How Long Should You Run Ads Before Deciding They Do Not Work?

Run an ad long enough to observe the customer journey you actually sell through, but never keep spending merely because a platform says it is learning. The right decision point comes from business evidence, not the calendar alone.

What is the short answer?

Before you decide, the campaign should have reliable tracking, enough opportunity to produce the target outcome, and enough time for delayed conversions or sales follow-up to appear. For some high-volume e-commerce campaigns, useful evidence can arrive within days. For a lower-volume service with a two-week consultation process, the same number of days may reveal almost nothing.

Google recommends allowing a ramp-up period of at least two weeks or three conversion cycles when evaluating some value-based bidding changes. Its reporting also shows that conversions can arrive days after the click, making recent CPA look worse and recent ROAS look lower than they eventually become. Those are useful cautions, not universal guarantees that every weak campaign deserves two more weeks.

The four clocks that determine a fair test

A campaign that ran for 14 days at a very small budget may have less evidence than a campaign that ran for three days at meaningful volume. Conversely, high spend does not make bad measurement trustworthy. Time and spend are inputs; neither is the final verdict.

Define the decision before the campaign starts

The most expensive time to invent success criteria is after the results disappoint you. Before launch, write down the target business outcome, the maximum amount the business can afford to pay for it, the expected buying delay, and the maximum test loss.

For lead generation, work backwards from a customer rather than a form fill. The following is an illustrative planning example, not a market benchmark:

This calculation does not say how much you must spend. It tells you what the evidence means. Five leads at $20 each may look efficient, but not if none fit the service area or budget. Two leads at $80 may look expensive, but not if one becomes a profitable long-term customer.

When should you stop or fix the campaign immediately?

Learning periods do not excuse a structurally invalid test. Pause the affected part of a campaign when continuing would only buy more of a known problem:

• Tracking is wrong: Purchases are duplicated, test events are counted as leads, calls are missing, or the campaign optimizes toward page views instead of real inquiries.
• Traffic cannot buy: Google search terms show jobs, free resources, unrelated products, or locations you do not serve.
• The offer is unavailable: Stock, booking capacity, pricing, payment, delivery, or the landing page contradicts the ad.
• The lead path is broken: The form fails, WhatsApp opens the wrong number, calls go unanswered, or no one follows up within the promised window.
• A hard financial boundary is reached: The pre-agreed test loss is exhausted without evidence that justifies extending it.

Fix the cause, document the change, and restart the evaluation window deliberately. Do not mix data from a broken setup with data from the corrected one and call it one clean test.

When is it reasonable to keep running?

More time can be justified when the campaign is reaching the intended market, the measurement is sound, the buying cycle has not completed, and early indicators are moving in the right direction. For a service business, that may mean relevant searches, qualified calls, attended consultations, and proposals even before revenue closes. For e-commerce, it may mean product views, carts, checkout starts, and purchases that match the expected funnel economics.

Google Ads attribution reports can show average days or hours to conversion and the number of ad interactions involved. Conversion-lag reporting can also estimate how recent CPA or ROAS may change when delayed conversions arrive. Use account-specific evidence like this instead of borrowing someone else's “three-day rule.”

On Meta, separate normal short-term variation from a creative or offer problem. If people are not stopping, clicking, or taking the intended next step after meaningful delivery, waiting alone is unlikely to repair the message. Test a materially different angle or format. Meta's own business guidance points advertisers toward controlled A/B testing for creative and placement decisions.

Should Meta Ads and Google Ads get the same evaluation window?

No. Google Search often captures people already expressing demand. You can inspect their search terms and quickly learn whether the campaign is matching relevant intent. Meta commonly creates or develops demand through creative, so the message, format, frequency, and post-click journey carry more of the burden.

That difference changes what you diagnose first:

• Google Ads: Check search intent, matched locations, conversion goals, landing-page relevance, and lost demand before assuming the platform needs more time.
• Meta Ads: Check whether the creative earns attention from the right people, communicates the offer, qualifies the buyer, and gives the algorithm a meaningful outcome.
• Both: Judge lead quality and revenue in the same business system. Platform dashboards should not be the only source of truth.

What changes for Lebanon and GCC businesses?

Many regional sales journeys cross channels. A customer may see a Meta ad, search the brand on Google, visit Instagram, ask a question on WhatsApp, and pay later by link, transfer, cash, or in person. If the business only records the first form or last platform click, the test window can look shorter or less productive than the real journey.

Agree on who owns WhatsApp and call follow-up, which countries and cities are serviceable, which languages the team can answer, and when a lead becomes qualified. Record confirmed orders, cancellations, no-shows, returns, and cash-on-delivery refusals where relevant. A reported purchase is not the same as collected, profitable revenue.

Use a three-level review cadence

1. Daily safety check: Confirm delivery, spend, links, tracking, geography, comments, inventory, and lead routing. Correct obvious failures; avoid redesigning the strategy every morning.
2. Evidence review: After meaningful delivery and the normal conversion lag, compare outcomes with the pre-agreed thresholds. Look at search quality or creative response, not just blended totals.
3. Business review: After the sales cycle has had time to mature, reconcile platform results with qualified leads, orders, margin, and revenue. Decide whether to scale, revise one major variable, or stop.

Record each material change and the reason for it. Frequent simultaneous edits make it difficult to know whether performance moved because of the creative, audience, bid strategy, budget, offer, landing page, or normal variation.

What should an owner ask the agency?

Ask for the evaluation window and stop-loss before spend begins. The agency should be able to explain which business outcome is primary, how tracking was tested, what a qualified lead means, how long conversions and sales normally take, what evidence triggers a change, and which metrics will be reconciled with actual revenue.

Be cautious with two opposite answers: “Give it time” without a financial boundary, and “It failed after three days” without enough evidence. A credible recommendation should say what has been learned, what remains uncertain, what the next test isolates, and how much that learning is allowed to cost.

The Verdict

Run ads until the campaign has had a fair opportunity to produce the business outcome, not until an arbitrary date arrives. A fair opportunity requires valid measurement, relevant delivery, enough evidence, and time for the real buying cycle. It also requires a loss boundary that protects the company.

If the setup is broken or the traffic cannot buy, fix it now. If the journey is valid but immature, wait for the appropriate lag. If the campaign has crossed its evidence and financial thresholds without a credible path to improvement, stop or change one major variable. That is disciplined testing—not impatience, and not blind optimism.