BlitzAPI's New Plans, Priced Against 350,897 Records From Our Own Logs

As of October 2026, BlitzAPI sells Start at $99 a month for 200,000 records and Scale at $499 for 5,000,000. We priced both against 22 days of our own spend logs, 28 September to 20 October, in which our account drew 350,897 records.

Three of those days, 4 to 6 October, drew 183,613, which is 91.8% of Start's monthly allowance. At least 110,504 of that was logged evaluation. Logged client work came to 75,397 records over the same 22 days, about 38% of Start. And 164,406 records, 46.9% of the total, aren't tied to any log we kept. Nearly half.

Our account has run on the legacy $399 plan since late September: 15,000,000 records a month, no email or phone, 50 requests a second. The $99 Start plan looks very interesting and is under consideration. The fit figures below apply the published plan rules to our usage; our own bill is a separate matter.

BlitzAPI pricing as of October 2026

Blitz now sells four plans. Start and Scale are self-serve, billed monthly and cancellable any time. Build Your Infra starts at $999 a month and is scoped on a call. Embedded, for software where your own users trigger Blitz features, starts at $2,000 a month on an annual agreement. Every plan opens every endpoint, verified work emails and US phone numbers included, so the plans differ by speed, monthly allowance and support.

The per-1,000 row is plain division: price over allowance, times 1,000. It's a floor, reached only by using every record. On that basis Start costs 18.6 times what our legacy plan did per 1,000 records. But the legacy figure buys no email or phone, so the two aren't like for like. Start costs 75.2% less a month than our legacy plan and carries 98.7% less allowance. Scale gives 2.5 times Start's speed and 25 times its allowance for about five times the price.

A record is one item the API returns: one company, one person, one email address, one phone number or one job posting. A call that finds nothing counts nothing, and a people search that returns 25 profiles counts 25. That makes Blitz a flat monthly fee with a fixed allowance, unlike credit-based pricing, where the bill grows with every lookup.

And speed is a rate limit per endpoint, so a batch of email lookups doesn't eat into the budget for company searches.

Two terms matter more than the price. When an account reaches its allowance, calls pause until the next billing cycle or until it moves up a plan, and unused records don't carry over. Section 9.1 of the Terms holds the limits. But Start and Scale carry no SLA and no signed DPA. The Terms incorporate Blitz's published DPA by reference on every plan; a DPA signed with your company comes with Build Your Infra.

And our BlitzAPI tool page covers the endpoints themselves.

When the plans changed, and what Blitz has said

Our own notes from 28 September show the old structure. When we read the pricing page again on 22 October it showed Start, Scale and Build Your Infra. The go-live date sits somewhere in between, and we couldn't pin it.

The vendor's pages bracket it without settling it. The Terms say the launch offer closed at 11:59 pm Eastern on 13 October and carry a "last updated" date of 14 October, while a changelog entry of 2 September gives a different cutoff, plans created before 30 September. None of the changelog's 47 entries, from 5 August to 26 October, announces the new plans. Blitz hasn't said why it changed them.

If you subscribed under the launch offer, the pricing FAQ says you keep your price, speed and allowance while the subscription stays active. Read section 10.7 of the Terms before you change anything. It says those terms lapse on cancellation, non-renewal or lapse, and don't revive if you re-subscribe. It also says a plan change "within the launch offer" keeps them, though we couldn't find where it defines what that covers.

What 350,897 records looked like, by workload

The count opens on 28 September, when the first spend log on the paid plan starts at a balance of 15,000,000. It closes on 20 October at the lowest logged balance, 14,649,103. In between we read 385 logged rows across 82 files, plus job-posting run-stats. Rows with a before and after balance were counted as balance before minus balance after; job pulls and the queue evaluation use the counts in their own run-stats files. Records per dive are counted as the first balance logged before the dive, minus the last balance logged after it, with the employment checks included.

Shares are rounded.

The evaluation rows are us testing our own research skills against live data. The three client rows are work for one data-infrastructure client: pulling rosters, checking that people still work where a lookup says they do, pulling job postings and scoring a queue of accounts. And the deep dives were research documents on four prospects, 590 records between them.

Then there's the last row. 164,406 records moved the balance without landing in any spend log. Some batch folders from the same weeks have no spend log, and they're the likely source. But that's an inference, not a count, so the table doesn't assign those records to a workload. And two logs may overlap: if the job pulls of 8 October and the queue evaluation that followed counted the same postings, the unattributed share rises by up to 10,623 records, to 49.9%.

And the fix is dull. A spend log on every batch, not most of them.

