New Relic Pricing: The Bill Is Users, Not Gigabytes

The New Relic homepage leads with the gigabyte. $0.40 a GB after the first 100 free, one rate for every kind of data, no per-host line to argue about. It is a simpler model than the Datadog quote we took apart last week, and the simplicity is the trap: teams price the data, sign, and find the invoice is mostly people.

New Relic pricing has two lines, and the second one is a seat. On the Pro edition a full platform user is $349 a month at list. Eight engineers who need the APM screens cost more than two terabytes of logs.

Our position: the seat count is the number to fight over before you sign, not the per-gigabyte rate. Decide who really needs the curated screens, keep everyone else on the free tier, and measure the data before anyone quotes you a commitment.

What is New Relic pricing actually made of?

Two meters. The pricing page says the bill is built from "users and data ingest". Data is one rate for everything: 100 GB a month is free, then $0.40 a gigabyte on the original data option or $0.60 on Data Plus. No separate charge for hosts, containers, custom metrics or an index: everything is bytes.

Users come in three types. Basic users are free and unlimited. Core users are $49 a month. Full platform users depend on the edition: one free on the Free plan, $10 for the first and $99 for each of up to five on Standard, $349 on Pro with an annual commitment or $418.80 paying monthly, and "contact sales" on Enterprise. The data ingest billing page adds two details: what counts is "the size of the data that actually ends up being stored", after New Relic's own trimming, and "monthly GB Ingested is rounded up to the nearest integer".

Why is the seat the number to argue about?

Because of what a seat costs relative to the data. At $0.40 a gigabyte, one Pro seat a month buys 872 GB of ingest; a year of it is $4,188 at list, which is ten terabytes. Most fintech accounts we see do not send ten terabytes a year. Several have ten people who could plausibly ask for a login.

So the estimate starts with a list of names and what each person does in the tool. New Relic's user type page is the guide. A basic user can build dashboards, run NRQL, receive alerts and view logs, but "can't use our curated experiences (for example, our APM UI, browser monitoring UI, or mobile UI)". A core user adds the errors inbox and the full log UI. Only a full platform user gets APM, infrastructure and synthetics as New Relic designed them.

In practice the people on the pager need the expensive seat and almost nobody else does. The CTO who looks at one dashboard a week is basic. The developer who lives in the errors inbox is core, at a seventh of the price. We usually land a 20-person engineering org at four to six paid seats, and the first draft always has 20.

Which users will New Relic bill that you did not plan for?

The ones somebody invited. The user billing page is clear: "even if a billable user has never logged in to New Relic and has a Pending invite tag in the UI, they're still billable". A contractor's unaccepted invite is $349 a month.

The month is the other trap. A user's billable type "is defined as the highest user type at which a user was set during a calendar month", in UTC. Upgrade someone for a three-hour incident on the 29th and downgrade them on the 30th, and you have paid for the month. New Relic prorates only the first month of a subscription and the last.

Then the one that catches teams on an annual contract. Pay-as-you-go has "no rules limiting the downgrading of users". On a commitment, if a full platform user is downgraded and later returned to full platform, and "this downgrade/upgrade cycle happens twice in a contract year, that user will be billed as a Full Platform user for the remainder of the year". A rota of six engineers who take the seat during their on-call week hits that rule by month three. Seats do not rotate.

How much data will an AWS account actually send?

More than the log figure, because the infrastructure and APM agents send samples as well as the logs you forward. New Relic does not publish a per-host number, and we would not trust one if it did. Run the agents on a slice of the fleet for a week on the free account. Then the usage queries page gives the query that splits it by source:

FROM NrConsumption SELECT sum(GigabytesIngested)
  WHERE productLine = 'DataPlatform'
  SINCE 7 days ago FACET usageMetric TIMESERIES 1 day

Scale each source up to the full fleet and out to the month. Logging is usually the largest line and the easiest to cut on the forwarder before it leaves the box; the log rules in the post on cutting a Datadog bill apply to a New Relic forwarder unchanged, because the waste is in the application, not the vendor.

