Consumer 28 May 2026

Why We Built Our Own Operations Command Center Instead of Buying a CRM

We spent two years trying to make generic SaaS fit how a creative studio actually runs in India, Indian fiscal year, multi-currency, two-entity invoicing, scope builder, project tracking. We gave up and built our own. Here's exactly what we replaced, what we kept, and what we learned about Indian SaaS gaps along the way.

**We spent two years trying to make generic SaaS fit how a creative studio actually runs in India. Then we gave up and built our own, the command center we now run our entire business on, replacing a 6-tool stack with one source of truth. Here's exactly what broke, what we tried, and what we learned about Indian SaaS gaps along the way.**

This is not a "build vs buy" think piece. It's a specific account of one studio's two-year journey from buying generic SaaS to building a custom CRM + scope builder + project tracker + multi-currency invoicing + Claude-powered AI assistants. We're shipping the result as a public product soon. If you're a creative or service business operating in Indian FY with multi-currency invoicing across two entities, this is the build cost vs the rent cost math.

— — —

What we were running before

In 2022, NOW Media's operations stack looked like this:

The cost wasn't the problem. The fragmentation was.

A typical client engagement touched all six tools. Lead lands in HubSpot. Proposal drafted in Notion, exported as PDF. Contract sent via DocuSend. Payment terms recorded in Zoho. Project plan in Google Sheets. Status updates posted in Slack. By month three of any engagement, asking "what's the current state of client X" required opening four tabs and reconciling three different versions of the truth.

We tried to fix it. Twice.

— — —

The two failed attempts

Attempt one: consolidate on HubSpot

HubSpot is the most credible generic CRM for a services business at our scale. We upgraded to the Sales Hub Starter tier (₹3,500/month), built custom properties for our stage gates, imported our proposal templates into HubSpot Quotes, and tried to make it the operating system.

What broke:

**Indian fiscal year reporting**. HubSpot's pipeline reports default to calendar year. Indian businesses run April to March. Building custom reports that aligned to Indian FY required HubSpot's Operations Hub (₹50,000+/month). Out of scope.
**Multi-currency invoicing**. We bill in INR, USD, and AED. HubSpot Quotes are single-currency per deal. Workarounds existed but felt hacky.
**Two-entity invoicing**. NOW Media is a brand of Bleep Design Private Limited. Some engagements bill under one entity, some under another. HubSpot's standard product model doesn't accommodate this, you'd need duplicate pipelines.
**Indian rupee formatting**. HubSpot displays ₹100000 instead of ₹1,00,000. Minor, but every Indian financial document our clients see uses the lakh-comma convention. Looked wrong on every PDF we sent.
**GST integration**. Zero. Required separate Zoho workflow that didn't talk back to HubSpot.

We accepted the gaps and ran with HubSpot for six months. Adoption broke. The team kept reverting to Notion + Sheets because the workflows there matched reality better.

Attempt two: Pipedrive + Zoho + custom Zapier glue

After HubSpot we tried Pipedrive (better pipeline UX, similar gaps) with Zoho Books for invoicing and a layer of Zapier automations to keep them in sync. Net stack:

Pipedrive for sales pipeline
Zoho Books for invoicing (Indian FY native, GST native, multi-currency)
Notion for proposals and knowledge
Google Sheets for project phase tracking
Zapier for cross-tool sync (~15 zaps)
Slack for notifications

The Zapier glue worked, for a while. Then a Pipedrive API change broke a critical zap, we missed three invoice deadlines before noticing, and the Zoho-to-Pipedrive sync started double-creating contacts because the dedup logic was string-match instead of email-canonical. We spent two weeks unfucking the data.

The Zapier-glue approach gives you the surface of one system. It gives you the failure mode of N + glue. When the glue breaks silently, you don't notice until a client emails asking why they got two different invoices for the same project.

— — —

The decision

In late 2024, after the second cleanup, we made the call: build our own.