Three days that used 92% of a Start month

Evaluation rows run from late on 4 October to the afternoon of 6 October, UTC. Across that window the balance fell from 14,941,536 to 14,757,923. That's 183,613 records. Or 91.8% of Start's monthly allowance, used across three calendar days (4 to 6 October).

And of that, 110,504 is logged to the evaluation. Another 11,830 is client work that ran on the same days (2,297 on the employment check, 9,533 on job pulls). The remaining 61,279 sits in no log. That puts the evaluation share somewhere between 110,504 and 183,613 records, and we can't narrow it further from what we kept.

But had our key been on Start from 28 September, the 200,000 would have run out at about 08:05 UTC on 5 October. Seven days in. Every call after that would have paused until the reset, client work included. But that's a hypothetical, and it assumes Start's month counted from the same day ours did.

Then it went quiet. From 14 to 20 October the account drew 1,363 records, 590 of them in the four prospect dives and about 770 in no log. After 20 October the balance moved by only 624 more: 33 in no log, and 591 on a company lookup test we ran on 27 October.

And there's no steady state to quote from this. A monthly average would blend 183,613 records in three days with 1,363 in six. It would describe neither period.

Does logged client work fit Start?

Mostly, yes.

Logged client work came to 75,397 records in the 22 days, about 38% of Start's 200,000: 27,064 on rosters and the employment check, 34,884 on job postings and 13,449 on the queue evaluation. That's 37.7% of Start. Of Scale, 1.5%.

But we lead with the observed figure because the pace overstates it. The last client folder in these logs is dated 14 October, and the account drew almost nothing after 20 October. And our key's cycle was due to reset at 18:04 UTC on 28 October, per the key on 27 October. That makes the first cycle close to a measured month: 351,521 records, 175.8% of Start and 7.0% of Scale.

Client work fits Start with room, but the whole account did not, because evaluation and the unlogged draw ran on the same key. An agency on Start that tests its own tooling on the key it uses for clients could hit the pause in the week it tests, and client work would pause with it.

What a record buys: two per company from a domain

On 27 October we scored 200 companies, drawn at random from 4,789 domains in our own archives, on the legacy key. Every company endpoint keys on a LinkedIn URL, so starting from a domain takes two calls: domain to LinkedIn, then company enrichment. The 200 companies used 398 records, 1.99 each, never more than 2. Of the 200, 199 resolved, and the one miss was billed nothing, which matches what the pricing page says. One default page of the employee-finder roster endpoint (10 rows) cost 9.55 records, about 4.8 times a company score.

And at that rate Start covers about 100,000 companies a month from a domain list, or 200,000 if you already hold the LinkedIn URLs, and Scale covers about 2.5 million.

But the caveats are real: the test ran on our legacy plan and we haven't seen Start or Scale bill the same calls, a crude name check flagged 8 of 199 resolved domains (one clearly wrong, the other seven unchecked), and 94% of the sample is headquartered in the US. Our enrichment waterfall teardown shows where a lookup like this sits in a full waterfall.

Where MoltSets' $27 plan fits the timing

MoltSets' $27 plan was in our notes on 21 September, a week before our 28 September read of Blitz's old structure. It prices by rate limit rather than by a monthly allowance: 1,000 enrichment records per five hours, per its pricing page as of October 2026.

Blitz's own benchmark, vendor-run and published under CC BY 4.0, names MoltSets among six providers. We can show the order of these dates, but not that one caused the other. And our MoltSets tool page covers its plans and limits, while a follow-up piece puts the two vendors side by side.

What to ask your agency about its data plan

If an agency runs Blitz on your work, three questions are worth asking.

Which plan is it on? Our legacy plan, Start and Scale differ on allowance, speed and data, and the data difference decides whether emails and phones can come from this vendor at all. Start and Scale both include email and US phones, while our legacy plan includes neither.

What happens to your work at the cap? On Start and Scale, calls pause until the next billing cycle starts or until the agency moves up a plan. But if your job pulls or employment checks tend to run late in a month, ask who sees the pause first and what waits behind it.

Does its own testing share the key with your work? In our logs, a three-day evaluation window drew 91.8% of a Start month, and logged client work drew 37.7% in 22 days. Same key for both.

A contract point too, for anyone with procurement or data-protection reviews ahead: Start and Scale carry no SLA and no signed DPA. And running Blitz to deliver client work is standard usage on any plan, per Blitz's own FAQ, which leaves volume and contract terms as the deciding questions.

For a wider view of where data sits in an outbound budget, see our cold outbound stack cost breakdown.

At Lead Line Partners, pricing is scoped per engagement. Ask any agency which plan sits under your work and how it is metered.