Watch the AWS side too. New Relic pulls CloudWatch metrics through a metric stream, and AWS's metric streams documentation says "metric streams pricing is based on the number of metric updates", plus Firehose underneath. That lands on the AWS bill, where nobody attributes it to monitoring. Filter the stream to the namespaces you alert on.

Does Data Plus buy the retention CERT-In wants?

No, and the reason matters. The retention page gives logging 30 days on the original option and 120 on Data Plus, with APM data at 8 and 98 days. Retention beyond that "costs $0.05 per GB ingested per month" for each additional 30 days.

CERT-In's 2022 directions, paragraph (iv), want logs kept "for a rolling period of 180 days" and "within the Indian jurisdiction". New Relic's data centre page lists three regions, the United States, the European Union and Japan, and says an organisation "can only be in a single data center region". So the 180 days cannot be bought from New Relic at any retention setting, because the residency half fails. The compliance copy is an S3 bucket in ap-south-1 fed from the log forwarder, and New Relic keeps the 30 days on-call actually searches. Whether payment data may leave the country at all is the analysis in the data localisation post; usually the answer is to strip it before the agent sees it.

How do you estimate New Relic pricing before you sign?

The same shape we priced for Datadog: 40 hosts, 20 of them traced, 60 GB of logs a day, and 20 engineers. At list, on Pro with an annual commitment:

• Full platform users: 5 on-call engineers at $349, $1,745.
• Core users: 6 developers who need the errors inbox at $49, $294.
• Basic users: the other 9, $0.
• Logs: 60 GB a day, 1,800 GB a month.
• Infrastructure, APM and metric stream samples: call it 400 GB, and measure it.
• Data: 2,200 GB less 100 free, 2,100 at $0.40, $840.

That is $2,879 a month, about $34,500 a year, with seats at 71 percent of it. Hand every engineer the full seat and the same account is $7,820 a month. Trim the logs by half and it drops by $360. The New Relic cost lever is the org chart.

Against the Datadog figure of $2,158 for the same fleet, the comparison is not clean: that quote indexed a fifth of the logs and this one stores all of them. Price both against your own list of names.

Should you commit, or pay as you go?

Pay as you go for the first quarter, on the seats especially. The monthly Pro seat is $418.80 against $349, a 20 percent premium, and it buys the right to downgrade people freely while you learn who opens the tool. After three months the NrMTDConsumption query on the usage page says what to commit to:

FROM NrMTDConsumption SELECT latest(FullPlatformUsersBillable),
  latest(CoreUsersBillable) SINCE 3 hours ago

Commit to the seats that were full platform in all three months and no others. Data is different: there is no on-demand premium on the gigabyte, so a data commitment is a discount negotiation, not a rate decision. Put the usage page's ingest alert on the account on day one, at your estimate, so a spike is a Slack message and not an invoice.

When is Datadog the honest answer?

When many people need the curated screens and the fleet is small. Datadog's bill scales with hosts; the New Relic cost scales with people. A 30-engineer product team on 15 hosts pays Datadog $225 a month for infrastructure at the list price in the Datadog post, and would pay New Relic $10,470 for the same team on Pro seats. The reverse holds: a five-person platform team running 400 hosts is New Relic's customer.

And for the team of three from the hire-or-partner post, with ten hosts and no platform engineer, the honest answer is the Free edition: 100 GB, one full platform user, and nothing to sign.

The short version

Write the list of names before you read the rate card. Full platform seats for the people on the pager, core for the people in the errors inbox, basic for everyone else, and no pending invites. Run the agents on a slice of the fleet for a week on the free account and read NrConsumption by source. Keep New Relic at 30 days and put CERT-In's 180 in S3 in ap-south-1. Pay monthly for a quarter, then commit to the seats that were paid every month, and never rotate a seat with the on-call rota. Filter the metric stream.

If the quote in front of you has a seat count equal to the engineering headcount, that is the finding. Send back the list.

This is engineering guidance, not legal advice. Regulations and their annexures are amended over time; read the current text and take your compliance position from counsel.