Three factors tipped it:

**The marginal generic CRM was getting worse, not better, for Indian creative practices.** Every SaaS round-tripping through the US-first product roadmap was adding features for US-first workflows. Indian FY, multi-currency with INR-default-AED-USD secondary, two-entity invoicing, none of these were on anyone's roadmap.
**We were already shipping custom internal tools for clients.** Building our own dashboard was within our service offering. The economics worked: 8 weeks of engineering time = ~₹15L of equivalent client engagement cost, but the resulting tool would save us ~₹2.16L/year in SaaS spend forever and (more importantly) collapse the 6-tool stack into one source of truth.
We could productize it later. Other Indian creative studios, accounting firms, law practices, and consulting groups face the same gaps. Building for ourselves first, productizing once the operational pattern was proven, that's the strongest go-to-market story.

We started the build in November 2024. The command center has been the single source of truth for NOW Media since March 2025.

— — —

What we built

The operations command center has 9 core capabilities:

1. Sales pipeline + scope builder

CRM with stage gates (Prospect → Pitched → Negotiation → Won → Lost), plus a proposal builder that ships shareable public scope URLs (the same one prospects use at nowmedia.in/#scope). Each public scope has view tracking, expiry, and a kill switch, so if a scope shouldn't be live anymore, one click removes the URL.

This replaced HubSpot's pipeline + Notion's proposal template + DocuSend's PDF sending.

2. The Discovery Blueprint

Adaptive AI-augmented client intake that kicks off every engagement. The questionnaire reads scope and only asks what's relevant. Auto-saves. Becomes the single source of truth that strategist, designer, copywriter, and engineer all work from.

This was the workflow innovation that didn't exist anywhere, there's no SaaS for "structured client brief that the whole team works from instead of a Notion doc that gets reinterpreted three times."

3. Project tracking (Pending → Invoiced → Paid)

Phase-by-phase delivery with scope checklists and retainer support. Each project has explicit payment phases tied to milestones, when phase 2 deliverables ship, the system surfaces the next invoice to be generated rather than waiting for someone to remember.

This replaced Google Sheets project trackers + manual invoice prompts.

4. Multi-currency invoicing (INR / USD / AED)

PDF extraction from Google Drive (we import paid PDFs and reconcile them automatically), Zoho sync (the Zoho integration finally goes one-way + automatic instead of bidirectional fragile), lifetime-value rollups per client.

This replaced Zoho's native UI for the day-to-day cases (we still use Zoho for GST returns, which it does well).

5. Indian fiscal year revenue targets

Year / quarter / month granularity with smart pro-rating. If you set a ₹4Cr annual target on May 15, the system pro-rates the May target to account for the 14 days already past. Q1 target = Apr+May+Jun pro-rated. Standard for any Indian business operator and absent from every generic SaaS we evaluated.

6. Intelligence nudges

Daily sweeps flag: stale leads (no contact in N days), won-without-project handoff gaps (deal marked Won but no project spun up), high-value silent deals (>₹10L deal that hasn't moved in 2 weeks), scopes not opened (proposal sent but never viewed), prospects nearing close date.

This replaced the manual "what slipped through?" weekly review.

7. Multi-channel notifications

In-app bell, Slack channel posts, weekly recap emails. Per-user opt-in. Different events route to different channels, a Won deal goes to the team Slack; a stale lead goes to the BD lead's email; a payment received goes to the founder dashboard only.

8. AI assistants powered by Claude

Three flows in production:

**Client health summary**: pulls the last 30 days of activity for a client (touches, deliverables, invoices, sentiment from email threads) and writes a one-paragraph status read.
Draft follow-up email: given a stale lead, drafts a contextual follow-up message in your voice, not a template.
**Suggest next step**: given a project state, suggests the next operational action (typically "send invoice X" or "schedule check-in with Y" or "spin up handoff doc for Z").

Built on Claude (Anthropic) because the response quality on judgment calls, "should this lead get a follow-up or be moved to lost?", was meaningfully better than the alternatives in our testing.

