Connect what you already have: integrations before a CRM
Most owners are not ready for a system of their own. They are tired of typing the same customer into four tools. That is a smaller problem, and it has a smaller fix.
By Eugene Romanov·Published
The short answer
Start by connecting what you already run. The site form, the inbox and WhatsApp write into one record; QuickBooks, Stripe and the job list stop being retyped by hand; supplier prices and outside leads arrive by themselves. One connection is an AI Automation Sprint at $999, scoped in writing and live in 1-2 weeks on your own accounts. When you outgrow it, a CRM is built onto the same routes rather than replacing them.
SHT-02What drives the number
What not being connected actually costs
It rarely looks like a cost. It looks like a person typing the same customer into a second tool, an enquiry that sat in an inbox overnight because nobody had the inbox open, and two screens that disagree about what a job was worth. Count it before buying anything: how many times last week the same customer was typed twice, how many enquiries waited more than a day for a first reply, and how often someone asked which of two numbers was right. If all three counts are small, keep the money. If one of them is not, that one is your first connection.
The connections owners ask for first
The website form, so an enquiry becomes a record with its source rather than an email. The inbox, read over IMAP from the real mailbox, so a reply counts as a reply and not as a guess from a tracking pixel. WhatsApp and Telegram, so a message lands in the same record as the form. The phone, where the provider's status callback writes how the call ended onto the lead without anyone clicking. Stripe, with deposits landing in the owner's own account. QuickBooks, connected on the owner's own account through its own sign-in; in one of the systems here the module that writes the customer, the invoice and the payment into it is already part of the code. And Google Sheets, for the price list or the job list nobody wants to give up yet.
Data that should arrive by itself
Some of what a business runs on lives on other people's sites. In one build a supplier's whole catalogue, with its prices, descriptions and photographs, came into the shop by a script instead of being typed. In another, an industry directory came into the lead list deduplicated by phone and email, so nobody was called twice. Reviews and leads from marketplaces would come in the same way. What has to be written down first is the source, how often it is read, and what counts as a duplicate, because a feed without a duplicate rule fills a list with the same person three times.
Why glue tools break quietly
Glue tools such as Zapier or Make are the right answer for a two-step hop: a form arrives, a row is added. They start to fail where real operations live. A rule with a condition in it, such as deduct the stock only once the client approves. An event that fires twice, which payment and form providers do, and which has to change the record once, not twice. A field someone renamed in one tool and not the other. When that happens the failure usually arrives as an email nobody reads, and the record is wrong until someone notices on a screen a client is looking at. Code that is yours fails loudly, is tested, and lives in a repository you own.
How a connection is scoped as a sprint
One route per sprint, written down before anything is built: what comes in, where it goes, what it must never do, and what counts as done. One operational number is written down with its value today, for example time to first reply or the customers typed twice a week, and measured again once the route is live. Everything runs on your accounts and your API keys, so provider fees are billed to you and nothing is rented back through the studio. At the end it is in production, not a demo, and it is built so a later system uses the same route instead of throwing it away.
When to stop at connections, and when it is a system
Stop at connections when every tool does its own job well and only the hops between them hurt. It is a system when the thing that makes you money is a rule no tool models, such as your pricing math or the document that has to come out right, and when the workaround has quietly become the process. The move does not cost you your history. The importer written for one owner's spreadsheet of past estimates keeps his own estimate numbers, invents no won or lost status, and brings every row in archived as history until he decides otherwise. The old tool stays readable until you decide to turn it off.
SHT-03The options, side by side
What each route actually costs you.
| Option | Price | Timeline | The trade-off |
|---|---|---|---|
| A glue-tool subscription | Their published monthly plans, priced by task volume | An afternoon per hop | Fine for a two-step hop. Conditions, duplicate events and renamed fields are where it fails, and the failure arrives as an email nobody reads. |
| One connection as an AI Automation Sprint | $999 | 1-2 weeks | One route, a written scope and one number measured before and after, on your keys. The right size when one hop is the leak. |
| A system that already runs, fitted to you | From $1,500 | 1-2 weeks | When your trade is close to one already built for, the record and its connections come with it, shaped to your stages and documents. |
| A CRM of your own | From $2,500 | 30-45 days | When the thing that makes money is a rule no tool models. Built onto the routes you already connected, not instead of them. |
SHT-04In this hub
How much does a custom CRM actually cost?
The commonly quoted range describes a different kind of company building a different kind of system. Here is the arithmetic underneath it.
Should you build a custom CRM or buy one?
Start from buy. Make the build argue its way in. Here is the argument it has to win, and the cases where it genuinely does.
How long does it take to build a custom CRM?
The schedule is not decided by how fast anyone types. It is decided by how quickly questions get answered and what state your data is in.
Who owns the code and the database when someone builds your system?
The answer that matters is not in the contract's intellectual property clause. It is in what stops working the day you stop paying.
Moving a business off spreadsheets: what actually has to happen
Nobody has ever been stopped by the number of rows. What stops migrations is that half the columns mean something only one person knows.
Custom CRM vs Jobber: when each one actually wins
Written by someone who builds the custom side, which is why the answer below is the product more often than not. The interesting part is the line where that flips.
How does AI lead scoring actually work inside a CRM?
The honest version. What a vendor means by a lead score, what is actually in the repository, and the four questions that separate the two.
SHT-05General notes
The questions people actually ask.
- N01Do I need a CRM to connect my tools?
- No. Most of the pain is in the hops between tools that each work fine on their own. Connecting them is a smaller job than replacing them, and it is the usual place to start.
- N02Will I lose what I built if I move to a CRM later?
- No. A later system is built onto the same routes, so the connection you paid for keeps working. Your history moves too: clients, deals, notes and files are moved before the switch and counted on both sides.
- N03Which tools can be connected?
- Anything with an API on your own account: QuickBooks, Google Sheets, Gmail, Stripe, phone providers, the WhatsApp Business API, Telegram and your website form. A tool with no API is harder, and that is said in the scope before anything starts, not discovered halfway.
- N04How much does one integration cost?
- One connection is an AI Automation Sprint at $999, live in 1-2 weeks, with a written scope. Provider fees, such as message volume or API usage, are billed to your own accounts.
- N05Why not just use Zapier?
- Use it, when the job is a two-step hop with no condition in it. It stops being enough when a rule has a condition, when the same event can fire twice, or when a mistake would change money or stock. Those are the routes worth owning as code.
- N06How do I know it worked?
- One operational number is written down before the sprint with its value that day, and measured again once the route is live. It is measured, not promised: what it did is shown to you, not forecast.
SHT-06Evidence
The systems this argument comes from.
Everything above was learned building these. Each sheet names the business, what the trade's arithmetic actually is, and the file in the repository that proves the capability. No client revenue and no client's customers appear on any of them.