← All posts Industry observations

The integration you need is probably already paid for, check what your existing tools can do before buying a third

A gap appears between two systems and the instinct is to buy something to sit in the middle. More often the capability is already on the invoice, switched off and unopened.

Published 2026-08-135 min read
Tools we live in
n8n
Make
Zapier
Supabase
PostgreSQL
Cloudflare
HubSpot
Stripe
Shopify
Airtable
Slack
Notion

A gap opens up between two systems. Orders live in one place, invoices in another, and somebody spends part of every week moving information from the first to the second. The obvious next step is to find a product that sits in the middle and closes the gap.

Sometimes that is the right answer. More often the capability is already paid for, switched off, or exposed through an interface that nobody in the building has ever opened.

The person is the integration

It is worth naming the current arrangement accurately. When two systems do not talk and a person moves data between them, that person is the integration layer. They are an unusually expensive one, they only run during office hours, they make occasional errors that nobody catches for weeks, and they are entitled to annual leave.

Describing it that way changes the maths. The question is no longer whether to spend money on a connection. Money is already being spent on the connection. The question is whether the current form of it is the one you would choose.

What is usually already there

Most business software sold in the last decade can send and receive data. It is rarely advertised on the pricing page, because it is not what wins a demo, but it is almost always present.

  • Features included in your tier. Plenty of platforms ship with an export schedule, a webhook, or a native connection to the other product you already run, sitting unenabled because nobody looked.
  • The interface behind the interface. The screen a user sees is one way into a system. Nearly all of them have another, designed for machines, and it usually does more than the screen does.
  • A connection that already runs in one direction. Often one system is already pushing something somewhere. Extending an arrangement that works costs less than inventing a new one.

Why the third tool is tempting

Buying is easier than connecting. It has a price, a vendor, a contract, and a demo where everything works. Connecting has none of that. It needs somebody to look at two systems in detail and make a judgement, which feels like a project even when it is a week of work.

The trouble with the third tool is that it is a third thing. It has its own login, its own renewal, its own release notes, and its own edges that do not quite line up with either of the systems it was bought to join. In eighteen months it is part of the estate rather than a fix to it, and the original two systems still do not talk to each other directly.

What to check, in order

None of this needs a supplier to start.

  1. Write down the fact being moved. Not the process, the fact. An order number and its status. A contact and their address. Being precise about what actually travels makes everything after it simpler.
  2. Search both platforms for that fact. Look up the field name in each set of documentation. If both systems can read and write it programmatically, the gap is a connection problem rather than a product problem.
  3. Check the plan you are on. Some capability is gated behind a tier. Occasionally the upgrade is the cheapest option on the table, and occasionally you are already on the tier that includes it.
  4. Count the current cost honestly. How many minutes, how many times a week, and what happens downstream when it is late or wrong. Without that number there is nothing to judge any option against.

When buying is right

This is not an argument that new software is always the wrong answer. If a genuine capability is missing, if volume has outgrown what the current tools were built for, or if the existing platform is being retired, then replacing it is sensible and connecting it is throwing good money after bad.

The distinction is worth making out loud before anyone books a demo. Missing capability is a buying problem. Present but disconnected capability is a plumbing problem. They look identical from a distance and they cost very different amounts to solve.

The duller question

Before pricing up a platform to close a gap, it is worth asking what the software already on the invoice can actually do, and what it would cost to connect it properly.

The answer is often that the licence renewed in March already includes most of it.

Before you buy anything

Check the invoice before the catalogue.

Two systems
One fact held in both, nothing joining them
The person
Currently doing the job an interface could
Twice a week
Where the manual join usually runs
Third tool
Adds edges of its own to maintain
Keep reading

You might also like.

Short pieces on the patterns we see across the work, and the principles behind how we price it.

How we think

Frequency, not difficulty, decides what to automate first.

Asked what to automate first, most teams nominate the task they hate most. It is almost always the one that happens least often, and fixing it saves an afternoon a year.

Read the post →
Behind the scenes

The first hour of a Sprint is spent watching, not asking.

Every business has a documented process and a real one. The gap between them is not sloppiness. It is where the workarounds live, and the workarounds are the actual requirements.

