What monitoring X for leads costs now the free tier is gone

You built the keyword monitor in an afternoon two years ago. A cron job, a filtered stream, a webhook into Slack, and for the price of nothing you got a ping every time somebody on Twitter asked out loud for a tool like yours. It ran quietl

You built the keyword monitor in an afternoon two years ago. A cron job, a filtered stream, a webhook into Slack, and for the price of nothing you got a ping every time somebody on Twitter asked out loud for a tool like yours. It ran quietly the whole time. Then it stopped, and the error was a billing error, and the fix was a credit card. Monitoring X for leads is a metered utility now, and the meter starts at half a cent per post.

That turns a question that used to be about engineering into a question about arithmetic. Not whether you can watch X for buying signals, but what each signal costs you before you have written a single word of reply.

X's own changelog puts the switch on 6 February 2026

X's developer changelog dates the launch of pay-per-use pricing to 6 February 2026 and describes it as a credit-based model built for indie builders, startups and hobbyists. Legacy free-tier users got a one-time $10 voucher. Apps X classifies as Public Utility Apps keep free access. Everybody else buys credits up front in the Developer Console and watches them drain.

Two months later, on 16 April 2026, the same changelog records a second change worth knowing before you plan any automation: following, likes and quote-posts came out of the self-serve tiers entirely. If your outbound playbook involved warming a prospect by following and liking them programmatically, that endpoint is no longer on a menu you can buy from.

The rates are per resource returned, which is the whole story

X's pricing page lists the meter plainly. Posts read cost $0.005 per resource. User reads cost $0.010. Reads of your own data, where the authenticated user owns the app, cost $0.001. Creating a post costs $0.015, and $0.200 if the post contains a link.

Per resource returned. Not per query, not per hour, not per keyword. Every post your search matches costs half a cent whether it was a founder asking which tool to buy or a bot quoting song lyrics that happen to contain your keyword.

One piece of good news sits in the docs. X states that the same resource requested twice in 24 hours is only charged once. Polling a thread through the afternoon for new replies does not multiply your bill that day. Re-reading the same search results across a week does.

The meter charges you for the noise, not for the leads

Here is the arithmetic that decides this, and it needs one number you already have: how many posts a day your keyword actually matches.

At 100 matches a day you read 3,000 posts a month and pay $15. At 500 a day, $75. At 2,000 a day, which is roughly what a generic term like CRM or invoicing looks like on a platform that size, you land at 60,000 posts and $300 a month before anything useful happens.

Now put your own precision number on top, because nobody can hand you that one. Every 200 matched posts cost a dollar. If one in 200 turns out to be a person who wanted to buy something, each real lead cost a dollar in reads. If one in 20, it cost a nickel. If one in 2,000, you are paying ten dollars a lead to a platform for the privilege of finding it, and you still have to read all two thousand.

On a free API, bad keyword targeting cost you an afternoon of reading. On a metered one, it shows up on an invoice.

That is the real change here. Precision used to be a productivity problem you could ignore by skimming faster. It is a line item now, and the line item is honest about your keyword choices in a way your own time never was.

Two million reads is the self-serve ceiling and it prices out at $10,000

The pricing page caps pay-per-usage at 2 million post reads per monthly billing cycle and points you at an Enterprise plan above that. Multiply the cap by the rate and the self-serve ceiling is $10,000 a month of reading.

Nobody watching keywords for a seed-stage product goes anywhere near that number. It still matters, because it tells you where X drew the line. The self-serve meter is priced for a small set of well-chosen keywords, not for a firehose you filter afterwards. Filtering afterwards is exactly what a broad keyword monitor does.

Hacker News is the control group and it is still free

Point the same monitor at Hacker News and the cost conversation disappears. The official HN API needs no key, no authentication and no payment, and its own documentation states there is currently no rate limit.

That is why our five-minute HN monitoring setup is a build-it-yourself post. On HN the do-it-yourself version is a cron job and a fetch call, and the only budget it spends is attention.

