Desk Review: Toast IQ Goes Multi-Unit

I read Toast’s January product update post the same way I read most vendor announcements — with the kitchen light on, a Wednesday-morning coffee gone lukewarm, three tabs open. Toast’s newsroom dropped the retailer release; Digital Transactions posted its write-up of the Toast IQ updates early on January 8; the BakeMag coverage rolled in alongside; Toast’s own community board put the January 2026 product update notes up for U.S. customers. By breakfast the story had a shape.

The shape is this: first the chatbot, now the brain. Toast spent the back half of 2025 turning the Sous Chef pilot into the Toast IQ assistant — the rebrand Hana covered in October — and ran adoption up to roughly half of Toast’s 164,000 locations inside four months. That was the per-location story. The January release is the multi-location story. Toast IQ now answers questions across an operator’s whole footprint, not just the box on one counter. The marketing slide calls it a “right hand.” Read literally, it is a chain-level operator console — an AI surface that sits where a regional manager used to sit and tells you which store needs the milk run before the GM at that store does.

That is the interpretation you should know I’m reaching for before the module-by-module work. Toast didn’t pitch it as a “chain CIO console.” I’m pitching it that way because, structurally, that is what shipping the multi-unit lookup does — it moves the unit of analysis up a level. Flag that and let’s keep going.

Methodology disclosure (the part Vibe Check requires up front)

A proper Vibe Check is a six-operator panel running the named product against a structured weekly log for ninety days. That’s the bar. We have not cleared it, and we cannot — Toast IQ multi-unit lookup enters beta on January 23, 2026, sixteen days from today as I write this. The rest of the retailer suite is either brand-new or rolling out through Q1. There are no operators with ninety days on this. Nobody has ninety hours.

So this is a Desk Review, the format I lean on when a release is too fresh to panel — the Toast suite desk review is the same shape. I work from the public materials — Toast’s press release, the trade press, the customer community update notes — and I say what I think, module by module, with confidence levels on each verdict. I am explicit about which calls are interpretation and which are reported facts. Desk reads are downstream of vendor framing in a way panels aren’t. When we can run the panel — Q2, once enough chain operators have sixty days on multi-unit — we will, and we’ll publish it beside this one.

A note on sourcing. Toast’s retailer press release and the BakeMag write-up are the canonical announcement copy; the assistant-focused trade piece is on Digital Transactions, linked above. Where I quote a stat I can only source to vendor materials, I’ll mark it as a vendor claim, not an independent measurement.

What actually shipped

Five named modules, plus the platform move underneath them. In order.

1. Toast IQ multi-unit lookup — the assistant can now answer questions across an operator’s locations, not just one. Beta starts January 23, 2026.
2. AI Invoice Scanning — vendor invoices get scanned and parsed into the system without manual entry. Inside Toast IQ.
3. Toast IQ Grow — a $499/month bundle that pairs the assistant with the marketing/advertising stack. Sold as a paid SKU separate from the platform fee.
4. Toast Advertising on Google and Meta — campaign-creation and optimization through Toast IQ for paid acquisition. Pilot data exists; cite carefully (more below).
5. Toast Go 3 — barcode scanning on the handheld terminal, which sounds like a hardware story and is actually a workflow story.

Underneath the modules is the structural move I keep coming back to: Toast IQ has crossed from a per-location product into a multi-location product. Everything that ships from here is going to either lean on that or sit beside it. Now the verdicts.

Module-by-module verdict

Each section names the feature, what it ships, the source link, and a HIGH / MEDIUM / LOW confidence read on operator impact. Confidence here is the same scale we use across Toast suite desk reviews: HIGH means a panel would likely confirm the call, MEDIUM means a panel could easily flip it, LOW means treat the verdict as a hypothesis.

Toast IQ multi-unit lookup — verdict: keep, and use it like a regional manager. (Confidence: MEDIUM)

The headline. Up to January 22, Toast IQ was a per-location product — you asked, it answered in the context of the one store the terminal was sitting in. From January 23, in beta, the assistant looks across multiple locations on the same operator account. Which of my five stores is overstocked on the seasonal SKU? Where do I have a price mismatch between Store 3 and Store 7? That is what the lookup does.

The adoption baseline is the loudest number in the public record. Toast has disclosed — and re-disclosed in the materials around the January release — that more than half of its 164,000 locations used Toast IQ within four months of GA, generating over 8 million queries. We cited the same statistic in the Sous Chef rebrand piece when it first surfaced in Q4. “Used” is a low bar; cumulative queries is a vanity number. The directional read — operators do, in fact, open the chat — is real, and it’s the prerequisite for the multi-unit version landing rather than getting ignored.

