AI lead enrichment tools fill the empty fields on a contact record so a rep can actually work it: work email, direct dial, job title, company size, technology stack, and the signals that say whether this person is in market. If a name lands in your CRM with no reachable address, it is not a lead yet. It is a task.
The market is crowded and the marketing is not honest. Almost no vendor publishes a match rate with a sample size and a date attached. Two vendors publish their own pricing in ways that contradict themselves. And the one comparison guide that does the most thorough job, Clay’s July 2026 list, gets the HubSpot enrichment price wrong too.
So this guide does not hand you a leaderboard. It gives you the credit arithmetic, the mechanism behind a “verified” email, published match-rate figures with their methodology, and a plain reading of the UK, EU and US rules that decide whether you can use the data at all.
The short answer: pick the credit model, not the logo
Almost every AI lead enrichment tool is the same five jobs wearing different clothes. They differ in where the data came from, how a credit gets consumed, what happens when a lookup misses, and whether anyone will tell you which provider produced a given email.
If your only gap is a verified work email, you do not need a platform. You need a verifier. Hunter.io charges $149 a month for 10,000 credits on its Growth plan, where one credit finds an email and half a credit verifies one. That is roughly $0.022 per verified address if you consume the whole plan.
If your gap is a direct dial as well, the arithmetic changes hard. On FullEnrich, a work email costs 1 credit and a mobile number costs 10. Its 15,000-credit Pro plan is $720 a month, so an email-plus-mobile record is 11 credits, or about $0.53 of marginal data cost.
If you need coverage on a list where the top provider leaves holes, a waterfall buys fill rate. Both Lusha and Hunter published guides in October 2026 that agree on this and on the cost: a waterfall raises coverage, not accuracy, and providers with looser standards quietly capture more of your list.
And if you are in Europe, the vendor choice is partly a compliance choice. Cognism and Dropcontact both foreground GDPR posture; the US-first incumbents do not.
How to read the match rates
The comparison uses vendor pricing, data documentation and terms checked on 16 October 2026. No trials or credit purchases were used to measure bounce or fill rates. Accuracy claims without a sample size, date and market should be treated as advertising.
Apollo’s per-seat rates and a separate Breeze Intelligence credit-pack price remain unconfirmed here.
What AI lead enrichment tools actually do (and how)
Enrichment is the process of appending fields a record is missing from external sources, then writing the result back into your CRM. Five distinct mechanisms hide inside that word, and they have very different failure modes.
Firmographic fill: the easiest and least interesting
Firmographics are company-level attributes: headcount band, revenue, industry, HQ country, ownership type. Lusha publishes its schema at lusha.com/data, updated 30 September 2026, and it is worth reading as a specimen of what a good data page looks like: every field carries a type, a refresh rate and a one-line statement of what it is for. Headcount, for example, refreshes weekly. Company phone refreshes monthly.
Firmographic data goes stale slowly, because companies change shape over quarters rather than days. It is also the least contested. Nobody argues about whether a company employs 51 people or 500.
Contact discovery: guessed patterns versus confirmed mailboxes
Contact discovery is where the category’s real accuracy problem lives. There are two fundamentally different methods, and they are not equivalent.
The first is pattern inference. Take a name and a company domain, generate first.last@company.com, confirm the domain has mail records, and return it. This is fast and cheap. It is also how you manufacture bounces.
The second is mailbox confirmation. Connect to the domain’s mail server and check whether the specific mailbox exists. Hunter documents its chain as: valid format, gibberish detection, disposable-address check, webmail check, MX record presence, SMTP server presence, SMTP check, accept-all domain check, and a separate proprietary accept-all verification step. Hunter states it does not send actual emails to the addresses it verifies.
That distinction is the whole game. Validation checks the address is correctly formed. Verification checks the mailbox exists. A pattern guess can be well-formed and still not exist. A verified mailbox can exist and still belong to someone who left the company four months ago.
Why a “valid” email can still be wrong
Three distinct failures produce a technically valid address that is commercially useless.
- Catch-all domains. A mail server configured to accept mail for any local part returns success for addresses that do not exist. No verifier can fully confirm the mailbox in that case. Hunter includes an accept-all check plus a follow-up verification step precisely because of this, and Dropcontact publishes separate rates for the general case and for catch-alls: 99% and 97% respectively.
- Role-based mailboxes.
info@andsales@are real, deliverable and useless for a named individual. They will not bounce, so bounce rate will not warn you. - Stale employment. The address works. The person left. Lusha argues this is where its waterfall earns its keep, and its measured data decay figure supports the argument.
The practical fix is not a better guess. It is recording where each field came from, running one verifier of your choosing over everything at the end, and deciding your catch-all policy once rather than inheriting whichever provider happened to label the row.
Technographic and stack signals
Technographics are the tools a company’s website reveals: the CMS, the analytics vendor, the payment processor, the email platform. Lead411 sells technology-stack search as a filter. UpLead advertises search across more than 16,000 technologies.
