Buying Sales Leads: What Happens When You Dial Them

Buying sales leads is almost always written about as an email problem. Page one for the term hands you spam traps, bounce rates, verification tools, and a CAN-SPAM warning. Almost nobody writes about the phone column, which is the column most outbound teams bought the list for in the first place. That gap is where the money quietly goes.

We build the dialer and CRM that outbound teams run their day on, so we tend to meet a purchased list on the day somebody loads five thousand records and starts dialing. The pattern repeats often enough to write down. Here is what the phone half of a bought list does on a real dialing floor, what the rules expect from contact data you did not collect yourself, and how to test a vendor for a couple hundred dollars before you spend five thousand.

What are you actually buying when you buy sales leads?

Two very different products get sold under the same word, and the price sheet rarely tells you which one you are holding.

• A compiled contact list. Names, titles, companies, emails, phone numbers, assembled from public filings, web sources, and resold databases. Nobody on it asked to hear from you. It is a research shortcut, and nothing more.
• A generated lead. Somebody filled out a form somewhere and, on most of these forms, agreed to hear from a list of partners they scrolled past. The consent is real in a paperwork sense and often very thin in a human sense.

Those two things behave nothing alike on a phone. The compiled record is a stranger who will ask who you are. The generated lead is a person who may have filled out a quote form nine days ago and has already been called by six companies. Both can work. Buying one while pricing it like the other is how a desk ends up furious at a vendor who delivered exactly what was ordered.

The first question for any vendor is therefore not how accurate the data is. It is where every field came from, field by field, and when.

Why does buying sales leads look worse on the phone than in the inbox?

Because a bad email costs you a bounce and a bad phone number costs you a rep's afternoon.

An invalid address fails silently and instantly. A stale phone number does something more expensive. It rings. Somebody picks up. Your rep runs a pitch at a person who has nothing to do with the record on screen, marks it wrong party, and moves on ninety seconds poorer. Multiply that across a five thousand record file and the wasted talk time dwarfs what you paid for the data.

Worse, a number that was disconnected and handed to somebody else is the kind of call that generates complaints. Complaints feed the analytics engines that carriers use to label outbound numbers, which is the road to calls showing as spam likely on caller ID. A cheap list can degrade the phone numbers you dial from, and that damage outlives the list.

There is a second cost nobody invoices you for. Reps read a list through their fingers. Three wrong party calls in a row and the eighth dial goes out flat, because the rep has stopped believing the screen. A file with a visible defect rate does more damage to a floor's energy than a smaller file of records people trust.

So the phone column deserves its own budget line: phone number validation before the first dial, not after the first complaint.

Where do the data accuracy numbers actually come from?

Here is the part worth slowing down on. Search for how fast business contact data goes stale and you will find confident figures everywhere. Two percent a month. Twenty two percent a year. Twenty five to thirty five percent for phone numbers. Thirty to seventy percent. Forty three percent of business phone numbers change inside twelve months.

We went looking for the study behind those figures in September 2026 and could not find one. Every number we traced led back to a blog post published by a company that sells contact data or data cleaning. Same for the accuracy guarantees on the other side of the trade, the ninety five percent and ninety seven percent badges. Those are vendor self reports about vendor performance, measured by a method nobody publishes.

None of that makes the figures wrong. Contact data clearly does decay, and anyone who has worked a year old list knows it. What it means is that you cannot plan against those numbers, because you have no idea what they measured. A vendor claiming ninety five percent accuracy may be counting a record as accurate if the company still exists.

The only decay rate that matters to your desk is the one your own dialer reports on your own list. You can have that number in a week, which we will get to.

Treat the published figures the way you would treat a rate quote from someone who has not seen your bill. Useful for knowing the category is expensive. Useless for budgeting.

What do the rules expect from a list you did not build?

This part is educational background, not legal advice. Run anything real past your own counsel or compliance team.

Buying the data does not buy the homework that goes with it. Four things are worth knowing before a purchased file touches a dialer.

The registry scrub is on the clock

Under the Telemarketing Sales Rule, the FTC's safe harbor for an accidental call to a registered number requires, among other conditions, that a seller accessed the national registry no more than 31 days before calling a consumer and kept records showing it. That is from the FTC's published Q&A, retrieved 22 September 2026. A vendor telling you a list is already scrubbed is telling you about a scrub on their clock, not yours. The same guidance requires you to maintain your own company specific list of numbers you may not call, which is a different job from the national registry and lives with you permanently. We wrote about building that internal do not call list separately.

Business lists and consumer lists are different products

The same FTC guidance says most phone calls to a business made with the intent to solicit sales from that business are exempt from the registry provisions. So a B2B file and a consumer file sit in genuinely different places, and a vendor selling both under one banner may be glossing over that. The exemption is about the registry. It is not a general exemption from every calling rule, and a small business owner's listed number is very often a personal cell.

A purchased consent does not carry everywhere

The FTC guidance on written consent to call a registered number says the seller must have obtained express consent directly from the consumer to place prerecorded calls, which it spells out as meaning a seller cannot place calls with prerecorded messages to consumers whose information the seller obtained from third parties. Read that line twice if your plan involved buying leads and dropping a recorded message on them.

On the related question of whether one form can consent you into calls from a whole partner list, the FCC rule that would have required separate consent per seller was vacated by the Eleventh Circuit in Insurance Marketing Coalition v. FCC on 24 January 2025, and the FCC later removed the vacated language. So the older standard applies again. That is the legal position, and it is not the same thing as a person remembering who you are.

