Executive Intelligence Dossier
GRADE B · 81/100 · BOARDREADY
Public sample
Demand is measured strong. Conversion isn't — and two unreconciled leaks compete to explain why. Datadog, Inc.
This is a real, unredacted Colleviate dossier from our live corpus of 7,401 continuously monitored companies. Generated by the same engine that produces every $599 dossier, grounded in measured evidence, evaluated by our deterministic boardReady gate. We publish it in full so the shape and standard are not a promise — they are the artifact.
Overall grade
B / 81
boardReady · 0 weaknesses
Technical debt
30/100
7.76s LCP (lab-only, no CrUX) · 4.7s script
Pattern
CONVERSION / AI-VISIBILITY GAP
Awareness 73 vs Conversion 47
Synthesis
CLAUDE · GOLD 3-PASS
draft · critique · revise
Evidence · datadoghq.com homepage, captured during the audit run · unmodified full-page screenshot
01 · Central question
The one question that ranks above the rest
Datadog's own funnel model shows demand generated and then leaking at Conversion — Awareness scores 73/100 against a Conversion score of just 47/100 — while its GTM model flags AI/generative visibility as the single weakest lever (0/100, uncited in all 3 buyer-intent AI-answer probes run). At the same time, the homepage posts a 7.76s Largest Contentful Paint (lab, Google PageSpeed synthetic test) with 4.7s of main-thread script work and flagged JavaScript console errors — lab-measured, with no real-user field data available to confirm it at scale. So is Datadog losing conversion to a page it fully controls, or to AI answers it does not appear in at all — and does leadership currently know which leak is larger?
Every other finding here is a candidate explanation for one fact: strong, well-measured demand is not converting proportionally, and two independent, unreconciled leaks sit at that stage — one on-page (lab-measured only), one in AI-mediated discovery (measured at 0% visibility) — rather than a single clean cause.
02 · Business model read
What the numbers say about how Datadog grows.
Per this audit's ICP model, Datadog's buyers are technical evaluators (DevOps/SRE) and engineering economic buyers (VP Engineering/Infrastructure) who assess and shortlist observability tools directly — a self-directed, technical buying motion rather than an extended enterprise procurement cycle. On the evidence here, Datadog already wins the demand-generation half of that motion (Awareness 73/100, a leading 39.9% Hacker-News-based share of voice against New Relic, Dynatrace and Splunk), so growth is constrained less by attracting evaluators than by whether they convert — whether that failure happens on-page or before it starts, in an AI-mediated shortlist that does not name Datadog at all. Confidence: medium.
Growth drivers (evidence-backed)
- Technical-evaluator, self-directed buying motion (per this audit's ICP model).
- Category leadership on classic-channel share of voice (39.9% vs. New Relic, Dynatrace, Splunk; rank 1 of 4).
- Large paid-media footprint (8,000–9,000 Google ad creatives; 8,378 LinkedIn ads, both measured) feeding a funnel that narrows sharply at Conversion (47/100 vs. Awareness 73/100).
03 · Strategic chain
Two leaks at the one stage that matters.
How the signal becomes a strategic problem.
Signal
The funnel model scores Conversion at 47/100 against Awareness at 73/100 — the model's own computed bottleneck — while the GTM model's lowest-scoring dimension is AI/generative visibility at 0/100 (0 of 3 buyer-intent AI-answer probes cited the brand); the homepage separately measures a 7.76s Largest Contentful Paint and 4.7s of main-thread script work — both lab-measured, Google PageSpeed synthetic test, with no real-user (CrUX) data available to confirm either at scale — against a 30/100 Technical Debt Index and flagged JavaScript console errors.
Consequence
Evaluators who reach the site meet a script-heavy, lab-measured-slow homepage at the exact moment they are deciding, while evaluators who instead ask an AI assistant may never be pointed to Datadog at all — two independent leaks converging on the same funnel stage.
Second order
Because Awareness is already a relative strength, the marginal growth dollar spent making more demand is spent on the lever that is not the constraint, while the actual constraint — conversion — has neither a lab-to-field validation nor an AI-visibility remediation in place.
Economic
For a self-serve, evaluation-led business, conversion friction (on-page or AI-mediated) suppresses the pipeline that future net-revenue-retention compounds from; the modeled ~4% conversion drag tied to the lab LCP is directional only (no first-party or real-user data confirms it at Datadog's actual traffic), and the AI-visibility loss is entirely unsized in dollars because no query-share or referral data exists in this audit.
Strategic
Datadog risks over-funding the lever that already works (demand generation) while the two candidate causes of the conversion leak — page performance and AI-answer invisibility — remain uninstrumented and unowned.
Future state
Over the next 1–2 years, as more technical evaluation shifts into AI assistants, an unaddressed 0% AI-visibility score risks converting a conversion-stage leak into a demand-formation gap that a more-cited competitor would quietly capture — while the on-page performance question stays open until real-user field data exists.
