Gartner's 2024 FinOps practitioner survey ran one of the least-read but most useful data points of the year: across 412 respondents spanning 23 countries, 73% of cost-optimisation recommendations from native cloud advisors ā AWS Trusted Advisor, Azure Advisor, GCP Recommender ā remained unactioned for 90+ days after they were surfaced. The median time-to-action for the remaining 27% was 42 days. For the Indian mid-market accounts we've audited in the last year, the ignored rate is closer to 80%.
This isn't because teams don't see the recommendations. They see them. They click through them. They nod. And then nothing happens. We've watched this pattern in twelve Indian mid-market audits in a row: the Trusted Advisor dashboard lights up like a cricket scoreboard, showing ā¹1.5ā4 lakh per month of potential savings ā and the account proceeds as if the dashboard isn't there.
The interesting question is not "why is AWS Trusted Advisor bad" (it isn't ā the recommendations are fine). The interesting question is: what structural property of a mid-market engineering org makes ignoring the recommendations cheaper than actioning them? That's what this post unpacks. Seven reasons. Then a founder-led framework to fix it. Then a real-audit pattern where a 50-person fintech found ā¹3.2 lakh/month of ignored recommendations sitting in their dashboard for 14 months.
The seven reasons recommendations get ignored
1. No named owner
The Trusted Advisor dashboard is visible to everyone on the account. Which means, in practice, no one. "Someone should look at that" is the universal response when we ask who reviews it. Shared ownership without a named human is the oldest failure pattern in engineering org design, and it applies to FinOps signals exactly as hard as it applies to production alerts.
The fix is not more visibility. It's less. One person owns Trusted Advisor review. Their name is on it. They report monthly. They have the authority to action or defer each recommendation with a written rationale.
2. Political cost of changing another team's resource
Trusted Advisor flagged an oversized db.r5.xlarge owned by the data team. The platform engineer who reviewed it does not have the authority to resize it. She files a Jira ticket. The data team has a Q2 roadmap full of real features. The ticket sits. Six months later the recommendation is still there, still correct, and the rightsizing is now politically awkward because "you've been telling us about this for months".
The fix is organisational, not technical: the Trusted Advisor owner needs a direct escalation path to the CTO or VP Eng with a 30-day SLA on cross-team remediations. Without the escalation, every recommendation that crosses a team boundary dies in the gap.
3. Unclear revenue or business impact of the fix
"Rightsize this m5.2xlarge to m5.large, save ā¹18,000/month" is a number, not a business case. An engineer weighing whether to action it needs to know: is ā¹18,000/month meaningful against our gross margin? Does the founder care? Will finance notice? For a ā¹6 crore/year SaaS with 70% gross margin, ā¹18,000/month is meaningful ā about 0.3% of revenue, which beats most feature work's contribution at the margin. But nobody in the engineering org has ever been told that. So the recommendation feels small.
The fix is translation. The Trusted Advisor owner, once a month, converts the rupee savings into percentage of revenue, percentage of cloud bill, and implied engineer-month equivalent. Then it lives on a slide the founder sees.
4. The recommendation is stale by the time anyone reads it
Trusted Advisor refreshes most cost checks every 24 hours. That is fast enough. But the review cadence in most mid-market teams is not ā the dashboard gets looked at during the annual AWS account review. By then, the workload has scaled twice, the recommendation is now wrong in three directions, and the engineer who opens it spends 40 minutes determining that none of the current listings apply. Next year, same thing.
The fix is a standing monthly review, not a quarterly or annual one. 30 minutes. Same day of the month. Recurring calendar invite. Attendees: the Trusted Advisor owner, one finance stakeholder, one engineer with commit access.
5. Aggregated totals without per-team breakdown
Trusted Advisor tells you "ā¹2.4 lakh/month of potential savings". It does not tell you which team owns each recommendation. Without tagging + a per-team rollup, the ā¹2.4 lakh is an org-wide number that belongs to everyone and therefore no one. Per-team chargebacks convert a shared-pool number into a per-owner one, and per-owner numbers get actioned.
The fix is tagging discipline upstream. Every EC2 instance, RDS cluster, EBS volume, Load Balancer has owner + env + product tags enforced at deploy time via IaC + AWS Config. Trusted Advisor output then filters to per-team dashboards, and each team lead sees their own ā¹ number.
6. No governance cadence
Trusted Advisor review is the kind of work that disappears under a sprint planning session. If it's not on a calendar, it doesn't happen. If it's on a calendar that gets cancelled when a production incident hits, it doesn't happen either. The governance fix is simple but requires leadership air cover: the monthly review is protected. Only a SEV1 displaces it.
7. No rollback plan for the remediation
Engineers are trained ā correctly ā to treat every change as a potential production risk. Trusted Advisor says "delete 38 unattached EBS volumes to save ā¹8,400/month". The engineer responsible knows that one of those volumes might contain data that someone, somewhere, thought was backed up. Without a documented rollback, the safe move is to not action it. And the safe move is what gets chosen.