Signals are derived from technographic and public-web observation, so treat them as probabilistic. A company can use a technology on one property it does not operate everywhere.
Job-change and role-change detection
Job-change detection is the highest-leverage signal in the category, because it is the one failure mode that turns a working record into a dead one. It is also where the public dataset with a stated methodology cited here lives.
Lusha has run a data decay study twice against a fixed cohort of 148,000 records, tracking twelve- and twenty-four-month windows. The run dated 12 September 2026 found 12.25% of US sales leaders at VP and C-suite level had changed roles within twelve months, and 25.67% within twenty-four. The earlier run, dated 15 August 2026, found 12.6% and 25.7%.
That is a third of the “30% of your CRM data is outdated” figure that circulates in this category, and it comes with a cohort, a date and a re-run schedule attached.
12.25% of US sales leaders at VP and C-suite level changed roles within twelve months. 25.67% did within twenty-four. That is Lusha’s own published figure, measured against 148,000 records and re-run monthly since August 2026, and it is a third of the “30% is stale” number most vendors still quote without a method.
Lusha itself runs 26 dated buying signals covering funding, hiring, leadership changes and headcount shifts, refreshing most weekly and some monthly. It marks every email and phone with its last verification date, and returns a confidence tier per email from A1 down to A4.
Intent signals
Intent data attempts to rank accounts by observed research behaviour rather than by firmographic fit. ZoomInfo packages it as Buyer Intent and Streaming Intent, sold as an add-on priced by keyword volume and account universe. Lusha and lemlist sell it as a dated signal rather than a score. lemlist lists buying intent signals as a $29 per month add-on, or $21 billed annually.
Intent is the least standardised thing in the category and the hardest to verify. Treat any vendor’s “intent accuracy” number as untestable unless they publish a method.
Waterfall enrichment: how multi-provider cascades work
A waterfall enrichment chain sends the same missing field to several providers in a defined order and stops at the first confirmed result. Both Lusha and Hunter published guides on this in October 2026, and they describe identical mechanics, which is useful, because neither is quoting the other.
- Input arrives as an identity: a name plus a company domain, a LinkedIn URL, or a record ID.
- The first provider runs on every record.
- The candidate value is verified before it counts.
- On a miss, the record cascades to the next provider.
- On a match, the run stops for that field.
- Records no provider can fill are flagged unresolved, never filled with a guess.
Why a waterfall raises coverage but not accuracy
Both vendors land on the same conclusion, and it is the sentence most buyers need. A waterfall stops at the first answer it gets and does not check whether that answer is right. Inside a chain, the forgiving providers claim a bigger share of the list, because the records they “find” never reach the stricter tools behind them.
There is a second trap, which both vendors describe independently. Position in the queue decides the statistics. The first provider sees the whole list and resolves the easy records. A provider at position three only sees leftovers. Swap them and the numbers swap with them. So a waterfall find rate tells you almost nothing about provider quality unless you record the source on every field and judge providers by bounce rate instead.
Why a waterfall raises the bill
The economics turn entirely on what a miss costs. If a miss is free, adding a provider only costs you when that provider wins a record. If a vendor charges each source that returns data, and then charges a validator separately, a single contact can bill several times.
| Vendor | Own database first? | Stops at first match? | Charges on a miss? | Published waterfall rate |
|---|---|---|---|---|
| Lusha | Yes, 515M+ verified contacts | Yes | No | 2 credits per work email, 8 per phone, on match only |
| FullEnrich | No, states it has no enrichment database of its own | Yes | No, but catch-all emails are charged | 1 credit per work email, 10 per mobile |
| Apollo | Yes, then partner sources | No, runs until a validated result or every source is tried | Yes, each source that returns data charges; validator charges separately | Varies by source |
| Clay | No, data marketplace | Stops once validated data returns | Each step that returns data, and each validation step, charges | Varies by provider |
Table compiled from Lusha’s published vendor comparison, October 2026. Treat it as Lusha’s reading of competitors’ published terms, not as an independent audit. Apollo’s own help centre could not be retrieved to confirm its waterfall billing rule directly, so that row is single-sourced.
Lusha is the clearest example of the pay-on-match design: its docs state that filling 200 missing work emails costs 400 waterfall credits whether one provider answered or the request passed through every enabled one. Its waterfall also runs on Pro plans and above, with per-provider toggles and a source value on every result.
When the extra cost is worth it
A waterfall is worth it when coverage is the binding constraint, the list is big enough that a few percentage points means hundreds of records, and someone owns the setup. It is not worth it on a list of a few hundred contacts you could research by hand, and it is actively harmful if a bounce costs more than a gap: for example if you send from a single high-reputation domain.
