Cloudflare
Each of these is a real decision Cloudflare is working through, written so you can make the call yourself — incomplete information, someone who pushes back, no answer key.
Cloudflare is a cloud infrastructure and connectivity provider operating a global network between users and Internet applications, delivering integrated application and network security, performance, reliability, and developer infrastructure. The network spans 330+ cities in 125+ countries; Cloudflare sits in front of more than 20% of the web; its software platform serves 115 million HTTP requests per second on average. Cloudflare provides authoritative DNS for millions of Internet properties and operates 1.1.1.1, a public recursive resolver, together handling around 3 trillion DNS queries per day. Tens of millions of certificate operations run daily across products like Universal SSL, SSL for SaaS, Origin CA, Cloudflare Access, and Cloudflare Tunnel. It serves individual developers, startups, small businesses, and large enterprises such as Shopify and VISA that need to protect and accelerate applications or secure employees, devices, and networks. Cloudflare One is its SASE platform; its threat operations and research team, Cloudforce One, identifies and disrupts cyber threats from sophisticated cyber criminal activity to nation-state sponsored advanced persistent threats.
Your first one is free. No card.
52 live right now.
The rest
Prove policy migration in a pilot without adding another console
Network security administrators cannot yet show which access policies survive a pilot rollout. You must weigh self-service speed against the controls needed to trust the migration evidence.
Commit the access cutover without stranding distributed employees
Valid employees are being denied while the old policy still permits a credential-abuse path. You must weigh a security closure against proof that authorized access will hold.
Block releases for Tunnel routes before retired fields fail lookups
Tunnel changes are reaching progressive deployment with retired route fields and then failing route lookups. You must weigh strict pre-release blocking against keeping legitimate connectivity changes moving.
Carry each upload into processing without a pipeline rebuild
The customer can store media in R2, but cannot yet carry each upload into its processing flow. You must weigh a fast migration path against an operating model their engineers can trust.
Prove retrieval of archives in 12 days or renew the old store
Some archived files appear in R2 while downstream jobs still receive missing-object responses. You must weigh a time-bound cutover against evidence that the production retrieval path is safe to trust.
Fix links that expire in the middle of long videos before cutover
Private video downloads expire before some viewers finish playback. You have to weigh short-lived access controls against a download path viewers can actually complete.
Rank accounts by renewal risk without turning Slack gaps into claims
R2 account history is split across usage, tickets, commitments, and Slack, so renewal risks surface late. You must weigh faster account preparation against evidence that a Customer Success / Solutions Lead can defend in conversation.
Prove R2 is more than a bucket while platform leads trim vendors
Developers see R2 as a bucket, so Iceberg interest does not yet become a committed workload. You must weigh a focused programme against the need to show a broader developer platform path.
Choose a design compliance can defend while self-hosted support ends
A regulated archive migration is paused because retention and recovery evidence do not yet support one shared decision. You must weigh a near-term storage exit against claims the customer can defend in an audit.
Route campaign media through Workers before Friday’s traffic spike
Campaign-preview users are seeing media failures through the public bucket URL. You must weigh a fast cutover against enough validation to protect the campaign.
Route domain alerts without losing the evidence behind severity
Suspicious-domain alerts are spending too long in context gathering before an engineer can set severity. You must weigh a faster route against the evidence engineers need to trust each suggestion.
Structure discovery evidence before technical buying windows close
Discovery records are leaving sellers unsure which AI Gateway concern the buyer actually raised. You have to weigh faster preparation against evidence a seller can stand behind.
Prove partners can originate deals without a generic reseller program
Prospects are hearing separate proposals for AI routing, security, and cost control. You must decide what a partner can credibly lead and what stays in the direct technical conversation.
Decide whether to sign the platform deal despite open residency gaps
A $2 million platform agreement is stalled because the buyer cannot reconcile production AI traffic with its residency requirements. You must weigh a credible multi-product commitment against terms that will survive technical and risk review.
Close unified billing while procurement questions routing costs
An active gateway deployment could be displaced before a $180,000 annual commitment reaches procurement. You have to weigh flexible token demand against a commercial structure the buyer can approve.
Trace overload causes while live sessions keep failing
Busy Durable Objects can slow down or reject turns while adjacent objects remain healthy. You must decide what evidence AI can assemble and what an engineer must verify before a fix moves forward.
Make reconnects safe when live sessions lose their last event
Developers are replaying uncertain events after disconnects, which can duplicate or lose session state. You have to weigh a simple recovery API against the runtime guarantees needed to make it trustworthy.
Contain match overload without breaking ordered state
Overload errors are disconnecting players and leaving some match turns uncertain. You must weigh immediate queue relief against the evidence needed to trust state recovery.
Trim wake-up delay before live sessions feel stalled
Objects idle for more than 30 seconds add visible delay before customer JavaScript starts. You must weigh a faster wake-up path against preserving storage and request ordering.
Rank renewal risk when telemetry and commitments disagree
Renewal risks are surfacing only after AI platform engineers ask why adoption or costs changed. You must weigh early ranking against evidence that account owners can verify before speaking with customers.
Prove pilots can justify spend in 90 days without chasing lower bills
Pilot customers cannot yet connect shared-wallet spend to provider and application decisions. You have to weigh a reusable programme against each customer’s different governance needs.
Set routing rules while fallback answers and wallet spend diverge
Fallbacks are changing customer-facing answers while the shared wallet drains faster than forecast. You must weigh resilient routing against the evidence finance and security need to approve it.
Separate bad credentials from outages before retries waste more tokens
Rejected credentials are being retried as if providers are unavailable, creating repeat failures and avoidable spend. You have to weigh immediate traffic continuity against a retry rule that correctly protects credentials and budgets.
