Three of Last Week's Eleven Account Packets Came Back NIGO. That's the First AI Number an RIA Should Baseline.

The rejection queue at 8:15 on Monday

The client service associate gets in at 8:05, pours the coffee, and opens Schwab Advisor Center before she opens her email. The first thing she checks is the alerts tab — the list of new account applications and transfer requests the custodian would not accept as submitted. Last Monday there were three sitting there. The firm had sent eleven packets the week before.

One was a Roth conversion form where the client initialed the tax-withholding election but never dated it. One was an ACAT transfer where the registration read "Robert J. Whitmore Trust" on the application and "Robert James Whitmore Revocable Living Trust u/a/d 6/12/2009" on the delivering firm's statement. The third was a trust certification missing the successor trustee page. None of these are difficult problems. All three cost the same thing: another email to the client, another signature request in DocuSign, and another week before the money is actually in the account.

Every registered investment adviser in the country has this queue. Almost none of them know their own number.

What "not in good order" costs a 400-household firm

A NIGO packet is a new account or transfer application the custodian rejects on first submission — a missing date, a mismatched account title, an unsigned page — and each rejection restarts the clock on the client's money actually moving. That last part is the part owners underprice. The rework is annoying; the delay is expensive.

The rework itself is real: call the client, explain that the form they already signed needs signing again, re-key the packet, resubmit, watch the alerts tab. At most firms that is twenty-five to forty minutes of a client service associate's day per packet, plus a piece of the operations manager's attention, plus — on the third or fourth rejection — a lead advisor picking up the phone to smooth over a client who is starting to wonder whether the firm has its act together.

The delay is worse. If your firm bills quarterly in advance off the last business day of the quarter, an $850,000 rollover that lands on April 2 instead of March 28 does not get billed until July. Nothing was lost permanently, but a full quarter of revenue on that account slid into the next period because a page was unsigned. Multiply that by the number of March transfers your firm handles and you have the actual cost of the queue.

What changed in 2026 is the size of the question, not the cleverness of the tool

Deloitte's State of AI in the Enterprise 2026 found that 84% of organisations investing in AI report positive returns. That headline gets quoted badly. The interesting part is the pattern underneath it: the firms that got a return did not start with a strategy. They took one messy process, put a machine in front of it, kept a human reviewing every output, and measured whether errors fell or hours dropped before they widened the scope by an inch.

For an RIA, new account and transfer paperwork is close to a perfect first process. It is high volume, it is repetitive, it is governed by rules the custodian publishes, it produces a hard pass/fail result from an outside party who does not care about your feelings, and the baseline is sitting in the alerts tab already. You are not asking a model to have judgment. You are asking it to read a packet against the same checklist your best operations person carries in her head, and say what is wrong before the packet leaves the building.

Here is the workflow that firms are actually running in mid-2026.

flowchart TD
  A["Advisor collects signed packet in DocuSign"] --> B["Checker reads packet against custodian rules"]
  B --> C{"Anything missing or mismatched?"}
  C -->|Yes| D["Flag: unsigned page, title mismatch, missing trust page"]
  D --> E["Client service associate fixes before submitting"]
  E --> B
  C -->|No| F["Submit to Schwab Advisor Center"]
  F --> G["Custodian accepts, assets in transit"]
  G --> H["Ops logs outcome in the NIGO tracker"]

The four numbers to write down before you switch anything on

If you do nothing else this month, do this. It takes an operations manager about two hours with a spreadsheet and the custodian's alerts history.

• Packets submitted per month, split by type: new individual, new trust, IRA rollover, ACAT full transfer, non-ACAT, LOA or standing letter of authorization.
• First-pass rejection rate — what percentage came back NIGO, and the top five reasons. Almost every firm finds the same short list: unsigned or undated page, registration title mismatch, missing trust certification page, wrong or missing beneficiary form, and a delivering-account number keyed off a stale statement.
• Days from client signature to assets in the account, median not average. The average hides the ugly ones.
• Minutes of staff time per rejected packet. Ask your CSA to log it honestly for two weeks. She will guess low.

Those four numbers are the argument. Without them you are buying software on a feeling, and in ninety days you will not be able to tell your partners whether it worked.

Worked example: eleven packets a week at a 27% rejection rate

Stated assumptions, all illustrative — substitute your own from the baseline above. A firm with roughly 400 households submits 11 packets a week, 48 working weeks a year. 27% come back NIGO on first submission. Each rejection burns 35 minutes of client service associate time at a fully loaded $38 an hour, and adds 6 days to funding. Average transferred balance is $310,000 at a 0.90% annual fee. A checking step cuts the rejection rate to 9%.

That is roughly $2,090 of recovered staff cost and about $8,370 of billing that stops sliding into the following quarter — against a tool that, at current pricing, costs less per packet than the postage your firm still spends on wet-signature trust documents. The point is not the exact figures, which are yours to replace. The point is that both lines are countable and both come out of systems you already own.

Where the human stays, and it is not a small place

The checker does not decide whether the account is appropriate. Suitability, the rollover conversation, and whether a client should move an old 401(k) at all remain the advisor's job and the advisor's documented reasoning — the Department of Labor's expectations around rollover recommendations have not softened, and a machine's opinion is not a defense.

The checker also does not sign anything, does not initiate money movement, and does not touch the standing letter of authorization conditions your firm relies on to stay out of the custody rule. Keep a named person approving every submission. And keep the CCO in the loop: whatever the tool reads, it is reading client personal information, so it goes on the vendor list, into the Rule 206(4)-7 policies, and into the safeguards documentation Reg S-P now expects smaller advisers to maintain.

Finally, it will miss things. Trust language is genuinely ambiguous, custodians change intake requirements without announcing them loudly, and a checker trained on last quarter's rules will confidently pass a packet that fails today. That is why the number you track is the rejection rate, not the tool's self-reported accuracy.

Frequently asked questions

Doesn't my custodian already validate this?

They validate at intake, which is exactly the problem — the validation happens after the client has signed and after the clock has started. The whole gain here is moving that same check to before submission, inside your office, where fixing it costs one email instead of one cycle.

We only open about eight accounts a month. Is it worth it?

Probably not as a purchase decision on its own, but it is absolutely worth the two hours of baselining. Small firms often discover their real bottleneck is somewhere else entirely — quarterly billing reconciliation, or unreconciled positions in Orion or Black Diamond before the performance reports go out. Measure first, then aim.

What do I tell my compliance consultant?

That you are adding a pre-submission review step, that a named employee reviews and approves every output, that no client data leaves your approved vendor set, and that you are logging outcomes. Compliance people object to unmonitored autonomy, not to checklists.

How long before I know if it worked?

One quarter. You want at least 100 packets through the new process to see a rate move, and you want it to include a March or an April, when rollover volume spikes and everyone is distracted by the Form ADV annual updating amendment.

One side effect worth planning for: when transfers stall, clients call, and they call the main line asking whether their money has landed. CallSphere builds AI voice and chat agents that answer the firm's phone and website chat around the clock, take the question, capture who is asking and what account they mean, and book time with the right advisor — so the client service associate is not being interrupted mid-packet by the same status question fourteen times a week. It does not touch your custodial paperwork. It just keeps the front of the house quiet while operations fixes the back.