On chain length: both Lusha and Hunter independently argue for a very short chain. Published figures cited by Lusha put the second provider at recovering 15% to 25% of what the first missed, a third at 8% to 12%, and a fourth at 3% to 5%, because many vendors license the same upstream data. Hunter puts the cap at two to three providers. Two competitors with different incentives converging on “two to four providers” is the strongest signal in this section.
Match rates: what is measured, and what is only claimed
This is the section the category is missing, so here it is stated plainly.
Lusha’s published sample sizes
Lusha publishes four relevant measurements on its own site.
- A hit-rate test. Two published tests run on 28 July 2026 on 322 US contacts returned a usable work email for all 322 and a mobile number for 97.5%, ranging between 93.9% and 100% by segment. That is a real number with a real sample size, and it is also a small one. Three hundred and twenty-two contacts is roughly one trading day of outbound for a mid-sized team. It is directionally useful and statistically thin, and you should hold it that way.
- A decay study. 148,000 records, two runs, dated.
- A competitive test. On 11 August 2026, across five US ICPs, Lusha reports returning twice as many US contacts with a direct phone number as ZoomInfo, with full disclosure that ZoomInfo led on work email fill rate in the same test and that accuracy was not tested.
- A waterfall outcome. In September 2026, Lusha reports its waterfall filling 1 in 2 missing work emails across all waterfall accounts, and on one 1,103-contact customer list moving email coverage from 90% to 97%.
All four are self-published by a vendor selling the product. They are still better than nothing, because they carry a date and a denominator.
Everything else is a claim
| Vendor | Figure | What is missing |
|---|---|---|
| Lusha | 98% email accuracy; 85% US and 86% global phone accuracy | Method and sample size for the headline figure; last updated 30 September 2026 |
| Hunter | Bounce rate below 1% on verified addresses | Measurement window and list size |
| RocketReach | 98% accuracy, verified on demand in real time | Sample, date, market |
| UpLead | 95% data accuracy rate | Sample, date, market |
| Dropcontact | 99% verified, 97% for catch-alls | Sample size, date |
| Seamless.AI | 100% credit back if an email is invalid | Not an accuracy measure; a refund policy |
All figures read from each vendor’s own page on 16 October 2026. Every one is a vendor claim. None of them states a sample size or a market alongside the percentage, which is the specific problem.
The published match-rate ranges, properly labelled
There is a set of widely repeated match-rate figures that nobody in this category owns. Lusha cites them as published industry estimates, not as its own measurements, in its October 2026 waterfall guide. I am repeating them with that same label attached, and you should treat them as directional rather than precise.
| Architecture | How it runs | Published match rate | Main trade-off |
|---|---|---|---|
| Single source | One database | 55% to 70% | Simple, one set of terms, hard coverage ceiling |
| Parallel | All providers at once, best value kept | 75% to 90% | Cross-checks each field, pays every provider |
| Sequential waterfall behind a first-party database | Own data first, then providers in order | 80% to 92% | Coverage at controlled cost |
| Managed waterfall, no own database | Vendor cascades across third parties only | 80%+ claimed | Every answer is a third party’s answer |
Published industry figures cited by Lusha, October 2026. Not measured results. No named study.
No independent match-rate benchmark is cited here. The published figures come from vendors. If a vendor quotes you a 95% match rate, the only way to know what it means is to run their free tier against your own list.
The tools compared
| Tool | Category | What you are buying | Price transparency on 16 Oct 2026 | Best fit |
|---|---|---|---|---|
| Hunter.io | Email specialist | Emails found and verified, plus sequences | Full, self-serve | Verified work email, lowest published unit cost |
| FullEnrich | Waterfall | 25+ sources cascaded for email and phone | Full, self-serve | Filling sourced lists |
| Lusha | Browser plus API | Credits for reveals, signals and waterfall on Pro+ | Full, self-serve | Rep-level lookups and signals |
| UpLead | Database plus guarantee | 1 credit per contact with email and mobile | Full, self-serve | Small teams wanting email and phone together |
| Clay | Orchestration | Actions plus Data Credits from a 150+ provider marketplace | Full, self-serve | Multi-provider waterfalls you build yourself |
| Apollo.io | All-in-one | Seats plus a pooled credit pool, sourcing and sequencing | Tier names public, per-seat figures not retrievable | Source, enrich and sequence in one tool |
