Can Datadog keep growing 20%+ into 2027?

What the 20% question actually is, and why RevOps teams outside finance should care

The question "can Datadog keep growing 20%+ into 2027" sounds like an equity-research question, and it is one. But it decomposes into a set of operating questions that any revenue-operations leader running a consumption-priced product will recognize immediately, because they are the same questions in a different costume: how much of next year's number is already sitting in the installed base, how much has to be manufactured by new logos, how sensitive is the base to a customer deciding to use less, and which product attach motions can be relied on to close the gap.

Start with the arithmetic, because everything else is commentary on it. Datadog's FY26 revenue lands somewhere around $3.4–3.5B on current guidance. Twenty percent growth on a $3.45B base means FY27 revenue of roughly $4.14B, which means adding roughly $690–720M of incremental revenue in a single year. For context on the shape of the ramp, the company added something in the neighborhood of $650M in FY25 and is on pace for $700M-plus in FY26. So 20% in FY27 is not a step change — it is holding the current absolute dollar pace one more year. That is the crucial reframe. The percentage decelerates automatically as the base compounds; the dollars do not have to grow at all for 20% to land. They just cannot shrink.

That is why the honest probability sits around two-thirds rather than at either extreme. The bear case is not a tail scenario — it is a real and reachable branch that requires only one of three things to go wrong at moderate severity. And the bull case does not require anything heroic either; it requires the security and AI-observability attach motions to convert at rates the company has already demonstrated in APM and log management, just applied to a newer product line.

For a RevOps audience the interesting part is not the stock call. It is that Datadog is the cleanest public example of a consumption-revenue engine at scale, which makes its mechanics readable in a way most private consumption businesses are not. Every lever visible in Datadog's disclosures — net revenue retention, $100K+ customer count, multi-product attach rate, international mix — is a lever a RevOps team can instrument in their own business. If you run usage-based pricing, Datadog's quarterly reporting is effectively a free operating benchmark for what good looks like at $3B+ and what deceleration looks like when it arrives.

The adjacent lesson matters as much as the direct one. Consumption businesses fail their forecasts differently than seat-based businesses. A seat-based SaaS company misses when renewals churn or new bookings slip — discrete, countable events that show up in a pipeline review. A consumption business misses when nothing visible happens at all: no logo churns, no deal is lost, customers simply ingest fewer logs, sample fewer traces, and shorten retention windows. Revenue erodes without a single closed-lost record. That is precisely why the forecasting apparatus for a consumption business has to look different, and why Datadog's own guidance philosophy — conservative initial guides, consistent beats — is a rational response to a demand signal that arrives as a drift rather than an event.

The three-condition test and how the growth actually gets assembled

Rather than treating FY27 as one number, break it into the three sources of incremental revenue and pressure-test each. This is the same decomposition a RevOps team should run on its own annual plan, and it makes the fragile assumption obvious.

Source one: expansion within the installed base. This is net revenue retention doing the work. Datadog's NRR has settled in the mid-110s after the pandemic-era peak well above 130%. Hold NRR at 115% against a $3.45B base and the base alone generates roughly $515–525M of incremental revenue before a single new customer signs. That is about three-quarters of the requirement, produced by customers who have already made the buying decision. This is the single most important number on the page. At 118% the base throws off closer to $620M and the year is close to solved; at 110% it produces around $345M and the gap doubles.

Source two: new logos. If the base delivers ~$520M, the remaining ~$180M has to come from customers who did not exist last year. Datadog's mix of small self-serve accounts and large enterprise lands means blended first-year contribution per new customer is modest; the volume of new accounts the company adds annually is in the low thousands. Historically this is the piece Datadog has over-delivered on, and it is also the piece with the most operating leverage available — a modest lift in sales productivity or a strong partner-sourced quarter moves it. But it is the smaller lever, and it cannot rescue a base that has stopped expanding.

