Scale when qualified economics repeat

One unusually strong week is not a scaling system. I look for repeated qualified outcomes, a cost the gross profit can support, enough delivery headroom and operational capacity to serve more customers. Increase the budget in controlled steps and watch marginal performance, not only the blended historical average.

Stop when the economics or market evidence fails

Stop or redesign when a fair test reaches the intended market but cannot produce a credible path to profitable customers. Also stop when measurement is too unreliable to protect the next pound, dollar or euro. A platform learning label should never override a business stop condition.

The CEO paid-growth scorecard

Use this five-question scorecard at each commercial review. A “no” identifies the next business decision more clearly than another dashboard export.

The paid channel is only one part of this operating system. My growth partnership services connect acquisition with creative, landing pages, measurement and sales feedback. For current examples of that connection, review the case studies; each one separates Thomas's contribution from wider market and client factors.

Practitioner note: I do not call a campaign successful because it has left a platform learning phase. I call it ready for the next investment decision when the evidence has reached the deepest business outcome available at that point.