| Regie.ai | Research plus outreach | Credits across research, enrichment, drafting and dialling | Full, self-serve | Agents that research and write as well as enrich |
| Instantly | Sequencer with enrichment | Credits including waterfall work email across 5+ providers | Full, self-serve | Outreach-first teams wanting enrichment inline |
| lemlist | Outreach platform | Seats plus optional enrichment and intent add-ons | Full, self-serve | Multichannel outbound |
| Lead411 | Database | Exports per month | Full, self-serve | Budget US B2B search with rollover exports |
| Dropcontact | EU-first finder | Credits, pay on success, re-credited on a miss | Full, self-serve | GDPR-sensitive EU outreach |
| People Data Labs | Data API | Records retrieved, one credit per successful request | Full, self-serve | Engineering-led enrichment you wire yourself |
| RocketReach | Database | Credits, pooled, refunded on a miss | Floor only, from $6,000 annually | Mid-market teams wanting volume |
| Seamless.AI | Real-time search | Credits, free to 50 then sales-led | Free tier only, then quote | Net-new lead discovery at volume |
| Cognism | EU database | Credits, one credit reveals one contact | Quote-only | EU and UK coverage and phone verification |
| ZoomInfo | Enterprise platform | Credits, one per exported contact or company profile | Quote-only | Enterprise intent and inbound capture |
| HubSpot Breeze Intelligence | CRM-native | Credit packs on a paid HubSpot tier | Not confirmed on HubSpot’s own pages | Inbound form enrichment inside HubSpot |
| Kaspr | ( | ) | kaspr.com no longer resolves to a product site | Not a buying option |
What these tools cost, checked 16 October 2026
| Tool | Entry price | What that buys | Credit rules |
|---|---|---|---|
| Hunter.io | $0 | 50 credits, 100 verifications | Free verifications capped at 100 a month |
| Hunter.io Starter | $49/mo, $34/mo annual | 2,000 credits/mo (24,000/yr) | Credits reset monthly, no rollover |
| Hunter.io Growth | $149/mo, $104/mo annual | 10,000 credits/mo | 1 credit per email found, 0.5 per verification |
| Hunter.io Scale | $299/mo, $209/mo annual | 25,000 credits/mo | Unlimited seats on every paid plan |
| FullEnrich | $0.055 per credit at 1,000 credits | 1,000 work emails or 100 mobiles | Misses are not charged; catch-alls are |
| FullEnrich Pro | $720/mo for 15,000 credits | 15,000 credits at about $0.048 | Credits roll over on paid plans |
| Lusha Free | $0 | 40 credits/mo, 1 seat | Resets monthly, no rollover |
| Lusha Starter | $37.45/mo annual | 4,800 credits/yr | 1 credit per email, 5 per phone |
| Lusha Pro | $52.45/mo annual | 7,200 credits/yr | Adds API, CSV, signals, webhooks |
| Lusha Premium | $299.95/mo annual | 40,800 credits/yr | Adds waterfall, 5 seats, unlimited intent results |
| UpLead Essentials | $99/mo or $74/mo annual | 170 or 2,040 credits | 1 credit unlocks email plus mobile direct dial |
| UpLead Plus | $199/mo or $149/mo annual | 400 or 4,800 credits | Enrichment included on Plus and above |
| Lead411 Spark | $49/mo or $490/yr | 1,000 exports/mo | Unused exports roll over |
| Lead411 Ignite | From $150/mo or $1,500/yr | 1,000+ exports/mo | Adds API access |
| RocketReach | From $6,000 annually | Pooled credits | Lookup refunded if nothing verified is found |
| Seamless.AI | Free to 50 credits | Then sales-led | Credit refunded if an email is invalid |
| People Data Labs Pro | From $98/mo | 350 person enrichment records | 20% off annually |
| Dropcontact Starter | EUR 79/mo | Email Finder credits | Re-credited when no email is found |
| Regie.ai Pro | $41/mo, $49 list | 5,000 credits/mo | Enrichment that finds nothing is free |
| Regie.ai Team | $125/seat/mo, $150 list | 15,000 credits per seat | One shared pool, two-seat minimum |
| Instantly Growth | $47/mo | 1,500 credits, waterfall across 5+ providers | Credits, not contacts, are the unit |
| lemlist Email | $69/mo, $55/mo annual | Seats plus 650M+ lead database | Unlimited users |
| lemlist add-ons | $20/mo enrichment, $29/mo intent | Both cheaper annually ($16 and $21) | Per user |
| LinkedIn Sales Navigator Core | $119.99/mo or $1,079.88/yr | 50+ filters, 50 InMail credits/month | Annual saves 25% on Core |
| LinkedIn Sales Navigator Advanced | $159.99/mo or $1,799.88/yr | Adds content-engagement alerts, TeamLink | Annual saves 6% on Advanced |
| Clay Launch | From $185/mo | 2,500 Data Credits, 15,000 Actions | Data Credits from $0.05, Actions under $0.01 |
| Clay Growth | From $495/mo | 6,000 Data Credits, 40,000 Actions | Bring your own keys and skip Data Credits |
| ZoomInfo | Quote only | 1 credit per exported contact or company profile | Marketing Demand alone includes 75,000 credits |
| Cognism | Quote only | Standard and Pro, 5 seats included | One credit reveals one contact |
Where the sources disagree
Three conflicts worth knowing about before you budget.