Source three: new product monetization. This is where the case is genuinely uncertain. Security (Cloud SIEM, application security, cloud security posture management) and AI/LLM observability are the two lines expected to carry disproportionate incremental weight. Penetration of the security suite within the large-customer base remains far below the penetration of the core APM and log products, which is simultaneously the bull argument and the risk: the headroom is real, but headroom is not attach. LLM observability sits in an earlier stage still, closer to a land-and-adopt motion than a monetized one, and its revenue contribution in FY27 depends on paid-tier conversion behavior nobody has a long history on.

Read the diagram as an operating model rather than a forecast. The branch point that determines the answer is the NRR node, not the product node. Product attach can add or subtract perhaps a hundred million at the margin; a three-point swing in net retention moves the number by more than that in either direction, and it moves it silently, quarter by quarter, without any single event to point at.

There is a fourth source worth naming because it is frequently double-counted: price. Datadog has periodically adjusted list pricing and introduced new SKUs and tiers, and analysts sometimes model a price contribution on top of the retention math. In practice, list price changes on a consumption product mostly show up *inside* NRR rather than beside it, because the same customer paying more per unit for the same volume is expansion. Modeling both separately inflates the plan. RevOps teams building consumption forecasts should pick one home for price and stick to it.

What the numbers look like in ranges, and what the comparable set says

Precision here is false comfort, so work in ranges and probability mass rather than point estimates.

The scenario spread. A reasonable distribution puts roughly 20% of the mass on a bull case in the mid-20s percent, roughly 55% on a band running from 20% to 24%, and roughly 20% on a bear case in the mid-to-high teens, with a small remaining tail for a genuine macro contraction that takes growth below 15%. The cumulative probability of clearing 20% is therefore in the 65–70% range. That is a strong base case with a real, non-trivial miss branch — not a coin flip, but not a lock either.

What each scenario requires quarterly. Twenty percent growth on this base means roughly $170–180M of net-new revenue per quarter, sustained for four consecutive quarters. Datadog has printed quarters at that level, but not many, and never four in a row without a favorable macro backdrop. That per-quarter framing is more useful than the annual one because it converts an abstract growth rate into a cadence you can check against every earnings release. The first two quarters of FY27 will tell you almost everything; a single quarter at $140M net-new makes the annual number very hard to recover.

Timelines for the swing factors. Security attach compounds on an 18–24 month sales cycle inside large enterprise accounts — meaning the FY27 security number was substantially determined by pipeline built during FY26, not by anything the sales team does in FY27. AI-observability monetization is faster-moving but earlier, and its FY27 contribution is more plausibly measured in the tens of millions than the hundreds unless paid conversion surprises to the upside. Cloud-optimization waves, by contrast, arrive with almost no lead time: a customer's cost-reduction initiative can compress spend within a single billing quarter.

The comparable set, and what each one actually teaches. Look at how other infrastructure and platform businesses behaved as they crossed the $3B threshold.

• Snowflake at a similar scale demonstrated that consumption businesses can re-accelerate rather than decay monotonically, when a genuine new workload category attaches to the existing platform. That is the structural argument for the bull case: AI workloads are a new workload category, and observability attaches to workloads mechanically.
• MongoDB's trajectory is the cautionary counterweight. Consumption revenue decelerated markedly as its cloud database business matured and customers optimized. The lesson is that consumption gravity is real: without a new workload category, usage growth converges toward the customer's own infrastructure growth rate, which is nowhere near 20%.
• ServiceNow held high growth well past this revenue level, but on a subscription model with contracted expansion and a TAM that kept widening into adjacent workflow categories. The read-across is limited precisely because the revenue model differs — ServiceNow's growth is negotiated in advance, Datadog's is metered after the fact.
• CrowdStrike shows what platform-consolidation attach looks like when it works: a security company selling additional modules into an installed base at high attach rates. If Datadog's security suite converts at anything close to that pattern, the bear case largely disappears.

