Lead Spam Filter Cost Calculator: Are You Blocking Real Enquiries?

Enter reviewed labels, not a filter's own verdict

Use one frozen, deduplicated enquiry cohort and one policy version. The four labelled cells must be mutually exclusive. Positive means spam; blocked genuine enquiries are false positives. Unreviewed records are outside the matrix. This local tool needs no messages, email addresses, tokens, API connection or audit credits. It does not install or modify a filter.

Each count accepts whole values from zero to one million; cost assumptions accept non-negative amounts up to one million with two decimal places. All money uses one currency and one declared cost boundary. Blank, fractional-count, negative, non-finite and over-limit inputs are rejected. Changing an input clears stale results. Zero denominators display Not defined and export as null.

Read each denominator and cost boundary

Reviewed N is the sum of all four cells. Coverage = N/(N + unreviewed). Block precision = spamBlocked/(spamBlocked + genuineBlocked). Spam capture = spamBlocked/(spamBlocked + spamAllowed). Genuine block rate = genuineBlocked/(genuineBlocked + genuineAllowed). Allowed-queue spam share = spamAllowed/(spamAllowed + genuineAllowed). These are descriptive ratios for the supplied reviewed subset, without confidence intervals or weighting.

Let c be review cost per allowed enquiry, v the assumed contribution loss per genuine enquiry wrongly blocked and k incremental policy cost. Baseline = N × c. Filtered cost = allowed × c + genuineBlocked × v + k. Net saving = baseline − filtered cost. Break-even v = (blocked × c − k)/genuineBlocked when genuineBlocked is positive. Negative break-even means operating cost exceeds review savings even when v is zero; it must not be clamped to zero.

Try the synthetic default

Use 160/40/10/190 in the four cells, 50 unreviewed, c = 1.50, v = 25 and k = 50. Reviewed N = 400, blocked = 170 and allowed = 230. Coverage = 88.89%, block precision = 94.12%, spam capture = 80%, genuine block rate = 5% and allowed spam share = 17.39%. Baseline cost is 600; filtered cost is 345 + 250 + 50 = 645; net saving is −45. Break-even v is 20.50.

Set v to 10 without changing the counts: net saving becomes 105. At v = 20.50, it is zero. This is a sensitivity exercise, not evidence that genuine enquiries have either value. A zero genuineBlocked count leaves the per-error break-even undefined; it does not establish a zero error probability in the population.

Do not turn a sample into a revenue claim

The comparison assumes equal review cost and equal expected missed-enquiry value. A smaller sales queue is not automatically a better queue. It ignores future attacker adaptation, customer frustration and conversion friction before capture unless you study those separately. Security outages and denied requests are not independently labelled spam. Unknown labels stay unknown, and reviewing only allowed records cannot reveal the blocked genuine share.

Use the evidence checklist to approve the scope, the prompt for a human-reviewed memo and the technical guide for reproducible row aggregation. Synthetic arithmetic is tested locally; no live form or paid model is called. Do not disable a control based on this calculator.

Sources and boundaries

Checked October 25, 2026: Cloudflare validation (updated October 22) separates server verification from a browser widget. HubSpot spam prevention (updated September 22) documents differing filter and reporting effects. Google classification metrics (updated January 12) supplies precision/recall definitions. These sources do not validate this scenario's costs or label accuracy.