Last updated: September 12, 2026
This page describes a federal regulation and is not legal advice. Your call policy belongs in front of your own counsel before it becomes a dialer setting.
What the 7 in 7 rule says
Regulation F presumes a debt collector violates the FDCPA by placing more than seven calls to a person about one particular debt within seven consecutive days, or any call within seven days of a telephone conversation about that debt. Both limbs live at 12 CFR 1006.14(b)(2), and both are presumptions that can be rebutted.
The prohibition underneath the numbers sits at 1006.14(b)(1), and it reads on intent. A debt collector must not place telephone calls or engage any person in telephone conversation "repeatedly or continuously with intent to annoy, abuse, or harass any person at the called number." Counting is how the CFPB made that testable, since nobody audits intent from a dialer log.
Paragraph (b)(2)(i) sets the compliance side, where a collector under both thresholds on a particular debt is presumed to comply. Cross either one and (b)(2)(ii) presumes the opposite. The version repeated on most collection floors, seven calls then stop, is half of one limb and none of the other.
Presumption is the load bearing word
A presumption is a starting position in an argument. It decides who has to explain themselves, and it can be pushed back from either side.
Commentary at 14(b)(2)(i)-2 lists factors that rebut the presumption of compliance, so a collector who never crossed seven can still land on the wrong side of (b)(1). Calls bunched into a short window, a pattern of voicemails, calls placed after the consumer asked the collector to stop, and earlier conduct such as abusive language all appear there. Six calls in one afternoon sits inside the number and well outside the point.
Commentary at 14(b)(2)(ii)-2 runs the other direction. A collector over the threshold can rebut with facts such as a call placed to comply with applicable law, a call tied to active collection litigation, a call the consumer asked for, or a call carrying time sensitive information that gave the consumer a chance to avoid a demonstrably negative effect.
Neither list is closed, and the factors get weighed against whatever else the file shows. Which is the honest limit of QA software here: scoring cannot make that argument for you, it only produces the record the argument runs on.
What counts as a call attempt
Paragraph (b)(3) excludes three kinds of calls. A call placed with that person's prior consent given directly to the debt collector, within a period no longer than seven consecutive days after the consent, comes out. So does a call that is not connected to the dialed number. The third exclusion covers calls to the persons at 1006.6(d)(1)(ii) through (vi), meaning the consumer's attorney, a consumer reporting agency where law otherwise permits, the creditor, the creditor's attorney, and the collector's own attorney.
Everything else counts, answered or not. A voicemail counts. Consent is narrower than it looks, since it has to reach your agency directly rather than through the creditor, and it ages out in seven days.
Our reading is that "not connected" describes a call that never reaches the line, which leaves an unanswered ring inside the count. Settle that one before it becomes a retry rule, since the difference is worth several attempts a week per account.
Calls placed to somebody other than the consumer to acquire location information appear to count the same way, which catches agencies that run skip tracing on a separate dial plan.
The count runs per debt
Paragraph (b)(4) defines particular debt as each of the consumer's debts in collection, with one carve out: student loans serviced under a single account number count as one.
Say a hospital system places three accounts for the same consumer with your agency on a Monday. Each account is its own particular debt, so twenty one attempts across seven days can sit inside the presumption on all three at once. Nobody picking up on attempt nineteen is counting per account, and those commentary factors on calls in rapid succession are right there for whoever reads the file later.
Comment 14(b)(4)-1 adds the part that breaks most counting systems. A call where the collector and the consumer discuss two debts counts toward both, and an unanswered call counts toward at least one. The count therefore depends on what the call was about, and that fact lives in the audio rather than in the dialer record.
The conversation limb is the one that bites
For seven consecutive days after a telephone conversation with a person about a particular debt, that debt is closed for calls to that person. The trigger there is a single date rather than a running count.
Commentary says the conversation counts regardless of which party started the discussion of the debt. An inbound call your agency answers on Tuesday starts the same clock an outbound right party contact would have.
We would have called the seven call count the harder one to manage, until you look at where inbound calls get logged. Inbound sits in the ACD or on the payment line while the campaign sits in the dialer, so a consumer who calls Tuesday about a balance gets two minutes with somebody who never touches the account notes, and Thursday's campaign dials anyway. That failure produces no alert and surfaces months later when somebody rebuilds a timeline for a dispute.
Hours, media, and the disclosure on the same call
Frequency is one rule among several on the same call. Absent knowledge of circumstances to the contrary, 1006.6(b)(1)(i) treats a time before 8:00 a.m. or after 9:00 p.m. local at the consumer's location as inconvenient, so a lawful seventh attempt placed at 9:15 p.m. fails a different paragraph. Prohibited communication media sit at 1006.14(h), and the mini Miranda sits at 1006.18(e), with the attempting to collect a debt language in the initial communication under (e)(1) and a shorter disclosure on later ones under (e)(2). Every call inside your seven is also a call somebody has to open correctly.