Datadog's own history is the most useful comparable. The deceleration curve from the hypergrowth years down through the mid-20s has been unusually gentle and unusually consistent. Companies with smooth deceleration curves tend to keep decelerating smoothly; they rarely fall off a cliff, and they rarely re-accelerate sharply. Extrapolating that curve alone — no model, just the trend — lands FY27 right around the 20% line. That is a meaningful independent check on the bottoms-up math, and it is why the two approaches agreeing on roughly the same answer raises confidence in the range even though neither is precise.

One more calibration point: the company has publicly framed a multi-billion-dollar long-term revenue ambition that implies a compound growth rate above 20% sustained for several years. Whatever you think of the probability of that ambition, it means 20% in FY27 is *below* the internally-communicated path rather than above it. Management is not guiding to 20% as a stretch; it is closer to the floor of what the organization is planned around. That matters for interpreting behavior — sales capacity, hiring, and R&D allocation are all being set against a higher number, which creates some cushion if the higher number is missed.

Where the analysis and the operating plan both go wrong

The failure modes here are instructive well beyond Datadog, because they are the standard failure modes of forecasting any usage-based revenue stream.

Mistake one: modeling new logos as the primary growth engine. The instinct is to look at the $700M requirement and start sizing a sales team against it. But three-quarters of that number comes from the base. Building the plan around new-logo capacity misallocates attention and headcount toward the smaller lever. In a RevOps context this shows up as an organization with a well-instrumented new-business funnel and almost no instrumentation on expansion — no leading indicators of usage trajectory, no early-warning system for a customer whose ingest volume has flattened, no named owner for the retention number. If NRR is producing 75% of growth, it deserves 75% of the operating attention.

Mistake two: treating net retention as a single blended number. Aggregate NRR hides everything that matters. The real signal lives in the cohort structure: NRR among the top decile of accounts versus the long tail, NRR among multi-product customers versus single-product ones, NRR among customers whose own business is growing versus flat. A blended 115% can be a healthy 122% in multi-product accounts dragging along a deteriorating 104% in single-product ones — which tells you exactly where to intervene — or it can be uniform, which tells you something entirely different. Reporting only the blend guarantees you learn the answer too late to act.

Mistake three: ignoring concentration. A meaningful slice of revenue at this scale sits in a relatively small number of very large accounts, several of them AI-native or hyperscaler-adjacent companies whose own infrastructure spending is volatile. A single large customer restructuring its observability approach can move a full quarter's growth rate by a noticeable margin. This is not hypothetical for Datadog — large-customer optimization has visibly compressed growth before. The operating response is boring and effective: name the top accounts, forecast them individually rather than in the aggregate model, and maintain a direct relationship with the engineering leaders who actually control ingest volume, not just the procurement contact who signs the contract.

Mistake four: assuming AI is monotonically additive. The bull thesis holds that AI workloads generate more telemetry, agents generate more traces, and therefore consumption rises. Probably true. But the same AI capability can be pointed at cost reduction — intelligent sampling, automated retention tiering, anomaly-triggered ingest rather than always-on ingest. The identical technology that inflates the numerator can deflate it. The direction of the net effect is an empirical question that will be settled by customer behavior over the next several quarters, and anyone claiming certainty in either direction is asserting rather than knowing. The honest position is that early signals point toward net-additive, on a sample too small to be conclusive.

Mistake five: dismissing the bundling threat as pure FUD, or accepting it as inevitable. Hyperscaler and platform bundling — observability and security packaged into a broader enterprise agreement — is a genuine pressure on attach rates, particularly in the mid-market where the buyer is price-sensitive and the technical requirements are less demanding. It is also chronically overstated for large sophisticated enterprises, who consistently choose best-of-breed observability because the operational cost of a worse tool during an incident dwarfs the license savings. The correct read is segmented: real risk in mid-market, limited risk at the top of the enterprise, and the mix of Datadog's revenue determines how much it matters.