The fix is to productise remediation: a standing runbook per recommendation type. Snapshot before delete. 30-day soft-delete window. Written rollback path. Once the process is cheap, actioning is cheap.
The founder-led framework to fix it
You don't need a FinOps platform to fix this. You need three things:
- One named owner , 4 hours/month, reports to the founder or CTO.
- A ROI threshold ā the line below which a recommendation is auto-dismissed (we default to ā¹5,000/month for Indian mid-market; above the threshold, the recommendation must have an action decision within 30 days).
- A standing monthly review , 30 minutes, three slides: top 10 recommendations by ā¹, decisions on each, recurrence rate.
That's the whole system. It fits on one page, it runs on any cloud, and it moves the actioned-recommendation rate from under 20% to above 70% in two quarters of running it. Not because the system is clever. Because the system exists.
The real audit pattern: ā¹3.2 lakh/month ignored for 14 months
This is a composite example based on audit patterns across several Indian mid-market fintech clients. Specific figures approximate the modal case we see.
A 50-person Indian fintech, Series A, ā¹9.4 lakh/month AWS bill, mostly in ap-south-1 with a secondary us-east-1 for analytics. Engineering team of 22. No named FinOps owner. Trusted Advisor enabled on Business Support; dashboard visited by two engineers, historically, when the bill spiked and the founder asked.
When we ran our free 24-hour audit in March 2026, the Trusted Advisor output alone showed:
- ā¹1,16,000/month of Reserved Instance and Savings Plan opportunities (42% coverage, with 18 months of stable baseline eligible for a 3-year commitment).
- ā¹68,000/month of low-utilisation EC2 instances ā four m5.2xlarge workers at median 8% CPU, three r5.xlarge databases reporting zero connections over the last 30 days.
- ā¹42,000/month of idle Load Balancers, 11 of them, from staging environments that had been torn down without the ELB cleanup step.
- ā¹24,000/month of unassociated Elastic IPs ā cost per unattached EIP is small but they accumulate.
- ā¹71,000/month of storage-tier recommendations ā S3 buckets holding analytics raw data in Standard that had not been read in >120 days, eligible for Intelligent-Tiering or Glacier.
Total: ā¹3.21 lakh/month. Of the recommendations, 62% were older than 9 months. Two were older than 14. When we walked the engineering lead through the dashboard, his response was genuine: "We knew all of this was in there. Nobody owns it."
The remediation was not technical. It was process. We installed the framework above: one owner, ā¹5,000 threshold, monthly review. We helped structure the first two reviews. Within 60 days the team had actioned ā¹2.38 lakh/month (74% of the identified total), deferred ā¹47,000 with a written architecture reason, and dismissed ā¹36,000 as not-applicable-at-current-scale. The Trusted Advisor dashboard, which had been a scoreboard of shame, became a governed queue.
The engagement paid for itself in 11 days. The customer is now on a ā¹50,000/month gain-share retainer where we continue to own the monthly review cadence against a frozen baseline. Verified savings through month three: ā¹2.14 lakh/month.
What a useful Trusted Advisor practice looks like at Indian mid-market scale
- One owner, one meeting, one threshold. Not a platform. Not a consultant on call. Not a Slack channel that nobody reads.
- Monthly rupee rollup to the founder. Three lines of a CFO slide: "identified this month", "actioned this month", "cumulative run-rate saved YTD".
- A documented reason for every ignored recommendation. Not to shame anyone; to avoid re-reviewing the same dismissal five times a year.
- A rollback runbook for the top 5 remediation types. Delete-with-snapshot, resize-with-revert, retag-with-audit, terminate-with-30-day-hold, downgrade-storage-with-read-test.
- Escalation to the CTO at 30 days for any cross-team recommendation the owner can't action alone.
This is what a founder-led FinOps practice looks like at this stage. It doesn't need a platform. It needs a cadence.
Want us to run the first review with you?
Our Cost module installs this operating model in 2ā3 weeks. The free 24-hour audit shows you the ā¹ number you're sitting on before you commit to anything. If you already know the number and want the framework delivered turnkey, FinOps QuickStart is ā¹40,000 for the first three customers in our founding cohort, ā¹75,000āā¹1,00,000 thereafter, and includes the first 60 days of the owned monthly review.
Start your free audit ā aicloudstrategist.com/audit.html
Or, if you'd rather talk first, book a 30-minute Cloud Cost Health Check call.
Founder-led by Anushka B. AICloudStrategist is a founding-cohort FinOps consultancy for Indian mid-market companies (ā¹5Lāā¹50L/month cloud spend). First three customers at ā¹40,000 for a full FinOps QuickStart. We publish our numbers honestly ā including the ones that don't yet exist. See how we prove what we claim.
AICloudStrategist Ā· Founder-led. Enterprise-reviewed. Ā· Written by Anushka B, Founder.
