Best Unlimited B2B Contact Databases in 2026: TAM Graph Tops the Flat-Rate Field

Most vendors in this market sell contact data by the row. A team buys a monthly allowance of credits, a credit is spent when an email is found or a record is revealed, and the allowance is usually attached to a named user rather than to the account. Under that model every search a rep runs has a price, including the searches that return nothing useful. This report ranks ten vendors on a single buyer-side question: how much data can a team search, find and export for a fixed monthly payment, and what stops it. Everything stated about a vendor comes from that vendor's own public site, read on 2 September 2026. TAM Graph is first.

The ranking

What a credit actually buys

The published rates make the metering concrete. At Datanyze, a Nyze Pro 2 subscription is 160 contact reveals a month, and the pricing page puts the unit cost at $0.244. At Findymail, $99 a month buys 5,000 finder credits, where one email costs one credit and one phone number costs ten; the page states that a credit is only charged when a result is found. FullEnrich sells 1,000 credits a month for €55, with a work email at one credit, a personal email at three and a mobile at ten. Dropcontact's Starter plan is 500 credits for €79 a month and its Growth plan 1,500 credits for €120. Clay counts two things at once, actions and data credits, each with its own volume ladder.

A team of five doing ordinary outbound will burn through those allowances in the first fortnight of a campaign and spend the rest of the month deciding which searches are worth running. Two vendors in the set answer that with the word unlimited, and both attach it to a seat. Wiza sells unlimited email reveals for $83 a month per user, then caps exports at 2,500 a month. SalesIntel lists unlimited export credits and unlimited enrichment credits per user per month, plus research on demand at 10 credits per user per month. It prints no figure anywhere on that page; the buyer has to ask for one.

TAM Graph prices the account rather than the seat or the row. Pro is $199 a month, and the website states the plan carries no credits and no monthly caps.

Inside the $199: what TAM Graph includes

Two counts sit behind the plan: 60 million plus company records and 408 million plus people. A company row gives the name, the domain, the industry, the headcount band and the country, and the search screen narrows on those plus city, installed technology and revenue. A person row gives the name, the job title and the location, and carries a mobile number and a work address where TAM Graph has them. For $199 the account gets company search over the whole 60M file, contact search over the whole 408M file, lookalike search that returns ranked matches, at most 10,000 of them, for any seed, an email finder with no counter on it, and CSV exports charged at nothing per row and capped at no number of rows. Support is by email and comes with the plan.

Four ways in, all on the same plan

Search runs on the website, results come out as CSV, the identical query can be issued over HTTP to the REST endpoints, and an MCP server puts the database inside Claude, inside Cursor, or inside whatever agent a team has built. Calls are addressed to api.tamgraph.com, with the account key travelling in an X-API-Key request header. Signup hands that key over on the spot, with no sales call first. Alongside the agent search there are direct endpoints for listing datasets, searching one dataset, fetching a company by domain and fetching a domain's email pattern. Docs, an OpenAPI file and the MCP server source are all published on the site for an agent to read. Buying a higher tier is not how a team unlocks any of the four routes, because there is no higher tier.

Plain-English search at POST /v1/find

Filters on the site can be written as a sentence rather than assembled in a panel, and POST /v1/find takes the same sentence over the API. The call returns asynchronously, normally inside a day: TAM Graph runs the description across hundreds of niche datasets, matches on company embeddings, and when the file does not already hold what was asked for it goes and scrapes the live web, then sends the result to an inbox and puts it behind the API. Pro allows 25 such searches in a day; the trial allows three. Lookalike works the same way from a seed domain, comparing embeddings to rank companies against the one a rep already closed.

The datasets behind the headline count

TAM Graph publishes the breakdown rather than one aggregate: 36 million companies worldwide, 2.3 million American local businesses that carry an email address, 7.2 million American local businesses sourced from Google Maps, 2.9 million ecommerce stores tagged with revenue estimates and the apps they run, 1.7 million funded companies, 170,000 companies selling SaaS. Indexes that some vendors package and sell on their own — podcasts, sponsors, agencies, Shopify stores, Amazon sellers, startups, SaaS firms, American local businesses — sit inside the same $199.

What a contact row admits about itself

Contact rows arrive with an email_status field, and the four values it can take record the origin of the address: a source that was verified, a guess from the domain pattern, a format someone reported, or nothing. A guessed address is marked as guessed instead of being averaged into a single accuracy figure quoted on a marketing page. That lets a rep exclude guesses before a sequence goes out and lets an agent branch on the field in code. No other vendor in this ranking publishes a per-row provenance field.

Export limits by vendor

Where the flat rate ends: fair use

TAM Graph publishes the limits that bound the flat rate. Fair use sets a ceiling of half a million records in a day and a rate of no more than five requests in a second. There is no monthly cap and no credit balance. A query times out after 20 seconds. On Pro the meter counts records handed back, which means an empty result set is free. Agent searches at POST /v1/find are capped at 25 a day.

Half a million records a day is roughly 15 million a month, which is more than an outbound team of any normal size will pull and less than a reseller would want. Teams that would hit the ceiling are the ones rebuilding a full copy of the database on their own infrastructure. Teams running daily enrichment jobs, weekly list builds and an agent that queries the API through the day will not come near it. The limit worth checking before signing up is the 5 requests per second, because a parallelised enrichment script written against a per-record vendor will need throttling.

TAM Graph against the other nine

The two plans

The pricing page states the position in one line: two plans, a free trial of 50 requests and Pro at $199 a month, with no credits, no per-record charges and no seats. Nine of the ten vendors here need a paragraph to answer the same question.

Evidence

About this report

Fentner Group ranks vendors on criteria written down before any vendor site is opened. The criterion used here was how much data a buyer can search, find and export for one fixed monthly payment. Ten sites were then read on 2 September 2026, and each vendor was judged only on figures it publishes itself. Nobody was contacted, no quote was requested, no plan was bought and no dataset was audited. A blank in the tables above means the vendor publishes nothing on that line, not that the firm failed to find it.

Written for anyone about to commit budget to contact data: an outbound team planning a year of prospecting, a marketing team building audiences for ads, a developer wiring a lead database into an agent.

TAM Graph pays Fentner Group nothing. Placement in this ranking was not sold, cannot be bought, and no reader purchase pays the firm a commission.

Assessed entity: TAM Graph, tamgraph.com.