The software that covers 10% of your volumes

A director told me recently about his search for a travel management tool. He'd seen several, all very good. All incapable of handling the way his business actually buys its tickets. His conclusion: "to manage ten percent of our volumes, we don't need a dedicated tool."

It's one of the most clear-sighted sentences I've heard on the subject, and it applies well beyond travel.

Coverage rate is the only question that matters#

The demo always shows the easy case#

Demonstrations all look alike, because they all show the standard case. The standard case is precisely the one you have no trouble handling today.

The useful question isn't "can this tool do X?" but:

What share of my operations, by volume, will this tool handle end to end with no manual intervention?

End to end matters. A tool that handles the request but not the internal recharging, or the order but not the special case that makes up half your lines, doesn't cover those operations: it covers their beginning.

How do you measure what software really covers?#

To measure it before signing, there's a simple and merciless exercise: take your last thirty real operations, in order, without picking any, and ask the vendor to show you how each one goes through their tool. Not three typical cases — thirty consecutive ones. The figure that comes out of that exercise is your coverage rate.

What the uncovered share costs#

This is the part of the calculation that price comparisons ignore, and it's almost always the biggest.

When a tool covers sixty percent of your cases, the remaining forty percent don't disappear. They carry on being handled as before — a spreadsheet, a homemade database, a folder — but with two aggravating factors:

Now you have to maintain both#

What's in the tool and what isn't, and reconcile the two at month end. Monthly reconciliation is work that did not exist before the tool.

And somebody becomes irreplaceable#

The person who maintains the manual part is the only one who knows the rules that were never written down. That's a business risk, not an organizational detail: the day they are away, the process stops.

So the real cost of software is:

subscription + integration + (human time on the uncovered share × loaded hourly cost × 12)

With an uncovered share that occupies someone half time, you exceed the subscription very quickly, and often by an order of magnitude.

Sandrine takes care of that. She's got a little file of her own.

Three cases where off-the-shelf wins every time#

I don't sell custom software to people who don't need it, for a selfish reason: a badly framed project ends badly, and word gets around.

Off-the-shelf wins when:

When the law has standardized the process#

• Payroll, the general ledger, VAT returns. Payroll, general ledger, VAT returns. Nobody gains a competitive advantage from running payroll differently, and the vendor tracks regulatory changes on your behalf. That's exactly what you want to buy.
• The field moves faster than you could follow. Approved e-invoicing platforms, for example: the approval and its upkeep are the product.
• Your volume is low. Fifty operations a month justify no development, however tedious they are.

In those three cases, custom development is a vanity expense.

Where it loses, and why#

When the particularity is the business#

It loses when the particularity it can't handle is your business.

The company I mentioned buys its flights as batches of prepaid coupons, because it's twenty to thirty percent cheaper and cancellable until take-off. No tool on the market knows how to count in coupons, allocate them, then recharge them to the right company in the group pro rata to the person's working time.

That isn't a quirk: it's a real financial optimization, built up over years, that makes money. Asking him to give up his coupons in order to fit into a piece of software is asking him to pay more for his flights for the pleasure of owning a tool.

When software asks you to change what makes you money, it isn't the right software.

The threshold, in one line#

Custom becomes profitable when the annual cost of the uncovered share exceeds the amortisation of the development over three years.

The figures assume a loaded hourly cost of forty euros; replace them with yours, the method is unchanged.

At what point does custom become profitable?#

In practice, with today's tooling — and this is what has changed in the last two years — a focused business application takes six to ten weeks to build. So the threshold has dropped a long way: five years ago you needed a full-time person tied up in the manual part for the calculation to tip. Today, half a person is plenty.

Which doesn't mean you should rebuild everything. The right answer is often: off-the-shelf software for the standard part, a custom application for your particularity, and a link between the two.

What we do with that ten percent#

When an off-the-shelf tool covers the bulk, you do not throw it away. You build what handles the share it cannot take, and you connect the two.

That is the most common case among our clients: a standard tool holding ninety percent of the volume, and a custom app for the remaining ten — the ones that cost the most because they are handled by hand. See how we build that share.

Coverage rate: what to remember#

Before signing, run thirty consecutive real operations through the tool you're shown. The rate that comes out beats any comparison table.

Cost the uncovered share in hours and euros, not in annoyance. That's what decides, not the subscription price.

And be wary of software that asks you to give up a profitable practice to fit its model. What it saves you in license fees, you pay back elsewhere.