Clay contradicts itself. Its pricing page plan cards show Launch at $167/mo and Growth at $446/mo, while the same page’s FAQ text says Launch starts at $185/mo and Growth at $495/mo. The difference is most likely annual versus monthly billing, but the page does not say so. Treat $185 and $495 as the FAQ figures and confirm before signing.
Nobody agrees on Breeze Intelligence. Clay’s July 2026 guide reports credit packs from $45 a month for 100 credits. Cognism’s June 2026 article reports packs of 100, 1,000 and 10,000 credits ranging from $30 to $700 a month. HubSpot’s own Data Hub pricing page lists contact enrichment and company enrichment from the Starter tier at $7 per seat per month billed annually with 500 HubSpot Credits included, and does not show a separate Breeze Intelligence pack price. Get it from HubSpot.
Provider counts differ too. Clay’s pricing page says 150+ data partners; Clay’s own July 2026 guide says 200+ providers.
The cost arithmetic for a five-person team
Effective cost per usable row equals total monthly spend divided by the number of contacts that came back with a usable work email and a way to reach them. That second number is not the same as the number you enriched, because fill rate is the multiplier nobody budgets for.
Take a concrete case: five SDRs, 2,000 new contacts a month from inbound forms and sourced lists. Assume a 70% single-source fill rate, which sits in the middle of the published 55% to 70% band cited above. That yields about 1,400 usable rows.
| Plan | Credits needed for 2,000 emails | Credits needed for 2,000 emails plus dials | Monthly cost | Cost per usable email row |
|---|---|---|---|---|
| Hunter Growth, email plus verification | 3,000 | Not offered | $149 | $0.106 |
| Lusha Premium, email only | 2,000 of 3,400 | 12,000 needed, plan covers 3,400 | $299.95 | $0.214 |
| FullEnrich Pro, email only | 2,000 of 15,000 | 22,000 needed, plan covers 15,000 | $720 | $0.514 |
| UpLead Plus annual, email plus mobile | 2,000 of 4,800 | 2,000 of 4,800 | $149 | $0.106 |
Three conclusions fall straight out of that table.
- Underbuying is cheap; overbuying is expensive. FullEnrich Pro costs $720 a month for 15,000 credits. If you use 2,000 of them, you have spent $0.36 per record when the marginal data cost was $0.048. Plan sizing dominates vendor choice.
- Annual commitment changes the answer by an order of magnitude. UpLead Plus is $199 a month for 400 credits, which is $0.50 per contact, or $149 a month for 4,800 credits on annual billing, which is $0.031. Same product, sixteen times different.
- Phones are where budgets break. Adding a direct dial takes Lusha from 1 credit to 6 and FullEnrich from 1 to 11. A team that dials should model phone coverage separately and expect to buy a larger plan or a different tool.
How to run an enrichment pilot that produces evidence
- Pick 200 to 300 contacts from your actual target segment. Not a vendor demo list. Lusha makes this point explicitly in its accuracy FAQ: a provider strong in US mid-market tech can be thin in European manufacturing, and only your own segment tells you which you are dealing with.
- Strip the data you already have. You are measuring fill rate on the fields that are genuinely missing, not on fields the vendor can echo back.
- Run every shortlisted tool on the identical list. Same records, same fields, same week.
- Count three things separately. Fill rate: does the field come back at all. Accuracy: is it right, checked by sending or by a person. Freshness: how many are for a role the person has left. Lusha’s decay data implies you should expect roughly one in eight senior contacts to be stale after a year.
- Separate direct dials and mobiles from switchboard numbers. A vendor can report healthy phone coverage made almost entirely of main company lines. Ask for the split.
- Record the source of every returned field. Without this you cannot judge providers later, and you cannot answer a data subject request.
- Check the billing rule on a miss before you run volume. Whether you are charged for a cascade, for catch-alls, and for a failed lookup changes the cost per row by more than any list-price difference.
- Re-run the same list after 90 days. Nothing tests a vendor’s refresh cadence like a second pass over a list you already enriched.
What breaks: failure modes and gotchas
- The coverage trap. A waterfall that reports 92% fill rate has told you nothing about whether those addresses work. You bought rows, not replies.
- Status collapse. One provider calls an accept-all address risky, another calls it valid with a footnote, a third says nothing. Merge them into one CRM column and the word “verified” stops carrying information. Run one verifier of your choice at the end.
- Silent overwrites. Check whether the tool will overwrite a field a rep corrected by hand. Cognism explicitly sells field-level controls and locked fields as a feature, which tells you the problem is real elsewhere.
- Credit expiry. Apollo’s pricing page states unused credits expire at the end of each billing cycle and do not roll over, including purchased add-ons. Hunter’s page says the same for monthly plans, with all credits granted up front on annual. Lusha allows rollover up to twice the monthly allowance on monthly plans only. Read the rollover rule before you buy annually.