State rules can be tighter than the federal presumption on both frequency and hours, and the federal count does nothing about them. Recording disclosure on the same call is a separate obligation again, turning on the state you are calling into.
What a recording and a call log prove together
A dialer log proves attempt frequency. The number dialed, the timestamp, the disposition, and enough history to rebuild a rolling seven day window per phone number all sit in the export, and none of it describes what anybody said.
A recording proves the content: whether the disclosure was given, whether the consumer asked for the calls to stop, whether the exchange amounted to a conversation about the debt or thirty seconds of a wrong number.
Neither one proves which debt a call concerned. Audio carries no account number, and a consumer with three accounts at the same agency generates a log that looks identical whether the collector worked one of them or all three. That gap is where compliance programs in this vertical actually fail, and it usually traces back to disposition codes written for collector productivity rather than for a frequency audit. Tying the attempt log, the recording, and an account level disposition together per debt is the whole job, and most agencies have the first two.
Scoring transcripts closes some of the rest. A criterion can ask whether the collector named the account or the original creditor, whether a cease request was made, and whether the exchange met the description of a conversation about that debt, each flagged with a timestamp a compliance manager can open. Rule based criteria hold up under machine scoring, while anything asking for a read on tone gets shakier, which Aircall and StackAI both note about automated QA.
CallSniper ingests recordings after the call ends, transcribes and separates speakers, then scores against criteria your compliance team writes, with a call continued after a cease request sitting in the auto fail column rather than costing points. More on QA for collections agencies, where the verification and disclosure criteria sit beside these.
Take one collector's week of audio and the dialer export for that same week, then see whether the two agree on which debt each conversation concerned. A pilot runs about 30 minutes on Zoom on your own calls, at a fixed pilot price.
Frequently asked questions
What is the 7 in 7 rule in debt collection?
Under 12 CFR 1006.14(b)(2)(i), a debt collector is presumed to comply with the FDCPA harassment prohibition by placing no more than seven calls to a person about a particular debt in seven consecutive days, and no call within seven consecutive days after a telephone conversation with that person about that debt. Crossing either threshold flips the presumption under (b)(2)(ii).
Is the 7 in 7 rule a hard limit?
No. Both directions are presumptions the official commentary says can be rebutted. A collector under seven can still violate 1006.14(b)(1) on a pattern of calls in rapid succession, and a collector over seven can rebut with facts such as a call required by law.
Do unanswered calls count toward the seven?
Calls not connected to the dialed number are excluded by 1006.14(b)(3)(ii). Whether an unanswered ring counts is a reading question for your counsel, since a voicemail plainly counts and the exclusion is written around connection rather than around pickup.
Does the seven day pause start if the consumer calls you?
Commentary to 1006.14 says a telephone conversation about a particular debt counts regardless of which party started the discussion. An inbound call your agency answers opens the window on that debt, which is why inbound and outbound have to write to the same account record.
What hours can a debt collector call?
Section 1006.6(b)(1)(i) treats a time before 8:00 a.m. or after 9:00 p.m. local at the consumer's location as inconvenient, absent the collector's knowledge of circumstances to the contrary. State law and a consumer's own stated circumstances narrow that window further.
Sources
- Both presumptions, the prohibition at (b)(1), the excluded calls at (b)(3), and particular debt at (b)(4). 12 CFR 1006.14, read on consumerfinance.gov and then confirmed against eCFR primary text on 2026-09-12. Verified from primary text, with the retrieval saved to this article's evidence folder.
- Rebuttal factors, counting per debt, and conversations started by the consumer. Official commentary 14(b)(2)(i)-2, 14(b)(2)(ii)-2, and 14(b)(4)-1, read in summary form on 2026-09-12. Partially verified, and the full comment text needs reading before publishing.
- Location information calls counting toward the same seven. Comment 14(b)(2)(i)-1.iii, read in summary form only. Unverified, and flagged in the body.
- Inconvenient time, the third party list the exclusion borrows, and the mini Miranda. 12 CFR 1006.6(b)(1)(i), 1006.6(d)(1)(ii) through (vi), and 1006.18(e), read on consumerfinance.gov on 2026-09-12. Verified.
- Reading of "not connected to the dialed number" and the twenty one attempt example. Editorial interpretation and a hypothetical written for this page.
- Aircall and StackAI, on automated scoring being more reliable on rule based criteria than on subjective tone. Third party, gathered 2026-09-08. Unverified.
- Recording ingestion, transcription, speaker separation, scorecard criteria, auto fail flags, and manual upload. CallSniper product documentation. Verified.