Evidence required
First-party trial-start-to-activation conversion segmented by real-user (CrUX/RUM) page-load cohort — not just the lab figure; a measured share of AI-assistant answers naming Datadog vs. New Relic, Dynatrace and Splunk; and first-party attribution linking AI-referred sessions to pipeline.
Evidence · mobile Lighthouse capture, network+CPU throttled (lab conditions) · source for the 7.76s LCP / 30-100 debt index above · no real-user (CrUX) data exists for this domain
04 · Forensic finding
The distress read sharpens rather than distracts from the thesis.
The 2026 10-K reads with a net tone of -5.19 and a modal conviction of -7.02 (Loughran-McDonald, EDGAR forensics, measured) — hedged, uncertainty-heavy disclosure language, not distress language — and the deterministic distress rollup lands at 42/100 ("some strain", stable trajectory, top driver: uncertainty language at 85/100). That combination is consistent with, not contradictory to, this brief's thesis: a company confident enough in demand to keep scaling paid media, but hedging on execution certainty — which lines up with an outside-in read of unmanaged conversion and AI-visibility risk rather than financial fragility.
05 · Economic stakes
What the leak costs, by year.
If true
If Datadog's conversion shortfall is driven by a combination of on-page friction (a lab-measured 7.76s LCP) and AI-answer invisibility (0% cited), then continued investment in top-of-funnel demand generation compounds the wrong lever.
Who loses
The trial-to-paid and expansion pipeline that net-revenue-retention — and by extension the multiple — depends on.
Year 1
Trial-to-activation conversion stays unverified against real-user load times, since no CrUX (real-user) data exists for this domain in this audit; AI-assistant omissions (0 of 3 probes) go unmeasured and unowned in parallel.
Year 2
As more technical evaluation shifts into AI-mediated discovery, the unmeasured 0% visibility gap compounds into fewer trials starting in the first place, independent of any on-page fix.
Year 3
A conversion-stage leak, if left unowned, hardens into a demand-formation gap owned by whichever competitor's name AI assistants surface more often.
Range note
Both leaks are directional in this audit: the lab LCP has no real-user (CrUX) confirmation, and the AI-visibility read is drawn from 3 probe queries. Sizing either in dollars requires first-party data outside this record.
06 · Blind spots
What the outside-in scan can and cannot see.
What the audit CAN see
- Funnel model: demand generated, computed leak at Conversion (Awareness 73 vs. Conversion 47).
- Lab-measured Largest Contentful Paint 7.76s and 4.7s of main-thread work (Google PageSpeed synthetic test), console errors, Technical Debt Index 30/100.
- GTM model: AI/generative visibility scored 0/100 (0 of 3 probes cited the brand).
What the audit CANNOT see
- Real-user (CrUX/RUM) page-load data — not available for this domain in this audit.
- Trial-start-to-activation conversion by performance or discovery-channel cohort.
- How often AI assistants actually name Datadog versus New Relic, Dynatrace or Splunk in live buyer queries.
Why it matters: the two data points that would settle which leak dominates — real-user load time at Datadog's actual traffic, and Datadog's true share of AI-assistant answers — are exactly what this outside-in audit cannot hold; treating the lab LCP as a stand-in for the field, or 3 probe queries as a stand-in for AI market share, would overstate this brief's certainty.
07 · Leadership questions
Five questions to put on the next board agenda.
1
Our funnel model shows demand generated and lost at Conversion — do we have real-user, not lab, data on how page load time actually correlates with trial activation?
2
When a technical buyer asks an AI assistant to compare observability tools, how often does the answer name us versus New Relic, Dynatrace or Splunk?
3
Who owns AI/generative visibility as a measured, board-reported outcome, and against what baseline?
4
If demand generation stayed exactly flat next quarter, how much additional revenue could we convert from the trial volume we already have?
5
The 7.76s Largest Contentful Paint driving our LCP-fix roadmap is a lab-measured, synthetic figure — has anyone validated it against real-user field data before we commit an engineering sprint to it?
08 · How the thesis could be wrong
Two ways to falsify our reading — and how to test them.
Assumption: the lab-measured 7.76s Largest Contentful Paint materially suppresses real trial conversion.
Disprove by Real-user (CrUX/RUM) data showing flat activation rates across fast- and slow-loading session cohorts.
Assumption: AI-assistant invisibility (0% cited) costs Datadog real pipeline.
Disprove by Evidence that AI-referred sessions are immaterial in volume or convert no differently than other channels.
Pattern CONVERSION_AI_VISIBILITY_GAP
Evaluated boardReady = true · 0 weaknesses · deterministic scoreBrief gate
Synthesis claude · gold-harness 3-pass · draft · critique · revise
Corpus context 1 of 793 board-ready briefs in the current Colleviate corpus of 7,401 audited companies
Note This is a public sample. Every requested dossier is bespoke to the target company and includes the peer-graph, evidence citations, and screenshots not shown here.
Get this on your company — or a target you care about.
Same engine. Same evidence bar. Board-ready in 5 business days. One-off $599. If you want continuous monitoring instead, the Intelligence Subscription starts at $399 per month for up to five companies.