- The bounce that costs you more than the gap. One hard bounce from a pattern guess can drag a sending reputation with it. On a single high-value domain, that is a worse trade than an empty field.
- Provenance you cannot produce. If a European contact asks where their email came from and you cannot answer, you have a compliance problem, not a data problem.
- Seat minimums. Cognism’s Standard and Pro both bundle five seats. Regie.ai’s Team plan requires two. For a solo operator that is dead money.
- Credit vs contact confusion. Apollo meters credits pooled across the whole team. ZoomInfo charges one credit per exported contact or company profile. UpLead’s one credit unlocks email plus mobile together. FullEnrich charges 10 credits for what Lusha charges 5 for. The word “unit” is doing a lot of work in this category and means something different at every vendor.
Law, consent and platform terms
This section is not legal advice and it is not a substitute for advice from your own counsel. I am describing published regulator guidance and statute text, not your obligations. Get a lawyer or a qualified DPO to sign off on your specific campaign before you send.
UK and EU: business contact data is personal data
The most common misconception in this category is that “B2B data” is exempt. It is not. The UK Information Commissioner’s Office states that if you can identify an individual directly or indirectly, it constitutes personal data “even if they are acting in their business capacity”, and lists holding a business contact’s name and number as processing personal data.
Where it gets confusing is PECR. The ICO’s table is clear: the PECR email-marketing rules do not apply to corporate subscribers, so you do not need consent to email a limited company’s employee. But sole traders and certain unincorporated partnerships are treated as individual subscribers, and there you need consent or a valid soft opt-in. Automated calls always need consent.
The lawful basis question is separate and still live. The ICO sets out the three-part legitimate interest test: identify a legitimate interest, show the processing is necessary to achieve it, and balance it against the individual’s interests, rights and freedoms. Two further obligations bite enrichment specifically:
- Right to be informed. If you collected contact details from a third party or a public source, you must provide privacy information within a reasonable period and no later than one month from collection.
- Right to object. Under the UK GDPR this right is absolute for direct marketing. There is no balancing test to beat it once someone objects.
The ICO also notes that a waterfalled list multiplies your answers. With one provider you can describe one methodology. With five you have five, and one opaque source leaves its share unaccounted for. That is exactly the failure mode both Lusha and Hunter describe in their waterfall guides.
United States: three different regimes
California. The CCPA, as amended by the CPRA, applies to for-profit businesses doing business in California with gross annual revenue over $25 million, or that buy, sell or share the personal information of 100,000 or more California residents or households, or that derive 50% or more of annual revenue from selling or sharing personal information. Personal information excludes information from government records and certain publicly available sources. Private enforcement is limited: you can only sue over certain data breaches, with statutory damages up to $750 per incident. Note that the exemptions for employee data and for B2B transactional information expired on 31 December 2022, so do not assume a B2B carve-out survives.
On automated decision-making: the California Privacy Protection Agency’s own rulemaking page states that its board adopted regulations on 24 July 2025 implementing consumers’ rights to access and opt out of businesses’ use of automated decisionmaking technology, that the Office of Administrative Law approved them on 22 September 2025, and that the effective date is 1 January 2026. Any litigation-related delay or stay remains unconfirmed here; check the current docket before relying on the effective date.
Texas. The Texas Data Privacy and Security Act, Business and Commerce Code Chapter 541, took effect 1 July 2024. The definition that matters most here is narrow: “consumer” means an individual who is a resident of Texas “acting only in an individual or household context”, and the term expressly does not include an individual acting in a commercial or employment context. The chapter applies to persons that conduct business in Texas or produce a product or service consumed there, process or engage in the sale of personal data, and are not an SBA-defined small business, except that a small business may not sell sensitive data without prior consumer consent.
Email generally. CAN-SPAM applies to commercial email and makes no exception for B2B. Penalties are up to $53,088 per violating message. Opt-outs must be honoured within 10 business days and the mechanism must work for at least 30 days. Critically for this category, the FTC lists harvesting email addresses or generating them through a dictionary attack among the aggravated violations that can carry criminal penalties. Pattern-generating enrichers sit uncomfortably close to that line, which is another reason to verify rather than guess.
The LinkedIn trap
This is the single most expensive mistake in the category, and it is contractual rather than criminal.
LinkedIn’s User Agreement, effective 3 November 2025, says in section 8.2 that you agree not to:
- “Develop, support or use software, devices, scripts, robots or any other means or processes (such as crawlers, browser plugins and add-ons or any other technology) to scrape or copy the Services, including profiles and other data from the Services”.
- “Copy, use, display or distribute any information (including content) obtained from the Services, whether directly or through third parties (such as search tools or data aggregators or brokers), without the consent of the content owner”.
- “Rent, lease, loan, trade, sell/re-sell or otherwise monetize the Services or related data or access to the same, without LinkedIn’s consent”.