9. Privacy cloak + role-based access

One-click revenue masking for over-the-shoulder demos and screen shares. When the cloak is on, all currency figures display as XX.XX (preserving column widths). Useful when sharing your screen in a client call and you don't want them seeing your other clients' deal values.

Role-based access at Founder, Admin, Manager, BD, Accounts, CRM-only levels, with per-page permission overrides.

— — —

The build cost vs the rent cost (real numbers)

Build cost (engineering time only, no infrastructure):

Engineering: ~280 hours across the build
At our internal blended rate of ₹4,500/hour (loaded cost including overhead): ₹12.6L equivalent
We built it ourselves so the cash cost was zero, opportunity cost is what's relevant
Ongoing maintenance: ~10 hours/month = ~₹54K/month in equivalent opportunity cost

**Rent cost we replaced**:

HubSpot Sales Hub Starter: ₹3,500/month
Zoho Books (kept): ₹6,000/month
Notion paid: ₹2,000/month
DocuSend: ₹2,500/month
Slack paid + Zapier middle tier: ₹4,000/month + ₹2,000/month
Various small tools: ~₹2,000/month
**Total rent**: ~₹22,000/month = ₹2.64L/year

3-year math:

Build path: ₹12.6L upfront + ₹54K/month × 36 = ₹32.04L total
Rent path: ₹22K/month × 36 = ₹7.92L total
Build is **4× more expensive over 3 years on direct costs alone**

So on direct cost the rent path wins. The reason build still made sense for us:

**Productizing the result**. The command center is becoming a public product. The build cost gets amortized across other customers, not just NOW Media. The rent path produces no asset.
**Hidden costs of fragmentation**. The 6-tool stack cost us roughly one founder-hour per day reconciling data between systems. At a founder rate of ₹10K/hour, that's ₹2.6L/month of hidden cost we weren't tracking. The build collapsed that to zero.
**The Indian FY + multi-currency gaps would have gotten worse, not better.** Every year on the rent path, we'd have spent more on glue (Zapier, custom Apps Script, manual reconciliation) as the gap widened.
**Operational discipline became the moat.** When prospects ask "how do you run NOW Media?" we can show them. The command center is now part of our pitch.

If you're a creative or service business with under 20 people, mostly single-currency, and no specific Indian-FY or multi-entity constraints, **buy. Don't build.** Our cost case only worked because we had specific structural mismatches with generic SaaS plus a productization upside.

— — —

What we kept buying (and why)

We did not build everything. We kept buying when the off-the-shelf option was genuinely better:

**Zoho Books for GST returns.** GST is regulatory. Zoho already runs the entire compliance workflow including filing. We sync data from our system into Zoho, but Zoho is the system of record for GST.
Google Workspace. Email, drive, calendar. No upside to building.
Slack. Team comms. Network effect with vendors and clients makes alternatives impossible.
**Vercel for hosting.** The Vercel + Next.js stack is what we ship on every client engagement; no point switching for our own platform.
**Sanity for content.** Same reason.
Plausible + GA4 + Meta Pixel. Standard analytics. Build wouldn't differentiate.

The build/buy line for us: **build where structural Indian-context mismatch + productization upside exists. Buy everywhere else.**

— — —

What we learned about Indian SaaS gaps

Three categories of gap stand out:

Fiscal calendar

Every Indian business operator running an Indian-incorporated entity needs April-March fiscal year alignment. Generic SaaS that lets you set a custom fiscal year via admin settings, fine. SaaS that hardcodes January-start year-over-year comparisons, broken for the entire Indian market. The latter is more common than the former.

Currency formatting

"Lakh and crore" is the dominant way Indians write large numbers. ₹1,00,000 (one lakh) and ₹1,00,00,000 (one crore) are the conventions. Most US-first SaaS displays ₹100000 and ₹10000000. Functionally readable but visually wrong on every Indian-facing document.

Multi-entity by default

Indian businesses commonly operate multiple legal entities, one for services revenue, one for product, one for IP holding. Generic CRMs assume one company per account. Workarounds exist (duplicate pipelines, complex tagging) but they're workarounds, not first-class support.

