Search the guides

Traffic fundamentals

When a strict filter wastes paid traffic

A cleaner traffic report can hide lost customers. Measure the cost of rejected visits before tightening your protection rules.

The blocked percentage is not the target

A stricter filter blocks more visits, but accepted leads fall. That result needs investigation: the rule may have removed spam, genuine customers, or both. A rising blocked percentage cannot distinguish them.

A false positive is a legitimate visitor rejected by mistake. If the person already clicked the ad, your later rejection does not necessarily remove the acquisition cost. Judge the rule by useful business outcomes and the abuse it prevents, not by how busy its blocking counter looks.

A small example exposes the trade-off

Consider a hypothetical campaign that starts rejecting an unfamiliar network category. Among the remaining visitors, a larger share buys. The conversion rate looks better, yet the campaign produces fewer orders because some rejected people would also have bought.

The denominator—the group used to calculate the rate—has changed. Compare results for the whole paid audience as well as for admitted visitors. Keep the spend and the definition of a useful order consistent.

This does not mean filtering is always harmful. Spam handling and service overload also cost the business. The decision needs evidence of both the benefit and the legitimate activity lost.

Track both benefits and losses

The comparison below separates two questions: how the allowed group behaves, and what the whole campaign produces for its spend. A rule can improve the first while hurting the second.

Keep both views in the same report. Otherwise, one team may celebrate cleaner traffic while another sees fewer accepted orders, with neither looking at the complete result.

Two approachesTwo measures answer different questions

Among admitted visitors

  • What share purchased?
  • How many forms were useful?

Across the campaign

  • How many orders arrived?
  • What did acquiring them cost?

An improvement on the left does not establish an improvement on the right.

Make rejection reasons visible

Ask for the decision reason and the rule version. “Suspicious” is difficult to investigate; “submission rate exceeded on this form” tells you what to reproduce. Retain the information needed for that investigation without collecting unrelated personal details.

Treat broad network or browser labels as clues. A shared address can belong to an office, and missing browser information can reflect privacy settings. Neither observation alone establishes abusive behavior.

Match customer complaints to the decision record using a time and request reference. Then open the actual advertising route on an affected type of device. Your desktop may never encounter the screen that stopped the customer.

Change one rule with a rollback plan

If observation-only mode is available, use it to record proposed rejections without applying them. Review whom the rule would affect before a limited rollout. Where this mode is unavailable, keep the initial change small and reversible.

Write down the expected benefit, the useful outcome you must preserve, and the condition for reverting. Compare similar traffic, not a handful of clicks under different offers. Change one meaningful setting where possible.

During an active attack, coordinate with the security owner rather than removing essential protection for an experiment. A more precise limit on the affected action may be safer than weakening protection everywhere.

Check lead quality too. More forms are not a recovery if they cannot be contacted or were never intentionally submitted. Agree on what makes a lead valid before examining the results.

Step by stepEvaluate a new restriction
  1. Hypothesis

    Which abuse should decrease?

  2. Observation

    Who would the rule affect?

  3. Limited rollout

    What happens to useful actions?

  4. Decision

    Keep, narrow, or revert.

Agree on the keep-or-revert criteria before enforcement.

Put the filter into the campaign's economics

Choose the business result first: a reachable qualified lead, an accepted order, or another defined outcome. Apply the same definition before and after the rule. Include how cancellations are treated.

Compare acquisition cost with the burden the rule removes, such as repeated form processing or server load. You do not need a made-up monetary value for every minute, but you should show both the cost saved and the useful actions lost.

In a hypothetical lead campaign, staff handle less junk but also receive fewer genuine enquiries. Whether that trade-off helps depends on the team's capacity and the value of those enquiries. Neither “cleaner leads” nor “fewer leads” answers the question alone.

For a rare, costly failure, document the incident the rule is intended to prevent and when it should activate. A targeted emergency restriction may be useful even if leaving it on permanently would waste traffic.

Make the comparison credible

Record other changes that could explain the result: a new creative, a weekend audience, a different offer, or a new placement. A before-and-after chart does not isolate the filter's effect when several things changed together.

Give compared groups the same time to produce an outcome. A lead may be qualified later and an order may be cancelled. A newer group can look worse simply because its final results are not yet available.

Do not keep a clearly broken rule running to reach an arbitrary sample size. A reproducible blocked checkout or confirmed customer complaints can justify a rollback while you continue investigating.

Review allowed visits as well as rejected ones. Missed abuse and mistaken rejection are different errors; counting only one can hide the other.

Recover the route after a false positive

After confirming a false positive, fix the narrow cause and repeat the affected journey. A permanent exception for one customer's address may help that person while leaving the next legitimate visitor blocked for the same reason.

Check for a pending order, accepted lead, or payment attempt before asking the customer to retry. Restoring access does not remove a duplicate that another attempt might create.

Save a reproducible test with only the necessary diagnostic details. Tell the team what failed, which visits were affected, and what the retest established. The useful result is a more precise rule and a check that catches the mistake if it returns.

Keep, narrow, or remove

Compare paid visits, allowed visits, confirmed outcomes, and the abuse you observed. Note any campaign changes or missing evidence that limit the conclusion.

Keep a rule with a clear benefit. Narrow it when the problem belongs to one action or group. Remove it when it repeatedly rejects legitimate activity without a defensible benefit. If evidence is still insufficient, keep the conclusion open instead of calling the strictest setting the best.

Questions answered

Can I use the blocked share as a quality score?

It tells you how often the filter rejected a visit. To judge quality, check whether those rejections were correct and whether the campaign benefited.

ADSTRAFFIC / ENAll guides →