Contract Attach Rate Is Measured Against Every Customer You Ever Had

Why this matters

A shop can sell more agreements every year than the year before, keep a healthy share of them at renewal, and watch this number fall every single year. Owners who do not know why reach for the wrong fix: they lean on the technicians, change the pitch, or conclude the programme has stalled. Nothing has stalled. The denominator is growing faster than the numerator can, by construction, and no amount of selling changes that. Knowing the shape of the decay is what lets you tell a real problem from arithmetic.

What sits in each half

The numerator is customers holding an agreement that is active right now. The denominator is every customer on the books.

Neither half is period-scoped, and that is the entire story. The numerator is a snapshot of today. The denominator is a cumulative list that started the day the business opened and only ever grows, because almost nobody deletes a customer record. The person who called once four years ago about a problem you fixed in twenty minutes, who has moved out of the area since, is in that denominator forever and can never be in the numerator.

So the two halves do not merely have different time anchors. One of them has no time anchor at all.

The decay is arithmetic, not performance

Take a shop adding about 300 new customers a year, selling an agreement to one new customer in five, and keeping 78 percent of its active agreements at renewal. Nothing about their selling changes across the five years below.

Agreements grew from 180 to 238 over the four years, up about 32 percent. The list grew from 900 to 2,100 over the same four years, up about 133 percent. The rate fell by almost half because those two growth rates are not the same growth rate, and the reported figure is a ratio between them.

The agreement count converges, too, and that is the second half of the mechanism. Selling 60 a year and keeping 78 percent means the count settles just under 273 no matter how long you run it, because 60 divided by 0.22 is about 273. The list has no such ceiling. Run this shop another decade at the same performance and the attach rate approaches zero.

What holding the reported rate flat would have taken

Work year 1 backwards and the size of the problem becomes obvious. Holding 20 percent against a list of 1,200 needs 240 active agreements. Renewals carried 140 of the prior year's 180 forward, so they needed 100 new sales that year, against 300 new customers. That is an agreement sold to one new customer in three, where they were selling one in five.

One in three, on every new customer, every year, forever, purely to keep a reported number from moving. That is not a goal, it is evidence the number is not a goal.

Where the lifetime denominator is still the right one

It answers exactly one question honestly: of everyone we have ever touched, how many are under agreement today. That is a ceiling-of-opportunity count, and it is the correct base for sizing a win-back campaign, because the people outside the numerator are precisely the people you could call.

What it cannot do is trend. A falling lifetime attach rate is consistent with a programme growing strongly, and a rising one on a shop of stable size may only mean the shop has stopped acquiring customers. Use the level for sizing an opportunity and never read the direction as a verdict.

What corrupts the denominator before you even start

Records created for people who never bought. If a customer record is created at first contact rather than at first completed job, every quote request that went nowhere is in the denominator. In the example shop, 260 of the 2,100 records had no completed job ever. Excluding them gives 238 over 1,840, or 12.9 percent, which is 1.6 points above the 11.3 percent reported. That is worth knowing but it does not change the diagnosis, and it is a one-time correction rather than something that gets better.

One property split across two records. A rename, a spelling, a spouse's name on the second call. The agreement sits on one record and the other is a bare customer in the denominator.

A commercial account entered once per site. An account with nine buildings under one agreement shows as one agreement holder among nine customer records. In a shop with six such accounts averaging five sites each, correcting to one record per account takes the denominator from 2,100 to 2,076 and moves the rate by about 0.1 points, which is nothing. In a shop whose work is mostly multi-site commercial, the same correction can move it by several points, so the size of this one depends entirely on your mix.

The two denominators that answer a question

Customers served in the last 12 or 24 months. This asks whether the customers you actually work for are on agreements, and it is the number that should drive a technician conversation. Note the construction: an agreement holder receives plan visits, and a plan visit is a job, so every active agreement holder is already inside this denominator. The numerator is a strict subset of the base, which is what makes the ratio behave.

New customers offered an agreement in the window. This asks whether the offer is being made and whether it lands, and it splits cleanly into two figures that need separating: the share of new customers who were actually asked, and the share of those asked who said yes. A shop that never asks and a shop that asks and gets turned down have the same combined number and completely different problems.

Read any share over a customer list only when the denominator holds at least 50 customers, because below that a single sale or lapse moves the figure by 2 points or more and you will spend a month explaining noise.

There is no industry attach rate worth quoting, and there is a target worth setting

Any published attach-rate figure is uninterpretable, for the reason this whole card is about: two shops quoting one almost never share a denominator. A shop five years old with a lifetime base reads a different number from a shop thirty years old with the identical programme, and either of them roughly doubles on a customers-served-in-24-months base and rises further on a 12-month one. Comparing your figure to somebody else's is comparing record-keeping conventions.

What is comparable across shops is the offer rate, because its denominator is tightly defined: new customers in the window. So set the target there. A reasonable starting point is asking 80 percent of new customers, tuned down for the job types where a plan is genuinely irrelevant, and a shop should be able to name those job types rather than leaving the exception open.

Run the example shop against it. In year 4 they asked 140 of their 300 new customers, so 46.7 percent were asked, and they closed 60 of the 140 asked, so 42.9 percent of those asked said yes. Against the 80 percent starting point the ask rate fails badly. The close rate has no comparable target, because a close rate's denominator is whoever happened to be asked, so read it as this shop's own baseline and coach the ask, which is the half with a number to fail. Lift the ask to 240 of 300 new customers at the same close rate and the year produces about 103 new agreements rather than 60.

Then follow that through, because it is the mechanism this card exists to teach. Selling 103 a year and keeping 78 percent settles the agreement count near 468 rather than near 273. The reported lifetime attach rate still falls every year, because the list still grows by 300 a year and nothing about the offer rate changes that. What improves is the agreement count itself and the served-base figure, whose denominator grows with your actual workload rather than with your history. Note also where 103 lands: just above the 100 sales a year that would have been needed back in year 1 to hold the reported rate flat on a list a little over half the size. The target is reachable and the ratio is still not the thing to chase.

The same shop on three bases

Year 4 of the example, all three computed on the same day, all three correct.

Three warnings about reading that table across its rows. The first two figures share a numerator, so the gap between 11.3 percent and 22.7 percent is entirely a denominator fact and says nothing whatever about how well anyone sold. The third row carries a different numerator as well as a different denominator, counting only agreements sold in year 4 rather than all 238 active, so it is not comparable with either of the others and should never be set beside them as though all three were the same rate. And the 22.7 percent needs no zero-job correction, because a denominator built from customers served is built from completed jobs already, whereas the 11.3 percent does.

Pick one base, write down which it is, and keep reporting that one. The failure that costs a shop a quarter is two people in the same meeting quoting an attach rate on two different denominators and arguing about the business.

References

• See related: Contract Renewal Rate and the Link That Has To Exist
• See related: Recurring Revenue Share and the Band Worth Holding
• See related: The Membership Database: Keeping It Accurate, for the record hygiene that keeps the denominator honest
• See related: Customer Membership Programs, for structuring the offer itself