If you're an Indian SaaS founder building horizontal tools, these three gaps are still wide open in 2026. The reason they haven't been filled: the Indian market alone is hard to justify a product against, and the global product roadmap goes to the bigger market first. The opportunity for India-first vertical tools is real.

— — —

What's next

The command center is being productized for external availability. We're starting with creative studios + design agencies + small consultancies, the exact ICP of practices that face the same fragmentation we did. Custom builds for clients (₹15L to ₹30L, 8 to 16 weeks) are available today under the AI Automation service. When the productized version is ready, existing custom-build clients get migration paths.

If you're operating with a fragmented stack and the Indian-FY + multi-currency + multi-entity constraints sound familiar, start your scope, happy to do the same build-vs-buy math for your situation.

— — —

FAQ

When does it make sense to build your own CRM?

Two conditions together: (1) generic SaaS has structural gaps for your specific operating context (in our case: Indian FY + multi-currency + multi-entity), and (2) you can productize the result so the build cost amortizes across more than one customer. If only condition 1 is true, the rent path is almost always cheaper over 3 years. If only condition 2 is true, you're building a startup, not solving operational fragmentation.

How long does building a custom operations platform take?

Our command center took ~280 engineering hours across 8 weeks of focused build, plus ongoing maintenance of ~10 hours/month. For a client custom build (which doesn't require productizing), we scope similar projects at ₹15L to ₹30L over 8 to 16 weeks depending on integration complexity (Zoho, GST, banking, multi-entity setup).

What CRMs did you consider before building?

HubSpot (Sales Hub Starter and Professional), Pipedrive, Zoho CRM, Salesforce Essentials, Close, and notion-as-CRM patterns. HubSpot Professional + Operations Hub was the closest fit but required ₹50K+/month in subscription cost plus significant custom-property work to handle Indian FY reporting. The cost-to-fit ratio didn't work.

Why Claude over OpenAI for the AI assistants?

We tested both. For the specific tasks in our command center, client health summaries, judgment-call follow-up suggestions, contextual email drafts, Claude's responses were meaningfully better on the judgment-heavy work. For more straightforward tasks (data extraction, classification), both were equivalent. Claude won on cost-quality ratio for our specific use case. We re-evaluate quarterly.

Is the operations command center available to buy?

Coming soon as a public product. Available today as a custom build for clients under the AI Automation service. The custom build is built specifically for your business workflow, your fiscal calendar, your currency mix, your team roles, your operating discipline, rather than retrofitted. NOW Media's own command center is the reference build you'd point at.

How do you handle GST and filing?

We kept Zoho Books for GST compliance and filing, Zoho is the system of record for tax compliance. Our command center syncs financial data to Zoho via API rather than replacing it. Building GST compliance from scratch would have been months of regulatory work for no business advantage; Zoho does it well.

Did you consider building on top of Salesforce instead of from scratch?

Yes. Salesforce + Indian-FY + multi-currency + multi-entity customizations would have required Apex development at roughly the same engineering cost as building from scratch, plus ongoing Salesforce licensing (~₹6,500+/user/month for Sales Cloud Professional). We would have inherited Salesforce's product complexity without solving the underlying mismatch. Build-from-scratch had a clearer path to a working tool.

— — —

*Built by NOW Media, a Bangalore creative studio founded in 2019 by Nithin Koshy and Divya Maben, a brand of Bleep Design Private Limited. The operations command center is one of two products we have shipped, alongside Brandauditor.ai. Start your scope or view the AI Automation service.*

Internal links (add these in Sanity)

Pillar: /services/ai-automation, "AI Automation service"
Related: /products, "our two shipped products"
Related: /methodology, "the Discovery Blueprint"
Related article: /thinking/how-brandauditor-hit-first-chatgpt-citation-in-2-days, "the LLM citation playbook from Brandauditor.ai"
Scope CTA: /#scope, "start your scope"
Author byline: /about#nithin-koshy