Aoware

Custom Automation vs Zapier: When SaaS Isn't Enough

When custom automation beats Zapier or Make: three clear signals, the real cost of SaaS at scale, and a concrete lead routing example you can copy.

Aoware

Custom automation vs Zapier: when SaaS isn't enough

Your team already has HubSpot, Salesforce, Notion, Slack, and probably a Zapier account someone set up two years ago. The question isn't whether to automate. It's whether the glue holding it together can survive the next twelve months.

The "we already have Zapier" illusion

When a sales director tells us "we're already automated," what they usually mean is: there are 60 to 200 Zaps running, three people know what some of them do, and nobody touches them out of fear. That's not automation. That's accumulated technical debt with a friendly UI.

Zapier and Make are excellent at what they were designed for: connecting two tools with a clean, linear flow. New row in Sheets, send a Slack message. New form submission, create a HubSpot contact. That works. It will keep working.

The problem starts when your real process isn't "A triggers B." It's "A triggers B, but only if C and D match, and if it's a Tier 1 account it goes to the SDR team with a 30-minute SLA, and if the deal is reopened we have to reconcile data across two CRMs without duplicating the contact." No template covers that. And building it inside a SaaS automation tool means stacking filters, paths, webhooks, and code steps until nobody understands what runs on Mondays at 9am.

When SaaS automation is the right call

Before going further, the honest part: most companies should not replace Zapier or Make. If your process fits one of these patterns, stay where you are.

  • Linear flows between two or three tools. Calendly to HubSpot. Typeform to Slack. Stripe to Notion. Zapier nails this.
  • Volume under 2,000 monthly tasks. The Team plan at $69/mo covers it, according to Zapier's own breakdown.
  • The team that maintains it is non-technical and the cost of an error is low. A failed Slack notification isn't a billing incident.
  • The process changes every two weeks. A drag-and-drop builder beats a deploy cycle.

If you're in this zone, a custom build is overkill. We'd tell you that in the first call. Aoware doesn't sell projects that shouldn't exist.

The conversation changes when one of the next three signals shows up.

Signal 1: your process doesn't fit a template

Templates assume a generic workflow. Your business has rules nobody else has: how you score a lead, which exceptions trigger a manual review, how you split commissions when two reps touch the same account.

A real example we saw last quarter: a B2B SaaS company routing inbound leads. The official rule was "round-robin by territory." The actual rule was: round-robin by territory, unless the company has more than 500 employees (then it goes to the enterprise team), unless an existing opportunity is open (then it goes to the account owner), unless that owner is on PTO (then it bumps to their manager), unless the lead came from a partner referral (then a different rotation applies).

Building this in Zapier means seven nested paths, four lookup tables in Google Sheets, and a code step that nobody dares to edit. Building it as a small service that reads from HubSpot and writes back assignments is two weeks of work and one file to maintain.

The signal isn't complexity for its own sake. It's specificity. If your process is the reason you win deals, encoding it in someone else's template is a long-term mistake. It's the same dynamic teams face when they move from Excel, email and WhatsApp to a connected sales stack: the issue isn't the tools, it's that the rules holding them together don't live anywhere readable.

Signal 2: multiple tools are involved at the same time

The average sales rep juggles eight tools, according to Salesforce's State of Sales. The 42% of reps who feel overwhelmed by the stack are 45% less likely to hit quota.

SaaS automation tools solve this by adding a ninth tool. That's not always wrong, but it has a ceiling. A Zap that touches five systems is a Zap with five points of failure, five rate limits, and five sets of credentials to rotate.

When the process genuinely needs HubSpot, Salesforce, an ERP, Slack, and a custom database to coordinate, you have two options:

  1. Build chains of Zaps that call each other and pray the timing works.
  2. Build a service that orchestrates the five systems with proper retries, idempotency, and a single audit log.

Option two isn't more expensive than option one once you're past a certain threshold. Lowcode.agency frames the real breaking point well: the moment teams realize they have no git history, no PR review, no rollback, and no environment separation for the logic that runs their revenue operations.

The fix isn't to replace the CRM. It's to make sure the orchestration layer sits on top of a clean CRM that behaves as an invisible asset rather than a source of constant exceptions.

Signal 3: the team needs its own interface

This is the one most people miss. SaaS automation tools assume the user lives inside HubSpot, or Salesforce, or Notion. The automation runs in the background.

But sometimes your team needs a screen that doesn't exist anywhere:

  • A view that combines deal data from Salesforce with delivery status from the ERP and support tickets from Zendesk, filtered by account manager.
  • An internal tool where ops can approve, reject, or escalate flagged transactions with one click and full context.
  • A dashboard that shows the actual state of a multi-step onboarding across four systems, so customer success doesn't have to open four tabs.

