Skip to content
fiveleaf
Telecoms·

4 min read

·For operators

Ofcom automatic compensation: what each failure costs, and where AI actually helps

A missed appointment is £32.31. Delayed provisioning is £6.46 a day. Late repair is £10.34 a day. Here is what those numbers add up to across a growing subscriber base, and the honest limit of what automation can do about them.

Silviu Major·Founder, Fiveleaf··Updated

Every operational failure in a broadband business now has a price printed next to it.

From 1 April 2026, Ofcom's automatic compensation rates are £32.31 for a missed engineer appointment, £6.46 for each calendar day a new service starts late, and £10.34 for each calendar day service stays broken after two full working days. It lands as a bill credit within 30 days, and the rates climb every April with CPI.

Those are small numbers individually. That is exactly why they get ignored.

Do the arithmetic on a growing base

Take an altnet adding a few hundred subscribers a month. Suppose a modest share of installs slip, and a modest share of appointments get missed.

A hundred delayed installs running five days late each is £3,230. Fifty missed appointments is £1,615. A handful of multi-day faults adds another few hundred. Call it five thousand pounds in a month, before anyone has spoken to a customer about it.

That is not the part that should worry you.

The real cost sits underneath the payout

The compensation is the visible, quantified, easily-budgeted piece. It is also the smallest piece.

A delayed install means a customer who committed money and has no service. They contact you, probably more than once. Each of those contacts costs support time. If they cannot get a straight answer about when the engineer is coming, the frustration compounds, and the next thing that happens is a complaint, or a switch, or both.

The acquisition cost on that subscriber has not been recovered. Passing the premises has not been recovered. The £6.46 a day is almost incidental next to losing them in month four.

So the honest framing is not "reduce compensation payouts". It is "stop one operational failure becoming three commercial ones".

Where automation genuinely helps

Three places, and I want to be precise because this is where the category oversells itself.

Absorbing the chase. The heaviest volume during a delayed install is the customer asking where things are. Same question, repeatedly, from an anxious person. An agent connected to provisioning answers it instantly, at any hour, with the real date rather than a holding line. In our own deployment this was the surprise: we expected billing to be the big win, and the install-status chase turned out to be what was quietly eating the team alive.

Catching the at-risk order earlier. If an order has slipped its date, that is knowable before the customer notices. A triggered message that says the date has moved, here is the new one, here is what happens next, changes the emotional shape of the whole thing. The failure still happened. The customer was not left to discover it.

Making the compensation visible. Customers who find out they were owed money and never told it react badly. An agent that can explain what has been credited and why turns a grievance into evidence you are dealing with them straight.

Where it does not help, and you should hear this plainly

It does not stop your engineer being late.

If your field operations are under-resourced, if your subcontractors miss appointments, if provisioning is genuinely slow, no amount of conversational AI fixes that. It handles the consequences faster and more consistently. It surfaces the pattern sooner. It does not touch the cause.

I would be suspicious of anyone selling you automation as a compensation-reduction strategy. The mechanism that reduces payouts is fixing the operation. Everything else is managing the fallout better, which is worth real money but is a different claim.

What I would actually measure

If you want to know whether this is working, do not look at total compensation paid. Too many variables move it.

Look at contacts per delayed order. If a slipped install used to generate four inbound contacts and now generates one, that is the automation working, and it is measurable within weeks.

Look at complaint rate on delayed orders specifically, separated from your general complaint rate.

And look at churn in the first ninety days for customers whose install slipped, against those whose did not. That gap is the number that actually matters, and most operators have never calculated it.

The uncomfortable version

Automatic compensation is doing something useful to this industry. It puts a price on failures that used to be absorbed by the customer's patience, and it makes them appear on a finance report where somebody has to explain them.

If those numbers are climbing, the answer is not better messaging. It is that your operation is failing more often, and now you can see it.

What automation buys you is the ability to handle the failures you do have without them compounding, and better information about where they cluster. That is genuinely valuable. It is just not the same as fixing it.

Rates cited are Ofcom's automatic compensation figures effective 1 April 2026.

Frequently asked

What are the current Ofcom automatic compensation rates?
From 1 April 2026: £32.31 for a missed engineer appointment, £6.46 for each calendar day a new service starts late, and £10.34 for each calendar day service is not repaired after two full working days. Payments land as a bill credit within 30 days, and the rates rise each April in line with CPI.
Does AI reduce compensation payouts?
Not directly, and any vendor claiming otherwise is overselling. Compensation is triggered by operational failure, not by communication. What automation changes is how early you see an at-risk order, and whether one failure turns into a complaint and a switch on top of the payout.
Is automatic compensation mandatory for every provider?
It operates as an industry code that providers sign up to rather than a blanket legal obligation on everyone, so coverage varies. The commercial question for a smaller provider is less about whether you are obliged and more about whether the customer comparing you to a signatory expects the same treatment.

If you want help building this

Building AI agents into a mid-market business is what Fiveleaf does.

Bespoke build, fully integrated, continuously optimised. A 30-minute discovery call is enough to tell you honestly whether AI agents fit your team right now, or whether you’re better off waiting six months. No pitch.

About the author

Silviu Major, Founder, Fiveleaf

Silviu Major

Founder, Fiveleaf

10+ years building automation systems inside enterprise SaaS, now applying that same operational rigour to AI implementation for mid-market businesses. Writes about what works (and what doesn’t) from inside live deployments, not from the outside looking in.

Connect on LinkedIn →

Keep reading