The problems, in their own words.

By the time the customer starts using it, 3-4 months have gone by. They paid for a year; for 400k they have not got any value.

Most delays are not from the vendor side. Firewalls that cannot get opened, BRDs that cannot be shared, infra that was not provisioned before purchase.

The data we need is sitting in 100 different platforms, and there is no common sync between those data lines.

Implementation knowledge stays with the field delivery person, not captured anywhere.

It is happening because of lack of product adoption. And why? Because the product is hard to deploy.

The wrong person made a wrong decision early, and it is dragging the whole project.

Give me the data and I will roll this out in two weeks. These bottlenecks account for 40-50% of implementation delays.

Enterprise implementations go up from 3 months to 6 months to 12 months in some cases.

Around 50% of delays are due to customers not providing required information on time.

Requirements gathering is the most boring part of the job. I have a script of 10 standard questions I ask every client.

A lot of time my team spends creating different documents and artifacts: training manuals, guides, videos.

Name in one system: l.gupta. Name in payroll: lakshya.g. These discrepancies must be manually resolved.

Uploading data is not difficult. Cleaning that data so it fits the target schema is where all the time goes.

Two types of bottleneck. One because of humans: approvals, bureaucracy. Second is technical: data migration, cleaning, backup.

The management itself does not know why they are moving. The sales team just pitched everything.

In 4 sprints, 8 weeks, you should discover everything. But the people who own that are the ones slowing it down.

Is it a people issue? Is it a process issue? Is it a governance issue? Is it a technology issue?

A senior walks in, ego takes over, chaos ensues. We cannot predict when a senior will block a decision.

There is no standardisation. Every client is a custom build. That is 30-40% extra effort.

Sign-off cycles cause 15-20 day delays.

Deployment time is too much today. It is in months.

Professional services cost can equal or exceed the SaaS license: $70-80k consulting on a $100k deal.

One migration cycle: 45 days with repeated testing across local, staging, prod. Actual migration took 22-23 hours.

A migration failed inside a 24-hour downtime window and forced rollback.

Biggest time sinks: BRD, Jira stories, alignment, UAT. Dev and QA are not the bottleneck; documentation and alignment are.

Requirements evolve from 1.0 to 1.1, 1.2, but sometimes completely change direction. That causes massive rework.

Meeting AI tools stop at summaries. They tell you what happened but do not tell you what to do next.

Each migration requires four distinct roles in sequence: discovery, architect, implementation, validator.

Speed of implementation is directly related to how fast the customer can see value and to retention metrics.

Their bandwidth is shared. Someone has to prioritize this project, and that takes leadership push.

The knowledge base is misorganized. It needs to surface the right article at the right step.

Cloud cost estimates are rough guesses in sales pitch; actuals blow up during implementation.

POC-to-PS handoff needs full context preserved: what was promised, what data was used, what edge cases were found.

For a complex product and a big org, it could go from anywhere from three to six to maybe twelve months.

Lot of losses at handoffs... presales says customer paid for something, but now they want something else.

We have RocketLane, SFDC, Gong, Power BI... I try to bring all that into SFDC.

Discovery is something we are not able to optimize. Testing - we spend roughly 30% of time.

PS is expensive, customers pay $200-$400 an hour... we are always understaffed.

This world is 80% on-prem... can't give me cloud-only solution.

80% same agent, 20% configuration might be different.

I have to pay this guy for 120 days from my pocket.

If we could do a 70% integration in 2 weeks instead of 6-8 weeks, we'd win more deals.

First adoption is the most important part of professional services delivery.

If customer doesn't see ROI on first use case, he starts to lose interest.

Big banks... just to get one firewall rule, the cycle is almost a month.

Over a 30-day project cycle, people spend 12 to 20 hours building artifacts.

PS head with 2,000 people under him has no incentive to automate those people away. Control measured by how many people report to you.

Sales and implementation cycles are completely independent conversations.

Getting KT from client takes around a month... feedback cycles 1-2 months additional.

If they buy the product and we can't implement, they still pay - they lose value.

Implementation is the phase where you lose a customer... it starts there.

If I'm able to solve with two engineers instead of ten... company would be happy.

Time to value (TTV/TTR) is the north star metric for PSA and implementation teams.

Requirement itself is not clear. How would you solve it when you don't know problem?

What really would take time... what is my logical next step? Have not seen this in any AI tool.

25 status updates per week; Friday morning goes away in that.

Communication takes a lot of overhead... gather Loom videos, context from last week, prepare report, send to champion.

A customer can ask something three months later which he said on day one.

I know all things about all 4 projects. How will an AI have that much context?

I would say don't automate client-facing aspects for now, especially for high ticket products.

There was a lot of infosec that lasted around 3-4 months because they have a lot of security clearances.

People would still want someone there when it is getting deployed. The human touch is still required.

When a human leaves, context disappears.

It is very funny because we are all working on AI technologies, but they do not want to use AI technologies.

If we can bring the discovery phase down to seven to eight days, there's a 50% increase in efficiency.

Migrations are pretty tricky, and there's a lot of moving parts.

There's not a standardized path... we have to go step-by-step... a lot of manual steps in migrations.

Not all customers are happy giving us read-only access; we often just get a PDF bill and reverse-engineer.

It's hard to sell it to them... they see it as another integration they have to do.

Their VPs were using Claude Code to build some demo apps... 'why, if we can build it, then why so much time in getting these tools implemented?'

The QA is the largest workforce... there are a lot of QAs, almost like 30 QAs.

The client spent almost 5 million USD on this one.

If we can reduce the deployment cycle to a month, we can serve more clients and stay competitive.