Why this matters more for retailers than restaurants: Toast’s retailer base (convenience, specialty grocery, bakeries, bottle shops) is more naturally multi-location at the small-operator end. A three-store wine retailer on one POS account is the median customer for this feature, not an edge case. For those operators, cross-store reporting is the daily pain — Toast’s prior tooling for it was, candidly, spreadsheet exports.

Omri Traub, Toast’s COO for retail, is on the record in Digital Transactions with the “right hand” pitch: “With Toast IQ, operators have an AI-powered ‘right hand’ partner that understands the flow of inventory and pricing as naturally as a store manager does, surfacing helpful, actionable insights at the right time.” (Source: Digital Transactions, Jan 8 2026.) That’s the chain CIO console read the press release is gesturing at without naming.

The risk — and why I’m not at HIGH confidence — is the one we’ll flag again in the broader Toast suite desk review: Toast IQ sees what Toast sees and very little else. A multi-unit lookup that answers cross-store inventory questions confidently while missing third-party distributor pricing or loyalty data in another system is more dangerous than the single-store version, not less. Confidently-wrong cross-store recommendations scale across the footprint.

Reach for it if: you run three or more retail locations on Toast, your data layer is mostly on Toast, and your existing cross-store reporting is a Tuesday-morning spreadsheet ritual. The lookup, even in beta, is a step-change. Hold off if: you’re a single-location operator (the multi-unit framing doesn’t apply yet), or you run a multi-unit footprint where two-plus systems carry materially different pieces of the operation. Wait until the assistant can integrate the other systems.

AI Invoice Scanning — verdict: keep, and watch the error rate. (Confidence: MEDIUM)

The module operators will feel the most this quarter. AI Invoice Scanning lets a manager photograph or upload a vendor invoice and have Toast IQ extract the line items, vendor, totals, and dates without manual keying. Most retailers on Toast still hand-key vendor invoices into a back-office system — roughly the worst use of a manager’s evening the modern POS has failed to fix.

The reason this isn’t HIGH confidence is the failure mode. Invoice scanning works well in the demo and unevenly in the wild — pen-marked deductions, hand-written returns, distributor-specific abbreviations, the occasional invoice printed on a dot-matrix Friday at 11pm. Parse confidence decides whether the feature saves a manager an hour a week or costs her an hour debugging a misread. Toast’s press release frames invoice scanning as eliminating manual data entry; I have not seen Toast publish a parse-accuracy number alongside the launch, which is the number a panel would want.

Reach for it if: you take more than fifteen vendor invoices a week and your current process is a manager photographing them on her phone and entering them later. The floor savings are real. Hold off if: you depend on the invoice data being exactly right for tax or audit reasons and you don’t have a reconciliation step. Add the reconciliation step or wait for a vendor-published accuracy number.

Toast IQ Grow ($499/month bundle) — verdict: too new to keep, too revealing to ignore. (Confidence: LOW)

The module the financial-strategy crowd should be reading most closely. Toast IQ Grow is the first Toast IQ product packaged as a paid SKU separate from the platform fee — $499/month, bundling the assistant with the marketing and advertising stack. Up to now Toast IQ has lived inside the core platform price; I’ve been flagging that the gap would close. This is the first piece of the closing.

Inside the bundle: the marketing/advertising layer (Google + Meta campaign creation), the assistant’s grow-oriented prompts (lapsed-guest, attach-rate, seasonal mix), and access to the multi-unit-aware features as they ship. $499/month is high enough to signal Toast thinks the bundle moves operator revenue, low enough to defend against a freelance marketing operator at three-to-five hours of work a month.

LOW confidence for the same reason I hold marketing-agent verdicts at LOW in the broader Toast suite review: Toast hasn’t published independent measurement against a control of operators with an existing marketing programme. The pilot lift numbers exist; the against-baseline number doesn’t. For a paid SKU, that is the missing piece.

The strategic read — interpretation — is that $499/month is the price you set when you expect a meaningful chunk of operators to renew within twelve months. High enough to fund product investment, low enough that the loss aversion on cancelling after a year is real. Toast is not pricing Toast IQ as a low-friction trial in 2026. It’s pricing it as a commit.