Route resource-limit alerts with evidence so severity calls agree
Resource-limit alerts are reaching responders without the same evidence, so severity calls diverge. You must weigh faster routing against the proof an engineer needs before acting.
Structure discovery notes before technical follow-up loses the buyer
Discovery details are being reconstructed after calls, leaving technical buyers to repeat architecture and migration constraints. You must weigh faster seller follow-up against source-linked review before anything becomes a customer promise.
Rank causes of Worker failures with evidence while engineers decide
Recurring Worker failures leave responders with competing explanations and application developers without a clear path to recovery. You must weigh faster investigation against the evidence an engineer needs before acting.
Find connections that hang in migrated apps without leaking credentials
Migrated applications can hang when their existing database connection pattern meets request-time execution. You must weigh useful diagnosis against customer-data exposure and a narrow first release.
Commit recovery rules before durable state crosses the migration point
Concurrent writes may reappear in the wrong order after failover, despite acknowledged updates. You must weigh migration progress against proof that recovered state remains correct.
Restore Queue delivery before delayed jobs pile up
Messages are waiting minutes after entering affected Queues despite healthy consumer Workers. You must weigh faster delivery against the batching behavior and delivery guarantees customers rely on.
Build a review kit for connected Workers without touching production
Reviewers can inspect a page today, but cross-Worker checkout and account flows still reach production bindings or go untested. You must weigh a narrow pilot kit against a broader pattern the customer can operate after the engagement.
Prove the ordering of claim updates before Monday's cutover
Seventeen of 24,000 concurrent claim sequences show possible out-of-order updates before Monday’s cutover. You must weigh a deadline-bound migration against evidence that has to stand up in a claim dispute.
Recover checkout speed without undoing the new key namespace
Checkout p95 latency is 690 ms after origin eligibility reads rose from 3% to 19%. You must weigh a reversible Worker-side pattern against a rollback that protects today’s change window.
Draft a timeline for failed puts in 30 minutes so engineers verify
Intermittent puts leave application developers waiting while on-call engineers reconstruct evidence across the request path. You must weigh faster incident context against the risk of treating an AI-generated causal story as fact.
Prove object integrity without slowing the R2 request path
Developers lack a supported way to verify that retrieved R2 objects match what they uploaded. You must weigh integrity evidence for pilot workloads against compatibility and latency on every request.
Choose a repair for uncertain metadata before the next traffic shift
Some applications see a recent object through one request and an older view through another. You must weigh faster regional availability against a recovery path that proves committed object state remains correct.
Separate upload retries from new pressure without weakening limits
Avoidable 429 responses are forcing multipart upload clients to retry the same object parts. You must weigh smoother uploads against the bucket protection that keeps shared request capacity stable.
Sign partners for migration deals without unapproved discounts
Migration conversations are producing developer interest but not enough enterprise evaluations. You must weigh a focused first partner package against a broader motion that could dilute qualification.
Commit platform terms despite a 38% usage gap
A $2.5 million agreement rests on usage forecasts that differ by 38% across customer models. You must weigh platform scope against contractual exposure that cannot be cheaply reversed.
Win procurement approval this quarter instead of deferring to renewal
Workers AI and R2 usage rose 41%, but procurement has delayed the expansion twice. You must weigh predictable commitment against the customer’s need to keep scaling.
Keep failover within an allowed geography while showing the cost
Developers moving regulated state to Durable Objects lack a testable way to constrain failover geography. You have to weigh strong locality controls against the availability customers surrender during regional impairment.
Choose an enforcement line for phishing while 146 routes stay unproven
Thirty-one confirmed domains point to an active phishing campaign, but 146 related routes have incomplete attribution. You have to weigh rapid disruption against a false-block boundary that customers cannot easily escape.
Stabilize cold starts while the cache window is still open
Cold-start failures rose from 2.8 to 28 per 10,000 requests in one location group. You have to weigh a quick rollback against protecting isolate capacity with a stricter admission guard.
Build a partner route for sovereignty reviews while claims await review
European self-serve accounts are stopping before enterprise discovery as sovereignty questions multiply. You must weigh regional credibility against a package that stays truthful about the platform.
Commit terms for the next ingest without setting a blind precedent
A regional write failure paused a three-year AI agreement 10 business days before the next ingest. You must weigh customer continuity against terms that could shape every similar negotiation.
Convert media usage before procurement renews its storage commitment
R2 reads rose 240%, but two launch-day incidents have stalled the enterprise expansion. You must weigh a fast commercial close against a proposal the CTO and procurement lead can verify.
Predict runtime regressions while developers still control the release
Eighteen percent of severe latency regressions were first detected after deployment. You must weigh early warnings that developers trust against alerts that slow ordinary releases.
Contain hostile context before account actions reach more cases
Eleven of 240 reviewed traces contained retrieved text that tried to redirect a tool call. You must weigh wider automation coverage against controls that people can actually verify under load.
Trim repeat retrieval before the morning queue backs up
P95 first response has risen from 1.1 to 2.4 seconds after redundant retrieval calls appeared. You must weigh a faster cache path against session-safe context and answer quality.
Keep evidence in Threat Events so analysts stop copying it into notes
Analysts are copying source evidence into separate notes because Threat Events lack it at the point of use. You have to weigh a simple API against enough provenance for a defensible mitigation decision.
Publish phishing domains in 6 hours without blocking real customers
The cluster contains seven times more related domains than yesterday, but some links are inferred. You have to weigh campaign disruption against the cost of blocking a legitimate customer destination.
Drain IOC lag before fresh indicators miss investigations
During bursts, only 46% of imported IOCs are queryable within 10 minutes. You have to weigh faster draining against the stability of the shared enrichment flow.