Small Business Tool Stack: What to Keep, Cut, and Connect

Every couple of years the same instinct arrives: there are too many tools, so the answer must be fewer tools. Someone starts hunting for one platform that does everything. Six months later the business is running the new platform and most of the old ones, because each old tool turned out to be doing something nobody had written down.

The number of subscriptions is rarely the problem. The problem is that the same fact lives in four places and nobody has decided which one is right.

That is what a tool stack audit is actually for. Not counting licences. Deciding where the truth lives.

Audit the facts, not the subscriptions

Start by listing the facts your business keeps, not the software you pay for. For most small businesses the list is short:

Who enquired, when, and what about
What a job is worth
What stage that job is at
What has been invoiced, and what has been paid
Who has been contacted, and about what
Who is doing what this week

Six to twelve lines, usually. Next to each one, write every place it currently lives. Not where it should live. Where a person would actually go to look it up today.

That list is the audit. Tools only appear as answers to the question.

Two things tend to surprise people. The first is how short the list of facts is. The second is how many places each one turns up: the CRM, a spreadsheet somebody maintains, a column in the accounting package, and an email thread that three people treat as authoritative.

Give every fact one owner

For each fact, name one system as the owner. Everything else holds a copy.

The useful rule for picking the owner: choose the system where the fact gets created as a by-product of work someone is already doing. Not the one with the nicest reporting screen.

Payment status belongs to the accounting package, because it lands there when the bank feed is reconciled, work that happens anyway. The "Paid?" checkbox in the CRM is a copy, and it is only as current as the last person who remembered to tick it. Job stage belongs to wherever the person doing the job updates it in the course of doing it. If your operations lead lives in a spreadsheet, the spreadsheet may genuinely be the owner. That is allowed. An honest spreadsheet beats a CRM field nobody fills in.

Once the owner is named, every copy has to get one of three fates: deleted, fed automatically from the owner, or clearly labelled as an estimate that can be out of date.

Most of the time and money in a stack audit gets recovered right here, before anybody touches an integration.

Cut: what nobody will miss on Monday

A tool is a candidate for cutting when any of these are true:

Nobody has signed in for a month. Most admin or billing screens show last login per user. Check it rather than guessing.
Everything in it is a copy. Every fact it holds is owned by another system. It is a display, and usually a stale one.
It was bought for a project that has finished. The project ended; the renewal did not.

One caution before you start cancelling. Seasonal tools look dead in a 30-day window. Something touched once a quarter for tax, year-end, or an insurance renewal will fail all three tests in March and be badly missed in June. Walk the year, not the month.

The licence savings are real, but they are not the main prize. The prize is that there are now fewer places for a fact to quietly go stale.

Keep: the systems doing real work

Two kinds of tool are worth keeping even when nobody in the business particularly likes them.

The first is any system of record. Moving one of these costs far more than the subscription, and the data export is the cheap part. The expensive part is every habit, template, and half-remembered workaround built on top of it. Replace a system of record because it is genuinely blocking the business, not because the interface looks dated.

The second is any tool that is load-bearing in someone's day. A plain tool that a person uses fluently is worth more than a better one they would have to learn during your busiest month.

Neither of these is an argument for keeping everything. It is an argument for being honest about what a replacement actually costs.

Connect: follow the re-typing

Now walk a normal week and note every point where a person moves a fact from one screen to another by hand. Each of those is a candidate for a connection.

Rank them by three things together: how often it happens, how long it takes, and what it costs when it goes wrong. A five-minute re-key that happens twice a day will usually beat a two-hour export that happens once a month. And the cost of being wrong varies enormously. A mistyped address on an invoice is an apology, a deal sitting in the wrong stage is a customer who never hears back.

One ordering rule matters more than the rest: connect after ownership is settled, never before.

If you connect two systems that disagree about a fact, you have not fixed the disagreement. You have automated it, and made it harder to notice, because the numbers now arrive looking official and nobody checks them by hand any more. That is the same failure that makes a monthly report impossible to trust, worth settling before you automate the report. When you do start connecting things, the patterns that survive their first year are worth reading first.

The output should fit on one page

A finished audit is one row per fact, with five things on each row:

The fact itself, in plain words
The system that owns it
Everywhere else it currently lives
How each of those copies gets updated today, by a person, or automatically
The decision: delete the copy, connect it, or label it as an estimate

If yours runs to four pages, you have listed tools again. Go back to facts.

Re-run it once a year, and any time you add a system. New tools arrive claiming to own facts that already have an owner, and that is exactly how a clean stack turns back into four versions of the truth.

Where to start

If you suspect your stack is holding the same facts in too many places but you are not sure which ones matter, our Automation & Growth Scan is a focused 30 to 45 minute session. We map where the facts live, identify what is worth connecting first, and leave you with a 30 to 60 day plan you can run with, even without us. If you already know which process is costing you the most, the Automation Sprint picks up from there.

Toop Technologies is an Atlanta-based technology consultancy. It runs US launch IT for international brands and builds automation and AI decision systems for small and mid-sized businesses. Learn more at tooptechs.com.