That second clause is the one that catches agencies. If a data broker sold you records it obtained through a LinkedIn-derived source, the reseller prohibition in section 8.2 may already have been breached upstream, and your own redistribution of those records can compound it. Ask vendors to state their sourcing methodology in writing, and prefer vendors that publish it. Lusha states it never scrapes professional networks; Hunter states it tells you how every email address was found.
Related trap: reselling your own enrichment. Apollo’s pricing page states that the plans are for internal business use only, and that using Apollo data to power external products, share with customers or resell is not permitted under standard terms without a separate agreement. If you run an agency and enrich on behalf of clients, check the terms of every tool in the stack, not just the one you think is the data source.
The bottom line
Start with the mechanism, not the shortlist. Decide whether you need a verified email, an email plus a direct dial, or firmographics and intent, because those three needs have prices that differ by an order of magnitude.
Then buy the smallest published plan that covers your actual consumption, commit annually only if you are confident you will use the credits, and run 200 to 300 of your own contacts through every candidate before you sign anything. Lusha’s free tier gives 40 credits a month with no card, enough for 30 to 40 contacts; Hunter’s free plan allows 100 verifications a month. Neither is enough for a real test, but both prove the workflow.
Finally, get the lawful basis documented before the first send, and check that every provider in your chain can tell you where its records came from. The vendor that cannot answer that question will not be the one that saves you when a contact asks.
FAQ
What is the cheapest AI lead enrichment tool in 2026?
On published prices checked 16 October 2026, Hunter.io is cheapest per verified work email: $149 a month for 10,000 credits on Growth, where one credit finds an email and 0.5 verifies one, or roughly $0.022 per verified address if you consume the full plan. FullEnrich’s work email is $0.048 per credit on its 15,000-credit $720 Pro plan. Lusha’s Starter plan is $37.45 a month billed yearly for 4,800 credits, about $0.0078 a credit. Cheapest per credit is not cheapest per usable sales-qualified row, because plan size and fill rate dominate.
What is a realistic match rate for lead enrichment?
It depends on the segment and where the list came from. Vendor-citing “published industry estimates” put a single provider at 55% to 70% of a typical B2B contact list, parallel multi-provider lookups at 75% to 90%, and a sequential waterfall behind a first-party database at 80% to 92%. Those are not measured results and carry no sample size. The figure cited here with a stated sample size is Lusha’s: a test run on 28 July 2026 across 322 US contacts returned a usable work email for all 322 and a mobile number for 97.5%.
How does waterfall enrichment work?
A waterfall sends the same missing field, usually a work email or mobile number, to providers in a set order and stops at the first confirmed result. Both Lusha and Hunter describe identical mechanics in their October 2026 guides, and both state the same limitation: it raises coverage, not accuracy, because it stops at the first answer without checking whether that answer is right. Providers with looser standards therefore claim a larger share of the list, and you meet their guesses later as bounces.
Why does a verified email still bounce?
Usually a catch-all domain. When a mail server accepts mail for any local part, an SMTP probe returns success for addresses that do not exist. Hunter’s documented verification chain includes an accept-all domain check plus a separate proprietary accept-all verification step, and Dropcontact publishes separate rates for the general case and catch-alls: 99% and 97%. A role-based mailbox such as info@ can also be deliverable but useless for a named individual, and it will never show up as a bounce.
Is business contact data personal data under GDPR?
Yes. The UK Information Commissioner’s Office states that if you can identify an individual directly or indirectly it constitutes personal data “even if they are acting in their business capacity”. PECR’s email rules do not apply to corporate subscribers, so consent is not required for that step, but the UK GDPR applies in full, including the absolute right to object to direct marketing and the requirement to send privacy information within one month when data came from a third party or public source.
Can I legally scrape LinkedIn profiles for lead generation?
LinkedIn’s User Agreement, effective 3 November 2025, prohibits developing or using software, scripts, robots or other means to scrape or copy the Services, and prohibits copying or distributing information obtained from the Services including through “search tools or data aggregators or brokers”. It also prohibits reselling or monetising the Services or related data. Those are contract terms, separate from criminal or copyright law, and LinkedIn enforces them.
How do credits get consumed when nothing is found?
It varies enormously, and this is where published prices stop being comparable. Lusha charges nothing when a lookup finds no match, and its waterfall charges 2 credits per work email and 8 per phone only on a provider match. Regie.ai states enrichment costs nothing when it finds nothing. RocketReach refunds a lookup when no verified email or phone is returned. Apollo’s waterfall, according to Lusha’s published comparison, charges each source that returns data and charges a validator separately, so one contact can bill several times.
How fast does B2B contact data go stale?
Faster than the “30% a year” figure most vendors repeat, but not as fast as that figure implies either. Lusha measured role change against a fixed cohort of 148,000 records twice in 2026. The run dated 12 September 2026 found 12.25% of US sales leaders at VP and C-suite level had changed roles within twelve months and 25.67% within twenty-four. The 15 August 2026 run found 12.6% and 25.7%. That is roughly 1% of senior sales contacts per month.
