One ranked list, four readers, four different meanings

What you are trying to do

Win fee-sensitive plans and benchmark the ones you hold.

A row is

plan in review

The action on it

Add to pipeline

The lists, in the order you meet them

1. 1.Fee Renegotiation1 in 3 at the top of the list · 1 in 9 at random · 2.88xPlans whose administrative cost per participant is about to fall, and stay down.
2. 2.Recordkeeper Change1 in 10 at the top of the list · 1 in 33 at random · 3.19xPlans most likely to change recordkeeper next year.
3. 3.Plan Termination Risk1 in 81 at the top of the list · 1 in 214 at random · 2.64xPlans most likely to terminate or merge away next year.

It misses the bar on being better than random, at 2.88x against a bar of 3.0. That is the wrong headline for this list, though: because 11.5% of plans do this anyway, the top of the list still comes in right about one time in three, far better in plain terms than either other list here. Sorting plans by what they pay, with no model at all, gives 1.22x against this model's 2.88x, so the mechanical bounce-back this was designed to survive is not what is driving it. The number to watch is how often this happens at all, which fell 22% between the years we learned from and the years we tested on. That is more movement than the other outcomes show, and worth re-checking on the next refresh.

What you are trying to do

Prospect competitors' books.

A row is

prospect

The action on it

Add to call list

The lists, in the order you meet them

1. 1.Recordkeeper Change1 in 10 at the top of the list · 1 in 33 at random · 3.19xPlans most likely to change recordkeeper next year.
2. 2.Fee Renegotiation1 in 3 at the top of the list · 1 in 9 at random · 2.88xPlans whose administrative cost per participant is about to fall, and stay down.
3. 3.Plan Termination Risk1 in 81 at the top of the list · 1 in 214 at random · 2.64xPlans most likely to terminate or merge away next year.

On the quiet plans, the ones where nothing obvious was going on, this scores somewhere between about 1.78x and 2.02x depending on how the test is run. The bar is 2.0, so it sits right on the line and we cannot honestly say it clears it. How it works is not in doubt: it leans hardest on how a plan's asset mix moved last year, which anyone reading the same public filings can see too. Treat this as a better ranked list with real economics behind it, not as something a competitor could never build.

What you are trying to do

Reach plans before their first audit engagement is decided.

A row is

audit prospect

The action on it

Add to pipeline

The lists, in the order you meet them

1. 1.First Plan Audit1 in 21 at the top of the list · 1 in 193 at random · 9.27xSmall retirement plans about to cross the line that requires an independent audit.
2. 2.Fee Renegotiation1 in 3 at the top of the list · 1 in 9 at random · 2.88xPlans whose administrative cost per participant is about to fall, and stay down.
3. 3.Recordkeeper Change1 in 10 at the top of the list · 1 in 33 at random · 3.19xPlans most likely to change recordkeeper next year.

Every bar is cleared, but say plainly why: the rule is written in participant counts, so sorting plans by the count the rule uses gets 9.75x at the top of the list against this model's 9.27x. The model does not beat that sort and we do not claim it does. What it adds is below the line. Among plans under 100, which a sort on the count would put at the bottom, it finds the ones that cross at 6.64x the rate of the rest, about level with the count itself there. The real work in this list is the plumbing: short-form and long-form small filers in one place, named, with the 2023 change to who counts applied year by year. That change is also why only 0.23% of small plans crossed in 2022, against about 0.65% a year before it.

What you are trying to do

Retain the clients you already serve.

A row is

client at risk

The action on it

Flag for outreach

The lists, in the order you meet them

1. 1.Plan Termination Risk1 in 81 at the top of the list · 1 in 214 at random · 2.64xPlans most likely to terminate or merge away next year.
2. 2.Recordkeeper Change1 in 10 at the top of the list · 1 in 33 at random · 3.19xPlans most likely to change recordkeeper next year.
3. 3.Fee Renegotiation1 in 3 at the top of the list · 1 in 9 at random · 2.88xPlans whose administrative cost per participant is about to fall, and stay down.

This list's headline result is not repeatable enough to call. Depending on how the test is run, the quiet-plans score lands between about 1.87x and 2.07x, against a bar of 2.0, so it sits right on the line and neither a pass nor a miss means anything. Only 0.48% of plans close in a year, so the top of the list holds a few hundred of them at most, and ordinary rounding in the arithmetic moves enough of those to flip the answer. Read it as roughly at the bar, not above it. Separately, plans that merge away are buried inside plans that close and we cannot yet tell them apart, so the one case that is an opportunity rather than a loss, a plan somebody acquired rather than lost, is not something we can point you at yet.

What you are trying to do

Buy providers, not plans.

A row is

target

The action on it

Add to pipeline

The lists, in the order you meet them

1. 1.Provider ConsolidationRecordkeepers, TPAs and advisory firms whose book is about to be absorbed.
2. 2.Plan Termination Risk1 in 81 at the top of the list · 1 in 214 at random · 2.64xPlans most likely to terminate or merge away next year.
3. 3.Recordkeeper Change1 in 10 at the top of the list · 1 in 33 at random · 3.19xPlans most likely to change recordkeeper next year.

It scored a perfect 1.0000 in testing, which is not a result. It is a sign something is wrong. What we asked it to predict was defined as one number crossing 0.40, and that same number was one of the things the model could see, so it simply learned where the line was and every other input scored exactly zero. Two real bugs were fixed on the way to finding it: a duplicated join that produced shares above 100%, and one provider consolidating its own registrations being read as eight separate acquisitions. The underlying series is plausible now, at 1.4 to 5.6% a year. The list still predicts nothing as it stands, and it stays held back until it is rebuilt to predict next year from what is known this year.