Mistake six: confusing the growth rate with the business quality. A consumption business decelerating from the mid-20s to the high teens while expanding margins and holding retention is a healthier business than one holding 25% by discounting into unprofitable accounts. The 20% threshold is an arbitrary line that matters to a valuation multiple, not to whether the underlying engine is working. RevOps teams internalize the wrong lesson when they treat a headline growth rate as the sole scoreboard.

Choosing what to watch: a decision framework for the quarters ahead

If you want to update your view efficiently rather than re-reading every earnings transcript, there is a short ordered list of indicators, and they are not equally informative. Watch them in this sequence, because the early ones dominate.

First, net revenue retention. Everything else is secondary. Above 117% and the year essentially takes care of itself; the base does the work and product attach is upside. Between 113% and 117% the outcome is genuinely contested and depends on new logos and attach. Below 112% and 20% requires new-logo performance the company has never demonstrated. One quarter below trend is noise; two consecutive is a signal.

Second, large-customer count and the growth rate of that count. The $100K+ ARR cohort drives the majority of revenue. What matters is not the absolute count but the second derivative — the rate at which the cohort is growing. A decelerating add rate in this cohort is the earliest visible sign that the enterprise motion is saturating, and it leads the revenue number by several quarters.

Third, multi-product attach. The proportion of customers using four, six, or eight products is the single best proxy for whether the platform strategy is compounding. Attach is what converts a monitoring vendor into an infrastructure standard, and it is what makes the security and AI lines credible rather than aspirational.

Fourth, gross margin and operating margin. These are not growth indicators, but they are quality-of-growth indicators. Growth purchased through discounting shows up in gross margin before it shows up anywhere else.

Fifth, commentary on optimization. Management language about customers "optimizing" or "rightsizing" is the tell for a compression wave. It shows up in prepared remarks before it shows up in reported numbers, usually a quarter or two ahead.

The same framework transfers directly to internal planning at any usage-priced company. Substitute your own retention cohorts for Datadog's NRR, your own large-account tier for the $100K+ cohort, and your own module attach for multi-product attach, and you have a quarterly review that actually predicts the annual number rather than explaining it afterward. The reason this framework works is that it is ordered by leverage: the indicator at the top moves the outcome most and moves earliest, and each subsequent indicator only matters conditionally on the one above it.

A note on the adjacent question people usually mean when they ask this one. "Can Datadog keep growing 20%" is often shorthand for "is the observability market still expanding, or is it consolidating into a few platform winners." Those have different answers. The market is expanding, driven by workload growth and by AI systems that are harder to debug than the systems that preceded them. But the market is *also* consolidating, as customers reduce the number of monitoring vendors they run. Datadog benefits from both dynamics simultaneously — expansion grows the pie, consolidation grows its slice — which is a genuinely unusual position and the strongest structural argument for the base case. It is also why a competitor gaining share does not automatically translate into Datadog losing revenue: in an expanding market, share can shift while absolute revenue rises for everyone.

Related questions

What single number best predicts whether Datadog clears 20% in FY27?

Net revenue retention. At 115% or above, the installed base alone produces roughly three-quarters of the required incremental revenue. Below 112%, the gap has to be filled by new-logo performance the company has not demonstrated at this scale.

Does missing 20% mean the business is deteriorating?

No. A consumption business decelerating from the mid-20s to the high teens while expanding operating margin and holding retention is functioning normally. The 20% threshold matters to a valuation multiple, not to whether the underlying revenue engine is healthy.

How much can AI workloads realistically contribute in FY27?

Meaningfully but not decisively. AI-generated telemetry raises ingest volumes across the base, which flows through retention rather than appearing as a separate line. Directly monetized AI-observability products are earlier-stage, more plausibly a tens-of-millions contribution in FY27 than hundreds.

Is hyperscaler bundling a real threat to the security attach thesis?

Segmented. It is a genuine pressure in mid-market accounts where price dominates and requirements are simpler. At large enterprises, best-of-breed observability usually wins because incident cost exceeds license savings. Revenue mix determines how much the mid-market pressure matters.

