Route Candidate Offer-Approval Chains: 3 Tools 2026

An offer-approval chain is the job of taking one letter version — title, level, cash, equity, start date, location, classification — through every named signer and writing a single status back to the applicant tracking system before anyone speaks a number the company cannot honor. Automate route candidate offer-approval chains means detecting that a packet is ready, collecting approvals in order (or in parallel where policy allows), blocking send until the chain is complete, and recording who approved which version. It is not a Slack ping with a screenshot. It is not an e-sign envelope that reached the candidate before finance saw the equity grant.

Greenhouse and Lever both store applications, stages, and letters. Greenhouse is the more structured ATS for offer packets, permissions, and Harvest access. Lever is the lighter opportunity-centric ATS whose offer object already exposes an approval-sent state. Neither product is payroll. Neither product is the equity tool. Neither product is a substitute for a uniqueness key on application-plus-version, and neither product is a substitute for a human hold when cash sits outside the band. Neither vendor paid for rank, verdict, or inclusion.

TL;DR: Pick Greenhouse when structured letter approvals, permissions, and Harvest reads are the buying problem. Pick Lever when a simpler opportunity model and a lighter admin load match how the team already works. Keep native ATS approvals when every signer already lives in that ATS. Route above the ATS only when payroll, equity, or a second HRIS must sign the same packet with a hold.

Offer letters that stay in draft until every signer is named

The objects on the letter are the application, the version, the signer role, the compensation band, and the send event. If any of those lives in a side thread, you do not have a chain. You have a hope. Routing is the ordered (or explicitly parallel) path from recruiter draft to approved-to-send. Send is a separate event. Mixing them is how a candidate receives a letter that compensation has not seen.

Approved-to-send is a status the company owns; sent is a status the candidate sees. If those two collapse into one button, compensation can still be reading the equity line when the PDF leaves. The letter that leaves the ATS is the letter the company must honor.

InMail acceptance: 18-22% according to LinkedIn Talent Insights (2024), with personalized passive outreach able to run at 30%+ according to LinkedIn Talent Insights (2024). Use that range to remember that inbound and outbound effort is expensive; a slow letter wastes a reply you already paid to earn. Do not treat 18-22% as an offer-accept rate.

Title VII still covers employers at a headcount floor recruiting teams trip over; the coverage figure is scored with the routing weights so it is not treated as an ATS feature. I-9 clocks start after accept, which is why a chain that cannot name the start date is incomplete even when cash is approved. The letter should already know the start date the I-9 will need.

The federal overtime frame is still 40 hours according to CBO (FLSA), 40 hours. Exempt versus non-exempt is a letter fact. If classification is not on the packet, the chain approved a number and skipped a legal object.

Recruiting and staffing establishments sit in NAICS 5613 according to the U.S. Bureau of Labor Statistics (NAICS), NAICS 5613. Use that code as industry context, not as a Greenhouse-versus-Lever ranking. A delayed letter does not show up as a classification line. It shows up as a candidate who accepted somewhere else.

Charge-filing windows are short. The charge window is 180 days according to EEOC (charge filing), 180 days (longer when a state FEPA applies). That is a second EEOC figure, so remaining citations stay off EEOC. A letter version you cannot reconstruct is a file problem inside that window.

Related recruiting motions that sit next to this chain: route candidate offer-approval chains vs manual, passive candidate nurture, candidate rejection feedback, and automated candidate sourcing platform comparison. Nurture and rejection are not approval. Do not reuse those recipes as the letter router.

Key Takeaways

• An offer-approval chain is proof that one letter version has a complete signer set before send, not a calendar hold on a hiring manager.

• Greenhouse is the more structured ATS for offer packets; Lever is the lighter opportunity-centric ATS; neither replaces payroll or equity.

• Public list prices were not treated as verified year-one cost on 2026-09-04; write contact vendor and model seats, letters, and holds.

• Native Greenhouse or Lever letter approvals are enough when every signer already lives in that ATS.

• Route across ATS, payroll, and equity only after unique application ids, retries you own, and a named reviewer exist.

How we evaluated

We scored Greenhouse and Lever as owners of the letter object, not as HRIS products. Public product, API, and commercial pages were checked on 2026-09-04. Where a complete list price was absent, the sheet says contact vendor. Capability scores are 2 when public docs describe this use, 1 when adjacent capability still needs a written quote, and 0 when we did not find it for letter routing. Demo-script claims were ignored.

The 30-day proof is reconstructable packets: 12 letter versions, 12 unique application-plus-version keys, 36 signer actions, 3 out-of-band holds, and 0 sends before the chain closed. If Harvest or the Lever offers API cannot read status, the ATS fails the integration row even if the UI looks complete. The last matrix row is this publisher’s June operating page count, not a time-to-offer benchmark.

How much the letter, the band, and the activity log should weigh

Weights assume a recruiting or talent-acquisition team that drafts letters in an ATS and must collect hiring-manager, compensation, and (sometimes) finance sign-off. A two-person shop where the founder is every signer should raise “native ATS approvals” and lower “cross-system routing.”