Reddit's constraint was never the price, it was the permission

Reddit sits in a third category, neither meter nor free-for-all. Reddit's developer access page says you cannot use its developer tools and services for commercial purposes without first getting permission, and it counts subscription services and fee-based data access as commercial purposes.

So the three surfaces fail in three different ways. X sells you access and bills you for every irrelevant match. Reddit will not sell it at all without permission and a contract. Hacker News hands it over. Which makes the general build versus buy question for sales intelligence a per-platform decision rather than a strategy. Build on HN. Pay the meter or a vendor on X. On Reddit, buy from someone who can describe their access without flinching.

Letting somebody else eat the meter is the honest middle option

If you want X coverage without a billing relationship with X, that is the entire job of the monitoring vendors. Syften covers X alongside Reddit, Hacker News, GitHub, YouTube, Bluesky, Slack communities and a long list of forums, states a worst case of 15 minutes on X, and runs $29.95, $49.95 and $119.95 a month for 3, 20 or 100 community filters with daily result caps of 100, 200 and 500.

Compare that to the meter without cheating. Syften's entry plan delivers up to 100 results a day, which is about 3,000 a month, and reading 3,000 posts through X's own API would run you roughly $15. So at the low end you pay a premium for not maintaining a pipeline, and pick up ten other platforms while you are there. Their cap counts results delivered to you, not posts read on their side, so it is not a clean unit-for-unit comparison. Above a few hundred matches a day the vendor stops being a premium and starts being the cheaper option outright.

Shadow Inbox is our product and it does not help you on this one. We watch Reddit and Hacker News, with Quora and LinkedIn in beta, and X is not on the list. If X is the platform where your buyers actually talk, Syften is the better tool and it is not close.

What happens after the alert is still free, which is where the leverage went

Every dollar in this post buys the finding half of the problem. None of it buys the half that converts.

An alert that lands 15 minutes after a founder asks for a recommendation and an alert that lands three days later cost the same half cent. Only one of them is worth anything. We keep saying timing is most of outbound because it is the only variable in this whole cost structure that is free to improve, and it is the one people skip while they shop for tooling.

Precision works the same way. The reason to classify intent before it reaches your inbox used to be that reading 2,000 posts a week is miserable. Now it is also that fetching those 2,000 posts costs $10, and a filter good enough to let you run a narrower keyword pays for itself twice: once on the invoice, once on the hour you get back.

Pick the narrow keyword. Pay for fewer, better matches. Spend the difference on the reply.

● FAQ

How much does it cost to monitor X for leads in 2026?

X's pricing page lists post reads at $0.005 per resource returned, so the bill scales with how many posts your keyword matches. A keyword matching 100 posts a day costs about $15 a month, 500 a day costs about $75, and a generic term matching 2,000 a day costs about $300. Pay-per-usage is capped at 2 million post reads per billing cycle, which works out to a $10,000 self-serve ceiling.

Is there still a free X API tier?

X's developer changelog dates the launch of pay-per-use pricing to 6 February 2026 and says legacy free-tier users received a one-time $10 voucher. Apps X classifies as Public Utility Apps keep free access. The current pricing page lists no free allowance for everyone else, only credits bought up front in the Developer Console.

Does polling the same X thread repeatedly cost extra?

Not within the same day. X's own API overview states that the same resource requested twice in 24 hours is only charged once. Polling a thread through the afternoon for new replies does not multiply your bill that day, but re-reading the same search results across a week does.

Is it cheaper to pay the X API meter or buy a monitoring tool?

It depends on how noisy your keyword is. Below a few hundred matches a day the meter is cheaper in raw dollars, and you pay for the pipeline in maintenance instead. Above that a vendor like Syften wins outright, and it covers ten other platforms in the same subscription.

Does Shadow Inbox monitor X?

No. Shadow Inbox is our product and it watches Reddit and Hacker News, with Quora and LinkedIn in beta. X is not on the list. If X is where your buyers talk, Syften is the better tool and it is not a close call.