Read the post →
Patterns we see

Most jobs are not lost on price.

When a quote does not turn into a job, the story is usually that someone came in cheaper. Walk the timeline back and the job was often gone before price was ever the point.

Read the post →
Industry observations

Nobody reads the dashboard.

Dashboards get built, admired for a fortnight, then quietly abandoned. That is a design problem, not a discipline problem. A number with no decision attached is decoration.

Read the post →
Patterns we see

Every business has one person who cannot go on holiday.

Ask what would break if one particular person were away for two weeks, and watch the pause. There is almost always a name, and that name is a risk nobody has priced.

Read the post →
How we think

The cheapest leads you have are the ones you are already ignoring.

Most businesses have demand they are not collecting before they have demand they need to buy. Search, maps and referrals come first. Paid ads are the last lever, not the first.

Read the post →
How we think

The ads are not the problem, the machine behind them is.

Turning on ads before the funnel behind them converts is the most common way businesses waste money. The funnel is the machine. Ads are just the moment you switch it on.

Read the post →
Patterns we see

The spreadsheet that became a system when no one was looking.

Almost every team has one workbook everyone is quietly afraid to touch. Here is how it got that way, and what to do about it.

Read the post →
Industry observations

Most automation projects fail before anyone writes code.

When an automation project goes wrong, the tool usually gets the blame. The real cause is almost always something that happened earlier.

Read the post →
Worked examples

What we built when sales spent nine hours a week on lead routing.

A 30-person SaaS sales team was losing nine hours a week to lead routing. Here is what the fix looked like in practice, and what we deliberately did not build.

Read the post →
How we think

The list of automations is a roadmap, not a shopping basket.

Most automation pitches arrive as a menu of pick-three options. We deliver them as a sequenced plan, ordered by where the hours actually come back.

Read the post →
Patterns we see

Three signs the boring bit of your day is fixable.

A short diagnostic for people who suspect the boring layer of their day is fixable but cannot quite name what it is.

Read the post →
How we think

Why we count the hours before we touch a tool.

The most useful number in an automation project is not a price. It is the hours a task quietly eats every week.

Read the post →
Finance and accounting

Cutting the monthly close from eight days to two.

A 28-person fintech was losing the first eight working days of every month to close. Three small connections later, the close runs in two days and around 120 hours a month come back across the team.

Read the case study →
Customer service and support

Triaging 1,200 weekly tickets before an agent opens one.

A 60-person SaaS team was getting 1,200 tickets a week, most of them in five buckets. A triage layer classifies, routes, and attaches the right macro. Real bugs surface to a human inside an hour, not five.

Read the case study →
Operations and logistics

Stopping orders falling between Shopify and the warehouse.

A 45-person homeware brand was pushing orders to its 3PL via twice-daily CSV exports. A webhook-driven feed replaced the CSVs and the oversells stopped.

Read the case study →
Professional services

Client onboarding in 20 minutes, not two hours of forms.

An 18-person law firm was spending two hours of partner or paralegal time on every new matter. One form, one routing layer, and the partner signs a pre-populated letter while the rest is already done.

Read the case study →
E-commerce and DTC

Syncing 4,000 SKUs across Shopify and Amazon every hour, not every Tuesday.

A 12-person DTC pet brand ran its inventory sync as a weekly manual job. The hourly automatic sync took the manual loop away. Tuesday came back.

Read the case study →
Free tool

Not sure where to start? Take the 5-minute audit.

11 questions, your automation score, an estimated time-loss number, and three personalised recommendations. No email needed to see your results.

Take the free audit → 5 minutes · 11 questions · free
The Sprint

A focused workshop. A written report you keep.

A focused day mapping your operations and scoring the automation opportunities. The ranked plan you can act on, with us or without, follows. From £1,500.

See how a Sprint works → Focused workshop · written report · yours to keep
Talk to us

Want to chat?

Two ways in. Send us a message about what's broken, or book a 30-minute call if you already know what you'd like built. Both reach the same person.

Hours
Mon to Fri, 09:00 to 18:00 UK
Reply
Within one working day

Drop us a line.

We'll get back within one working day, usually faster.

By submitting, you agree to us getting back about this enquiry. No newsletter.