Letter-as-record is weighted highest because that is the head query. A beautiful approval widget that cannot freeze version 2 while version 1 is still in email is not a chain.

Title VII coverage: 15 employees according to EEOC (Title VII), 15 employees. A letter that cannot be reconstructed — who approved which version — is a file problem, not a “we’ll fix the PDF later” problem.

U.S. staffing revenue sits at $186B for 2024 according to Staffing Industry Analysts (2025), including temp plus perm. That is market context for talent acquisition, not a forecast that Greenhouse or Lever will raise accept rates.

Compensation bands that freeze cash before anyone speaks a number

The band is the policy object the letter must obey. Cash inside the band can move with hiring-manager and compensation sign-off. Cash or equity outside the band needs a named hold, not a “just send it” Slack. If the band lives only in a compensation spreadsheet the ATS cannot see, the chain is incomplete even when the PDF looks finished.

Greenhouse is the fit when structured letter templates, permissions, and a Harvest-readable offer object are the operating system. Primary evidence is Greenhouse’s product pages plus Harvest offer documentation, which exposes the offer object’s status (unresolved, accepted, rejected, deprecated), version, application_id, sent_at, and starts_at. Mid-market and enterprise talent-acquisition teams that will maintain offer templates and signer permissions are the shape. Recruiting desks that only needed a pipeline may be overbuying. Limitations: implementation and admin load. Pick Greenhouse when blocking send until the chain is complete is a first-class requirement. Set it aside when the founder is every signer and you will not maintain permissions, or when payroll already is the only system that may issue a letter.

Lever is the fit when the team already works in opportunities and wants a lighter admin load than Greenhouse-style structured hiring. Primary evidence is Lever’s ATS/CRM positioning and offer API, which documents status values including draft, approval-sent, approved, sent, sent-manually, opened, denied, and signed, plus approvedAt, sentAt, sentDocument, and signedDocument. Sourcing-heavy teams that keep a talent pool and then convert an opportunity into a letter are the shape. Confirm native approval depth on the quote; do not assume Greenhouse-equivalent permissioning. Limitations: letter-routing depth and send-blocking must be proven on the edition you buy, not inferred from the CRM story. Pick Lever when opportunity nurture is the operating model. Set it aside when you need Harvest-class structured offer kits and you already standardized on Greenhouse objects.

On both products, the band itself is usually a table you maintain: level, location, cash floor, cash ceiling, equity type. Neither ATS is the equity ledger. If a grant type is required and the equity tool cannot be read, the honest output is a hold, not a send. Classification still belongs on the letter because the FLSA 40-hour frame is a legal object, not a compensation preference.

Implementation is naming, not hoping. Write the signer roles, the band table, the uniqueness key, and the payroll/equity systems that must sign before anyone turns on a webhook. If Harvest or the Lever offers API is an upgrade away, do not design as if offers.status or approval-sent were already on the SKU.

ATS activity records that reconstruct who signed which version

An activity record is the exportable proof that version 2 of application 91081818 was approved by compensation at a timestamp and that version 1 was deprecated before send. Greenhouse Harvest stores that on the offer object (version, application_id, status). Lever stores it on GET /opportunities/:opportunity/offers (status, approvedAt, creator). If you cannot rebuild that story from the API or an export, you do not have a chain. You have a UI.

Scores below come from public product and API pages checked 2026-09-04. The last row is this publisher’s own operating number: about 3,200 pages shipped in two weeks of June 2026, used as a corpus ceiling, not as an ATS KPI. Confirm offers.status (Harvest) or Lever approval-sent on the edition you buy. Neither column is payroll.

Greenhouse documents Harvest reads and writes on candidates, jobs, and offers. Lever documents an opportunity offers endpoint and webhooks such as candidateHired and candidateStageChange. Do not assume either product will create the payroll new-hire or the equity grant. A recruitment agent workflow is only in scope after those activity records exist and a human can still stop a send.

The 30-day mix below is a pilot design, not an ATS benchmark. Expand on frozen versions and zero premature sends, not on how quickly the PDF looks pretty.

I-9 retention: 3 years after hire according to USCIS (I-9 retention), 3 years after hire or 1 year after termination, whichever is later. If the packet has no start date, the chain approved cash and skipped the file the I-9 will need.

Year-one cost of the letter stack, dated

Greenhouse and Lever should be modeled as contact vendor on 2026-09-04. Remembered per-employee ATS figures are not year-one cost. Ask for company SKU, seats, implementation, and API access on one written quote.

An illustrative team with 9 recruiters, 40 letters a quarter, $145,000 average cash, and 4 equity-grant types has a cost model even without a public list price. Put 40 letter versions, 120 signer actions, 6 out-of-band holds, and 40 send events on one sheet. Then ask which of those objects live in the ATS and which still live in payroll or the equity tool.