You can't build that in Zapier. You can build it inside HubSpot's custom objects up to a point, then you hit the ceiling. A thin custom UI on top of the APIs you already pay for is usually the cleanest answer. The CRM stays the source of truth. The team gets the screen they actually need.

The real cost of SaaS automation at scale

Pricing pages look reasonable until you read the footnotes.

Zapier's Professional plan gives you 750 tasks/mo at $19.99. Team is 2,000 tasks at $69. Overages bill at 1.25x to 2.5x the base rate. AI actions count as 3x to 5x a normal task. A single Zap is capped at 100 steps, per Zapier's documentation.

Make uses a similar model: it counts operations instead of tasks, but once you blow past your plan, extras cost more per unit than what's included. The all-in invoice always lands above the headline price.

HubSpot has its own ceilings: Enterprise tops out at 1,000 workflows. Plenty for most companies. A real constraint for some.

None of these numbers are scandalous in isolation. The issue is the trajectory. A team that starts at 50 Zaps and 500 tasks/mo often lands two years later at 180 Zaps, 8,000 tasks, three Make scenarios for the edge cases, a Google Sheet acting as a database, and a monthly bill that's higher than a part-time engineer.

The cost isn't just the invoice. It's the time spent debugging silent failures, the deals that fell through because a Zap hit a rate limit on Friday night, and the institutional knowledge locked in someone's head. It's the same hidden cost that shows up when teams consider automating sales processes before hiring more reps: the bill never arrives as one big number, it arrives as a thousand small ones nobody books.

A concrete example: lead routing with scoring and handoff

Here's a process we rebuilt for a mid-market B2B client. Their stack: HubSpot (marketing), Salesforce (sales), Slack (handoff), and a homegrown scoring model in a Google Sheet.

The old setup was twelve Zaps:

  1. Form submission creates a HubSpot contact.
  2. A Zap looks up firmographic data from Clearbit.
  3. Another Zap pushes the contact to the Google Sheet for scoring.
  4. A scheduled Zap reads the score back.
  5. Three branching Zaps route based on score tier.
  6. Two Zaps create the Salesforce lead and assign an owner.
  7. One Zap posts to Slack with the handoff and a 30-minute SLA timer.
  8. A separate Zap checks if the SLA was missed and escalates.

It worked about 85% of the time. The other 15% was a quiet disaster: duplicate contacts in Salesforce, leads stuck in "scoring" forever, SLAs missed because the Slack message went to the wrong channel.

The custom replacement was a single service with one job: ingest a lead, score it, route it, and track the SLA. HubSpot stayed the marketing source of truth. Salesforce stayed the sales source of truth. The service sat between them with a small Postgres database for state, a queue for retries, and a thin web UI for the RevOps team to override routing when needed.

Result: 99.4% success rate on routing, average handoff time dropped from 12 minutes to 38 seconds, and the RevOps lead stopped spending Mondays cleaning up duplicates. Total build time: six weeks. Ongoing cost: lower than the previous Zapier + Make + Sheets bill.

The point isn't that custom is always better. It's that this specific process had all three signals: it didn't fit a template, it touched four tools, and the team needed an override UI. The same logic applies once you start layering follow-ups on top — see sales follow-up automation that doesn't sound like a robot for what happens when the orchestration layer can finally pass context between steps.

How to decide today: 4 questions

Before you call us or anyone else, run your process through these:

  1. Does the workflow fit a "when X, do Y" template, or does it have five exceptions you keep explaining to new hires? If it's the second, no SaaS tool will hold it cleanly.
  2. How many tools have to agree for one transaction to complete? Two or three: stay on SaaS. Four or more with state and retries: custom starts paying for itself.
  3. Does anyone on the team need a screen that doesn't exist in your current tools? If yes, you're not buying automation. You're buying an internal product.
  4. If the person who built the current automation left tomorrow, could someone else maintain it in a week? If the answer is no, you don't have automation. You have a liability.

Up to 94% of sales orgs plan to consolidate their stack, and reps still spend only 28-30% of the week actually selling. Consolidation doesn't mean fewer tools. It means fewer seams between them. That's where custom earns its place — not by replacing HubSpot or Salesforce, but by removing the duct tape between them.

What to do next

If your team is in the "two or three tools, linear flow" zone, keep your Zapier subscription and ignore everything anyone tries to sell you. You're fine.

If you recognized your process in any of the three signals — the template that doesn't fit, the five tools that have to coordinate, the screen your team keeps asking for — the next step is a one-hour conversation about what your actual workflow looks like, not a demo of what we build.

We build systems that fit how your company actually works, not the other way around. If that's the conversation you need, tell us how your process actually runs today and we'll tell you honestly whether custom is the right call — or whether you should stay exactly where you are.