What would make the bear case arrive fastest?

A cloud-optimization wave at the largest accounts. Unlike sales-cycle-driven factors that take quarters to develop, an optimization initiative compresses spend within a single billing period, with no churn event to flag it in a pipeline review.

FAQ

Why is $700M the number that matters rather than the 20% figure?

Because percentages decelerate mechanically as the base compounds while absolute dollars do not. Twenty percent growth on a roughly $3.45B FY26 base requires close to $700M of incremental revenue. Datadog added roughly $650M in FY25 and is tracking above that in FY26, so clearing 20% in FY27 means holding the current absolute pace rather than accelerating. Framing the target in dollars makes it obvious that the year is a continuation, not a step-change, which is why the probability sits meaningfully above 50%.

What does net revenue retention actually measure here?

It measures how much revenue the same set of customers generates this year versus last, including expansion, contraction, and churn but excluding new customers. For a consumption business it captures something subtler than upsell: it captures whether customers are using more of the product. A customer that signs no new contract but doubles its log ingest shows up as expansion. That is why retention is both the largest growth lever and the hardest one to influence directly through sales activity.

How is forecasting a consumption business different from forecasting seat-based SaaS?

In seat-based SaaS, revenue changes through discrete events — a renewal, an upsell, a churn — each of which appears in a CRM record and a pipeline review. In a consumption business, revenue changes continuously as customers adjust usage, and a serious miss can happen with zero closed-lost records and zero logo churn. The forecasting apparatus has to be built on usage telemetry and cohort trends rather than on opportunity stages, which is a genuinely different operating discipline for a RevOps team to build.

Why does customer concentration matter more here than in a typical SaaS model?

Because the largest accounts in a consumption model can change their spend unilaterally and quickly. A large AI-native or hyperscaler-adjacent customer restructuring how it samples traces or how long it retains logs can move a quarter's growth rate without any commercial negotiation taking place. In a contracted subscription model, the same customer would be locked to its committed spend until renewal, giving the vendor quarters of warning.

Does the security product line have to work for the 20% case to hold?

Not strictly, but it removes most of the margin for error if it does not. If retention lands at the high end and the base delivers close to $620M, the remaining gap is small enough for new logos and modest attach to close. If retention lands mid-range, the security and AI lines have to contribute real incremental dollars. Security attach is also the highest-headroom lever available, since its penetration in the large-customer base remains far below the core monitoring products.

What is the most useful signal for a RevOps team that wants to keep learning from this?

Read Datadog's quarterly disclosures as a free operating benchmark rather than as an investment thesis. Net retention, large-customer cohort growth, multi-product attach, and international mix are all levers that map onto any usage-priced business. Watching how a $3B-plus consumption company reports and explains those metrics is one of the cheapest sources of calibration available to an operator running the same model at a smaller scale.

Sources

• https://investors.datadoghq.com/ — Datadog investor relations: quarterly results, guidance, and shareholder letters
• https://www.sec.gov/edgar/search/ — SEC EDGAR full-text search for Datadog 10-K and 10-Q filings
• https://www.datadoghq.com/pricing/ — Datadog's published product and pricing structure across product lines
• https://www.gartner.com/en/information-technology — Gartner IT research covering observability and monitoring markets
• https://www.forrester.com/research/ — Forrester research on IT operations and application performance management
• https://www.idc.com/ — IDC market sizing and spending forecasts for cloud infrastructure software
• https://www.cnbc.com/technology/ — CNBC technology coverage including earnings and analyst commentary
• https://opentelemetry.io/ — OpenTelemetry project documentation, the open standard shaping observability data collection

Related on PULSE

• Can Outreach keep growing 20%+ into 2027?
• Can ServiceNow keep growing 20%+ into 2027?
• Can HubSpot keep growing 20% YoY into 2027?
• Can Salesloft keep growing 15%+ post-Vista acquisition?
• How does the growing use of AI for procurement compliance checks lengthen the legal review stage in 2027?