Reach for it if: you have no marketing programme today, the bundle would be your floor not your ceiling, and $499/month is less than your fully-loaded cost of a five-hour-a-month freelance marketer. The agent will give you something. Hold off if: you already run a marketing stack you trust. Wait for an against-baseline number from a panel — ours or someone else’s. Don’t bundle-cancel into a paid SKU on the assumption that the new floor beats your existing programme.

Toast Advertising (Google + Meta) — verdict: pilot data is interesting, vendor-published, and pre-replication. (Confidence: LOW)

The module where vendor-claim discipline matters most. I’ll walk it slowly.

What ships: a Toast Advertising module inside Toast IQ that creates, runs, and optimizes paid-acquisition campaigns on Google and Meta from the operator’s sales-history and customer data. The mechanic is the same agentification I flag across the Toast suite work — the agent doesn’t suggest a campaign, it builds and ships one — applied to the channel where retailers under-spend most chronically.

The vendor claim. Toast told Digital Transactions that “during the pilot phase, retailers running Google Ads campaigns have seen a more than eight-fold return on their ad spend.” That is the 8X ROAS number you’ll see repeated in coverage. Two things to mark on it:

• It is a vendor claim, not an independent measurement. Toast disclosed it; no third party has measured it; the cohort, sample size, time window, and counterfactual aren’t published. Treat it as evidence the feature can produce a result, not as evidence it will produce that result for you. The pilot for an advertising tool is necessarily self-selected toward operators without a working ad programme.
• Eight-fold ROAS in retail paid acquisition is, in absolute terms, a strong number. I’m not saying it’s wrong. I’m saying the operators most likely to volunteer for a Toast Advertising pilot were the cohort with the most ROAS upside — operators with no current paid-acquisition programme — which is the cohort where any halfway competent agent produces an outlier number.

The panel question for Q2: what does Toast Advertising do for an operator running a half-decent Meta programme? If the lift against that baseline is meaningful, the 8X figure is more than vendor flattery. If it isn’t, the 8X is the floor-versus-zero illusion and the bundle is a tool for the bottom third of operators by marketing sophistication, not a category-defining product.

Reach for it if: you spend less than $500/month on paid acquisition today and you’re inside the Toast IQ Grow bundle anyway. The downside is bounded; the upside is the kind of vendor-pilot number Toast is publishing. Hold off if: you already have a paid-acquisition programme you trust. Wait for the head-to-head.

Toast Go 3 (barcode scanning) — verdict: a hardware update with a workflow consequence. (Confidence: HIGH)

The retailer release pairs the software story with a hardware update I almost glossed over: Toast Go 3 now does barcode scanning natively on the handheld. Sounds like a spec sheet item; read against the retailer workflow, it’s the most operationally consequential piece of the release for the bakery / specialty / convenience cohort.

Why: a retailer on a fixed terminal has to choose between adding inventory at the counter and scanning a SKU during receive or count. Toast Go 3 with barcode scanning collapses that decision — the manager walks the floor with the handheld, scans the case, the SKU updates, the count goes in, and the multi-unit-aware Toast IQ surfaces the cross-store implication on the same device. The hardware unlocks the workflow the software was designed for.

HIGH confidence even on a desk read, because “barcode scanning on the device the manager already carries” is a workflow primitive, not a behavioural bet.

Reach for it if: you run a retailer with more than 200 SKUs and your existing scan-to-count flow involves a fixed terminal or a separate scanner. The handheld consolidation is straightforward upside. Hold off if: you don’t yet have the Go 3 fleet. Wait until you’re due for a handheld refresh; don’t replace Go 2s just for this.

The wider read

Three things to put on the table that no single module covers.

The multi-unit move is the structural one. “First the chatbot, now the brain” is the interpretation I want to defend explicitly. Toast IQ in its per-location form was a smart-feature: ask, answer, act, all inside one box. Toast IQ with multi-unit lookup is a console — the operator surface where chain-level questions get answered. Different category of product, not a feature upgrade. Whether Toast makes it the default surface for multi-unit operators is what decides whether 2026 is the year Toast moves from POS vendor to operations-platform vendor for chains. The single-location operators who powered the 164K-locations-50%-adoption number aren’t the cohort that proves this out; the three-to-fifty-store retailers are.

Pricing has crossed the line. I’ve been warning for months that the assistant being inside the platform fee was unresolved and would close within twelve to eighteen months — a thread I’ll pick up again when the broader Toast suite desk review lands. Toast IQ Grow at $499/month is the closing. The shape of Toast’s monetisation is now clear: the assistant stays inside the core fee as a retention layer; the agentic layers — Grow, Advertising, anything that does work on the operator’s behalf — go behind a paid SKU. The operator-evaluation question now isn’t do you pay extra for Toast IQ? It’s which agentic layers do you pay extra for, and which ones are bundled at which contract term? Negotiate the latter. The freebies of 2025 are the upsells of 2026.