Reassigned numbers have a database, and it wants a date

The FCC runs a Reassigned Numbers Database, live since 1 November 2021, that tells a caller whether a number was permanently disconnected after a date you supply. Carriers must age a disconnected number at least 45 days before handing it to somebody new, and they report disconnections monthly. Per the FCC's database page, the safe harbor it offers requires that you had consent from the intended recipient, that you or your agent checked the database before calling, and that the database returned a "no" that turned out to be wrong.

Notice the shape of that. The query takes the date you obtained consent, or the last date you verified the person was at that number. A compiled record you bought on Tuesday gives you neither. The safe harbor was built for callers who have their own consent history, which is exactly what a purchased list does not come with.

How do you test a list before you buy the whole thing?

Every vendor guide says start with a sample. Almost none of them say what to do with it. Here is the version we hand to teams, built for a phone desk rather than an inbox.

Buy 500 records. Pay full retail for them and accept no hand picked sample, because a sample the vendor assembled is a marketing asset. Ask for a random draw from the same selection criteria you would use for the real order, and say so in writing.

Then run them through five checks in this order, and write down the number at each step.

1. Validation. Run every phone number through validation before anyone dials. Record how many come back invalid, disconnected, or a line type that does not match what you were sold. This is your first honest accuracy figure and it takes an hour.
2. Scrub. Push the survivors through the national registry and your own internal list. Record how many drop out. A file that loses a large share here was sold to you unscrubbed regardless of what the order page said.
3. Dial the whole remainder. Not the good looking half. Three attempts per record across different days and different hours, using a connect rate math baseline you already trust from your own lists.
4. Disposition wrong party separately. Give it its own code, distinct from not interested and distinct from bad number. Wrong party rate is the single most revealing number about a purchased file and almost nobody tracks it.
5. Count conversations, then divide. Right party conversations, not dials and not connects. Divide what you paid by that count.

Five hundred records at twenty cents is a hundred dollars and maybe two rep days. That is a cheap way to find out whether a five thousand dollar order is worth placing.

One more thing to watch while the test runs. Note how many records you could have found yourself in under two minutes on a company website or a public filing. If most of them are that easy to find, you are paying for assembly rather than access, and the price should reflect it.

Run the same five checks against a second vendor in the same week with the same reps and the same script. Buying sales leads from one supplier and judging the result against your memory of a different supplier from last year tells you almost nothing. Two files, one week, same conditions, is a real comparison.

What does a purchased lead cost once you dial it?

The sticker price is per record. The number you manage against is per conversation. They are rarely within an order of magnitude of each other.

Here is an illustration. These figures are made up to show the shape of the arithmetic, not a claim about any vendor or any list. Run yours.

• 5,000 records at $0.20 each. Spend: $1,000.
• Validation removes 600 as invalid, disconnected, or the wrong line type. Left: 4,400.
• Registry and internal list scrubbing removes 300 more. Callable: 4,100.
• Working those at three attempts each produces 380 conversations with the person named on the record.
• $1,000 divided by 380 is about $2.63 per conversation.

Thirteen times the sticker price, and that is before a minute of rep pay. Now change one input. If a fifth of those conversations turn out to be with somebody who never held the number, you have 304 real conversations and your cost per conversation is about $3.29. The wrong party rate you were not tracking just moved your true data cost by a quarter.

This is also the comparison that tells you whether to buy at all. Put cost per conversation from a purchased file next to cost per conversation from your own inbound, your referrals, and your reactivation list. Plenty of desks find that a bought list is their most expensive source of conversations and their only scalable one, which is a real answer and a very different one from the vendor pitch.

What should you ask a vendor before you pay?

Short list, and the answers matter less than whether the person can answer at all.

• Where did the phone numbers come from, and on what date was each one last verified as belonging to this person?
• How do you define an accurate record, in one sentence, and who measured it?
• Is this file exclusive to me, and if not, how many buyers have it and how recently?
• For generated leads: show me the form, the disclosure text, and the timestamp on this specific record.
• Do you refund or replace on a disconnected number, and what is the window?
• Will you sell me a random 500 from the same criteria at list price?

A vendor who answers those clearly is worth testing. A vendor who answers the accuracy question with a percentage and changes the subject has told you something too.

Exclusivity is the one people skip and regret. A generated lead sold to eight buyers is a person whose phone rang eight times in an hour, and the fourth caller has a different job from the first. If a vendor will not say how many buyers a record goes to, assume the answer is more than you would like and price the file accordingly.

Where this leaves a buying decision

Buying sales leads is a legitimate move for a desk that needs volume faster than it can build a pipeline. The teams that do well with it treat a purchased file as raw material with a known defect rate, measure that rate themselves in the first week, and manage against cost per conversation instead of cost per record.

The teams that do badly with it buy on an accuracy badge, load the file, and find out four months later that the wrong party code was never on the disposition list. The difference costs about a hundred dollars to discover.

Once a list is working, the next question is dial order, which is a separate discipline worth reading up on in scoring a cold list. And if you want validation, scrubbing, dispositions, and the dialing all reporting into one place instead of four tools that do not talk to each other, that is what one platform covers. You can start a 14 day free trial and run the 500 record test on your own data.