The hidden cost of bad LinkedIn data: enrichment vs scraping

Bad LinkedIn data costs you in three places: messages built on a job the person no longer holds, bounced emails that hurt your sender reputation, and rep hours spent on people who will never answer. Roughly 12 to 20% of professionals change employer each year, and vendor estimates put contact record decay at 22 to 30% a year. Scraped databases start ageing the day they are collected, and they now carry legal risk. Data pulled at the moment someone engages is fresher and easier to defend.

Here is the situation most reps will recognise. Your enrichment tool shows "Senior Director of Demand Gen, ScaleUp Inc, joined March 2024." It looks perfect. You send a DM and get no reply. The reason is that they left in November, and the record you read was cached months earlier. Your message referenced a job they no longer had.

Enrichment vs scraping: which data source should you trust?

How fast does LinkedIn data go stale in 2026?

Faster than most list buyers assume, but slower than some vendor marketing claims. These are the figures you can actually trace:

• Job changes. Lusha measured its own database and found 12.6% of US sales leaders changed roles over twelve months and 25.7% over twenty-four. UK sales leaders changed at 1.73 times the US rate. Cleanlist puts annual switching at 15 to 20% of professionals. Inbound Labs says the average professional changes roles every 2.5 to 3 years. Our earlier "about 10%" was too low.
• Whole-record decay. MarketingSherpa found B2B contact data decaying at roughly 2.1% a month. HubSpot's decay simulation annualises that to 22.5% a year. Firmable points out that these are one finding, not two independent ones. Landbase quotes a range of 22.5% to 70.3% a year and 28% a year for email lists.
• The 30% figure. It gets cited constantly. Lusha says it has no traceable source. Lusha sells data, so read that with some care, but we could not trace it either.
• Phones. Landbase puts direct dial decay at 15 to 25% a year and mobile numbers at 10 to 15%.

Job changes are also a selling opportunity. Our job change trigger playbook covers how to use them on purpose.

Is a record that is 60 days old unreliable?

We used to say that anything scraped more than 60 days ago is unreliable. That was our rule of thumb, not a sourced number, and it overstated the case. At 1 to 2% decay a month, a 60-day-old record is roughly 2 to 4% stale. That is not where the damage comes from.

The damage comes from age you cannot see. Bulk lists rarely tell you when each record was last checked. Many were collected six to twenty-four months ago, and at that age the job-change figures above start to bite hard. The question to ask about any record is "when was this last confirmed?", not "does it exist?"

What does stale data cost an outbound team?

Here is a worked example. It is an illustration with stated assumptions, not a measured result.

Assume 200 outbound touches a week from a list that is about 12 months old.

• Wrong context. At 12.6 to 20% annual job change, 25 to 40 of those 200 people are no longer in the role your message refers to. If the list is two years old, Lusha's 25.7% figure suggests around 50.
• Bounces. If email decay runs near 28% a year, a large share of addresses will need replacing. Every hard bounce counts against your domain. Our bounce rate guide and 2026 deliverability guide explain the thresholds that get you filtered.
• Dead dials. At 15 to 25% direct dial decay, a meaningful slice of call blocks goes to numbers that ring nowhere.

The touches overlap, so you cannot just add the percentages together. Even so, a team on year-old data can expect a quarter or more of its effort to land on the wrong person, the wrong inbox or the wrong number. The deeper cost is reputation. Bounces drag down placement for every future email, including the ones sent to good contacts.

Is scraping LinkedIn legal in 2026?

This is the biggest change since we first published this page. LinkedIn is taking scrapers to court and winning.

• Proxycurl. On January 24, 2025, LinkedIn sued Nubela, the company behind Proxycurl, in the Northern District of California. It alleged that Nubela created hundreds of thousands of fake accounts to scrape millions of profiles, including non-public data. Proxycurl shut down permanently on July 4, 2025 and was required to delete the LinkedIn data it had obtained.
• ProAPIs (iScraper). LinkedIn's October 2, 2025 complaint alleged that ProAPIs sold real-time access to member data through an API handling up to 150 requests per second, at up to $15,000 a month. On September 16, 2026, Judge P. Casey Pitts signed a final judgment on consent in LinkedIn Corp. v. ProAPIs Inc., No. 5:25-cv-08393. It orders the company to stop scraping, stop selling scraped data, stop using fraudulent accounts, and delete what it took. The data allegedly included posts, reactions and comments.
• Buyers are not automatically safe. A summary of the LinkedIn User Agreement effective November 3, 2025 says it covers information obtained through third parties such as data aggregators or brokers. That is a secondary reading, so check the agreement itself with your counsel.
• Europe. Lobstr reports that in March 2026 the CNIL closed its case on Kaspr after Kaspr deleted 160 million contacts and stopped collecting from LinkedIn. Verify that against the CNIL's own decision before you rely on it.

The lesson is that "real-time" alone is not a defence. Proxycurl sold real-time data too. What matters is how the data is collected.

Why are cheap lead lists so expensive?

Lists sold by the million at a low flat price usually share three problems. They were collected long ago. They are rarely refreshed. And they are sold to many buyers, so the same people get the same cold pitch from everyone. Add the legal question above and you are paying with your domain reputation, your reps' hours and your exposure.

Where Saava fits

Saava works the other way round. Instead of storing a big list and hoping it is still accurate, it watches LinkedIn posts you choose, including competitor and creator posts, and pulls a person's profile when they engage. The profile reflects that moment, and the engagement itself is the reason to reach out. We explain the method in turning post engagement into leads.

Each person is scored against your ICP, so you see who fits before you spend a touch. Outreach goes from your own connected LinkedIn account or email, within sensible limits (see our LinkedIn limits and warmup guide). If you are weighing us against a large static database, our Saava vs ZoomInfo comparison sets out the trade-offs honestly. Coverage is narrower by design, because you only get people showing a signal.

How do you check a data vendor before buying?

1. Ask when each record was last confirmed, and whether you can see that date per contact.
2. Ask how profile data is collected, in writing. Fake accounts and resold scrapes are the red flags.
3. Run a test batch through an email verifier before importing anything.
4. Track hard bounce rate and wrong-person replies for the first two weeks.
5. Compare against signal-led leads. Our piece on first-party vs third-party intent data helps frame that test.

FAQ

How often should I refresh contact data? Refresh before every campaign, and always before emailing anything older than a few months. At around 2% decay a month, a quarterly list has lost about 6% before you send.

Is the "30% of B2B data decays each year" statistic real? We could not trace it to a primary source. The best-documented figure is MarketingSherpa's 2.1% a month, which works out to about 22.5% a year.

If a vendor promises real-time LinkedIn data, am I safe? Not on that promise alone. Proxycurl and ProAPIs both sold real-time access. Ask how the data is collected.

Does speed matter once I have a fresh signal? Yes. Interest fades fast. See our guide to speed to lead on buying signals.

Should I stop using databases entirely? Not necessarily. Use them for coverage, verify before sending, and save your best effort for people showing intent now.

Want leads that are current the day you contact them? Start with Saava.

If you want this running for you instead of by hand, start a trial