Is HubSpot Breeze Intelligence still sold as credit packs?
A separate credit-pack price remains unconfirmed. HubSpot’s Data Hub pricing page lists contact enrichment and company enrichment from the Starter tier at $7 per seat per month billed annually with 500 HubSpot Credits included. Two third-party guides disagree, reporting $45 a month for 100 credits and packs of 100 to 10,000 credits at $30 to $700 a month respectively. Get the number from HubSpot before budgeting it.
Which enrichment tool should a small team buy first?
Buy the one whose free tier lets you test your own segment rather than a vendor demo list. Lusha’s free plan gives 40 credits a month with no card, which its own documentation says covers 30 to 40 contacts, and Hunter’s free plan allows 100 verifications a month. Run 200 to 300 of your real target contacts, count fill rate, accuracy and freshness separately, then size the paid plan to what you actually consumed rather than to a headline rate.
Sources
- Lusha pricing: entry tiers, credit costs per reveal, rollover rules
- Lusha data page: database size, field-level schema, refresh rates, accuracy figures, last updated 30 September 2026
- Lusha: B2B Data Accuracy FAQ: verification methods, 28 July 2026 hit-rate test on 322 US contacts, 11 August 2026 test against ZoomInfo, decay figures
- Lusha: Waterfall Enrichment: waterfall mechanics, vendor billing comparison, industry match-rate ranges, provider-count curve
- Lusha: How Lusha credits work: October 2026 credit rates including waterfall rates and rollover rules
- Hunter.io pricing: plan tiers, annual and monthly prices, credit-per-action rules
- Hunter Email Verifier: documented verification chain, accept-all handling, no-send policy
- Hunter: Waterfall Enrichment: independent corroboration of waterfall mechanics and the queue-position measurement trap
- FullEnrich pricing: $0.055 per credit, per-field credit costs, 25+ sources, 15,000-credit Pro plan at $720
- UpLead pricing: Essentials and Plus monthly and annual tiers, one-credit-per-contact model
- Clay pricing: Actions and Data Credits model, tier details, and the $167/$446 versus $185/$495 internal discrepancy
- Clay: 10 Best Lead Enrichment Tools for 2026: competitive landscape, G2 ratings, its own Breeze Intelligence price claim
- Apollo.io pricing: credit pooling and rollover policy, tier names, record selection limits, internal-use and no-resale terms
- LinkedIn Sales Navigator pricing: Core and Advanced prices, annual discounts, page updated 27 September 2026
- LinkedIn User Agreement: section 8.2 scraping, redistribution and resale prohibitions, effective 3 November 2025
- ICO: Business-to-business marketing: business contact data as personal data, PECR corporate subscriber rules, legitimate interest three-part test, right to object
- ICO: Direct marketing hub: index including data broking guidance
- FTC: CAN-SPAM Act compliance guide: $53,088 per-message penalty, 10-business-day opt-out, B2B applicability, dictionary-attack criminal exposure
- FTC: CAN-SPAM Rule: regulatory scope
- California AG: CCPA: thresholds, exemptions expired 31 December 2022, $750 per-incident damages, data broker definition, page updated 28 August 2026
- California Privacy Protection Agency: CCPA Updates: ADMT, risk assessment and cybersecurity audit regulations adopted 24 July 2025, effective 1 January 2026
- Texas Business and Commerce Code Chapter 541: consumer definition excluding commercial context, applicability thresholds, small-business consent rule, effective 1 July 2024
- Dropcontact Email Finder: 99% verification and 97% catch-all figures, algorithm-only approach
- Dropcontact pricing: Starter at EUR 79/mo, Growth at EUR 120/mo, re-crediting on a miss
- Seamless.AI: real-time search model, free tier to 50 credits, credit-back policy, sales-led pricing
- RocketReach pricing: $6,000 annual floor, pooled credits, refund on unverified lookups
- People Data Labs pricing: Pro from $98/mo for 350 records, 20% annual discount
- Cognism pricing: Standard and Pro tiers, 5 seats included, quote-only
- Cognism: Contact Data Enrichment: their enrichment workflow and the conflicting Breeze Intelligence credit-pack claim
- Regie.ai pricing: Pro and Team tiers, credit pools, free-on-miss enrichment
- Instantly pricing: Growth tier at $47/mo, waterfall work email across 5+ providers
- lemlist pricing: Email and Multichannel tiers, enrichment and intent add-ons
- Lead411 pricing: Spark and Ignite tiers, rollover exports, re-verification every 3 to 6 months
- HubSpot Data Hub pricing: Starter at $7 per seat annually, Professional at $720/mo, enrichment listed from Starter
- Kaspr: checked 16 October 2026, resolves to a GoDaddy domain-for-sale parking page