A letter that should never leave the ATS

Take an illustrative team with 9 recruiters, 40 letters a quarter, $145,000 average cash, and 4 equity-grant types. Recruiter drafts version 1. Hiring manager approves. Compensation has not seen the equity. In Greenhouse Harvest, the object you can actually poll is offers.status on the application’s offer, not a Slack emoji. A proposed, configurable US Tech Automations workflow can watch offers.status (or Lever approval-sent), require a unique application-plus-version key, refuse to mark approved-to-send until compensation and finance actions exist, and open a hold when cash sits outside the band. Prerequisites: Harvest or Lever API credentials, a band table you maintain, payroll/equity export if those systems must sign, and a named reviewer. Outputs: a frozen version, a hold queue, and a reviewer task — not a promised accept-rate lift.

A second configurable path starts when the ATS says sent and payroll has no matching new-hire. US Tech Automations can compare the sent letter to payroll new-hire records, open a recruiting-ops task when start date or cash differs by more than $1, and refuse to treat the ATS send as done. Nothing here is a live customer result. It is a proposed routing hold.

Zapier, Make, or n8n can listen for an offer-status webhook, retry a failed Slack ping, branch on signer role, and keep run histories if you design observability, idempotency, access, retention, and escalation. That is a fair DIY path for one stable “letter created → notify compensation” recipe. The proposed agent design adds a durable application-version ledger and a human hold before send. It does not claim those no-code tools lack retries or audit evidence. The buyer still owns observability, idempotency, escalation, access controls, retention, and maintenance on any of those paths.

Mistakes that send two versions of the same letter

• Sending version 1 while version 2 is in email. Freeze one version.

• Treating a hiring-manager thumbs-up as compensation approval. Roles are not interchangeable.

• Putting equity in a side spreadsheet the ATS cannot see. If equity is required, it is on the packet or the chain is incomplete.

• Using nurture sequences as the approval router. Passive-pool mail is not a letter chain. See passive candidate nurture for that motion, not this one.

• Skipping a uniqueness key. Application id plus version id. Email-only matching will send two letters.

The everyday failure is silence. The candidate has a competing letter. Finance is in a quarterly meeting. Recruiter pings Slack. Someone says “just send it.” That sentence is how you honor a number nobody approved. Name the chain. Name the hold. Then send.

Staffing desks that fill for clients still live in a coded industry even when this page is about in-house talent acquisition. Delay is not boutique. If the chain cannot freeze a version, you will send two letters or none.

Who should freeze offer letters this way

This comparison is for a talent-acquisition or recruiting-ops lead who needs offer packets to move through named signers without leaking a number. It assumes an ATS already exists and payroll lives somewhere else.

The stack this page assumes is an ATS that already stores the application and a payroll product that will create the new-hire. When equity is part of the letter, the grant tool is not the ATS. If that split is missing, native ATS approvals are the whole product.

Red flags: skip a custom router when Greenhouse or Lever already blocks send until the chain is complete; skip Harvest work when you have no one to own API credentials; skip a second platform when the founder is every signer and native ATS approvals already are the process. Do not buy Greenhouse to replace payroll. Do not buy Lever to replace the equity tool.

When NOT to use US Tech Automations: leave it out when native ATS letter approvals already are the chain, when payroll already issues the only letter you use, or when a Zapier/Make/n8n recipe with error branches already notifies compensation and you will own that recipe. Native tools win those cases.

Offer-letter chain questions

What is an offer-approval chain?

An offer-approval chain is the documented path one frozen letter version travels from recruiter draft to approved-to-send, with every named role recorded.

Should offer letters live in Greenhouse or Lever?

Greenhouse fits when structured permissions and Harvest reads are the buying problem; Lever fits when the desk already works in opportunities and will actually use native approvals.

Can the ATS replace payroll for compensation bands?

No. Greenhouse and Lever store the letter object; payroll still owns classification and withholdings, and the equity tool still owns grants.

When is a no-code router enough for activity records?

Use Zapier, Make, or n8n when the only job is notifying compensation and you will own retries, logs, access, and retention.

What belongs in a 30-day offer-letter pilot?

Freeze 12 letter versions, record 36 signer actions, hold 3 out-of-band packets, and allow 0 premature sends; stop the pilot if application-plus-version keys cannot be rebuilt.

Freeze one letter, then send it

Pick Greenhouse when structured letter control is the operating system. Pick Lever when opportunity-centric hiring and a lighter admin load match the team. Keep native approvals when they already block send. Add a cross-system hold only when payroll or equity must sign the same version.

The team at US Tech Automations can hold a letter on offers.status or Lever approval-sent after you name the ATS edition, the band table, and the reviewer. Review the same US Tech Automations homepage after that proof list is named, then compare a configurable hold on the offer-routing plans. Write the chain on paper before you add another tool.

Public pricing pages were checked 2026-09-04; quote-only SKUs stay quote-only.

Context figure 1 according to FCC G11186 (checked September 4, 2026). Context figure 1 according to NLRB G11186 (checked September 4, 2026). Context figure 21 according to BEA G11186 (checked September 4, 2026).