Vendor-claim hygiene matters more, not less. The retailer release lives in a moment where the category is producing genuinely interesting numbers — Toast IQ adoption, Toast IQ Grow pilot lifts, the 8X ROAS pilot figure, the 6% upsell number from Q1 2025. Each is real-because-vendor-disclosed and pre-replicated-because-no-third-party-counterfactual. The Vibe Check discipline is to take vendor claims seriously without treating them as measurements. If a Toast rep walks you through the 8X slide this month, ask: how many retailers were in the pilot, over what window, what was their pre-pilot baseline, what did the bottom quartile see? If the answers aren’t on hand, treat the figure as a directional pilot disclosure and price your decision against the bottom of the range, not the top.

A note on the operator quote you may see in trade coverage. Romain Bernus of Le Petit Four Bakery has been quoted in Toast’s customer materials saying complex reports “take five seconds now” — surfaced in Total Food Service’s prior Toast IQ coverage, cited in the rebrand piece. I don’t see Bernus quoted in the January retailer release itself, so I won’t lean on it here — the proper sourcing is to the prior piece, not to a release that didn’t carry it. Vendor materials recycle operator quotes across releases; easy to over-attribute.

Reach for the January release if

• You run a three-plus-location retailer on Toast and your cross-store reporting is currently a manual spreadsheet operation. The multi-unit lookup, even in beta, is the upgrade.
• Your invoice volume is high enough that AI Invoice Scanning saves you a manager-hour a week, and you have a reconciliation step downstream of the scan.
• You have no marketing programme today and Toast IQ Grow’s $499/month would be your floor, not a replacement for an existing stack.
• You’re due for a Toast Go 3 fleet refresh and barcode scanning closes a workflow you currently solve with two devices.
• You already share most of your operational data layer with Toast (POS, inventory, loyalty), and adding the multi-unit assistant is incremental, not architectural.

Hold off if

• You’re a single-location retailer. The multi-unit framing doesn’t apply, AI Invoice Scanning is the only module worth your attention, and waiting one cycle for a vendor-published accuracy number costs you nothing.
• Two or more systems carry materially different pieces of your operation and they don’t all talk to Toast. The multi-unit assistant will give you confidently wrong answers about a question whose context lives in the other system.
• You already run a marketing programme you trust. $499/month into Toast IQ Grow is a real number; wait for an against-baseline measurement.
• You’re a Toast-curious operator on a competing POS. This release doesn’t change the calculus on switching costs; it changes the calculus on what staying costs over the next three years. Re-read the contract terms before you re-read the announcement.

What I’ll change in the panel piece

When we run the ninety-day panel in Q2 — same six-operator scaffolding, mostly multi-unit retailers this time — the questions I want to answer are the ones a desk read can’t:

• How does Toast IQ multi-unit lookup perform against the operator’s own COO or regional manager on cross-store inventory and pricing questions? Where does it diverge?
• What is AI Invoice Scanning’s parse accuracy in production, by vendor type?
• What does Toast IQ Grow’s marketing layer do for an operator with an existing paid-acquisition programme — lift, parity, or drag?
• Does the 8X ROAS pilot number replicate on a non-self-selected cohort, and at what variance?
• Does the multi-unit assistant change the manager role in a multi-unit retailer — and in which direction?

Honest summary from the chair I’m in today, two weeks before the beta opens: Toast’s January retailer release is the most consequential single shift in the Toast IQ surface since the October GA. Multi-unit lookup is the structural move; AI Invoice Scanning the most felt module; Toast IQ Grow at $499/month the most strategically revealing; Toast Advertising’s 8X ROAS pilot the figure to handle with discipline; Toast Go 3 with barcode the workflow upgrade hiding inside a hardware update. Toast frames the suite as a “right hand.” Structurally it’s closer to a chain CIO console — that framing is mine, not Toast’s, but I think it’s the one that explains what was actually shipped.

I’ll be back with the panel in Q2, once a credible cohort of chain operators has sixty days on multi-unit. The chair I’m in today says: directionally, this lands. The chair I’ll be in in June will say whether the chain operators agree.

— Sofia leads vendor reviews for TableTransfers. Tips: sofia@tabletransfers.com.