Lab · business text messaging (A2P SMS) · 2026
Before your bank’s login code reaches your phone, several companies have been paid to carry it. This is a map of how businesses buy application-to-person (A2P) SMS — who buys, what they need, how they meet suppliers, how they decide, and why they stay.
信 xìn — a message; also, trust.
Designed by Zhouyi Xiaoxiao · Yuexin International, 2026
Start with one message ↓Who touches a text between a business and your phone, which way the money and the delivery reports travel, and where ‘delivered’ can go wrong.
Try: switch to the Grey route, then show Delivery reports.
The text passes through one or more wholesale hubs before the operator. More reach, more hops. Gold marks arrival.
Typically 160 GSM-7 or 70 UCS-2 units fit in a single SMS (Chinese text uses UCS-2). When a message is split, each part holds 153 or 67 units, and each part is billed.
A bank, a shop, an app, an airline or a public body. It pays for the message, but what it really buys is an outcome: a finished login, a traveller who knows the gate has changed.
Marketing and CRM tools, cloud platforms, systems integrators and resellers build messaging into software the business already uses. Many businesses never talk to a mobile operator at all.
Offers an API or SMPP connection, chooses routes, returns delivery reports, helps register sender IDs and templates, and sends the invoice. Most of the buying decision happens here.
Buy connections to many operators and resell them wholesale. Each extra hop adds reach — and a margin, a little latency and less visibility into where the message really went.
Some low-cost routes push business traffic through consumer SIM cards or unauthorised interconnects, disguised as person-to-person texts. It looks cheap; operator firewalls block it, delivery turns erratic, and some such routes return delivery reports that are not genuine.
Delivers the message on its own network and is paid a termination fee. Operators and regulators set the rules — registration of sender IDs and templates, consent for marketing — and run SMS firewalls against spam and fraud.
The only place where ‘delivered’ is certain. Serious tests put real handsets on the destination networks and compare what arrives with what the delivery reports claim.
On the chain · 3/7
Offers an API or SMPP connection, chooses routes, returns delivery reports, helps register sender IDs and templates, and sends the invoice. Most of the buying decision happens here.
Five questions, in order. Pick a buyer and the map re-sorts around them.
Try: pick Travel & logistics.
All buyersCentred on the business customer, linking channel partners and upstream supply.In order: who buys → what they need → how they meet → how they decide → why they stay.
Who is paying for the text?
Banks, insurers, payment and fintech firms: identity checks, transaction alerts.
Cross-border sellers, brands and retail platforms: order updates, loyalty programmes.
Social, gaming, tools and SaaS: sign-up, login, account security.
Airlines, travel, couriers and mobility: itinerary changes, delivery and service notices.
Government, public bodies and large companies: notices and bulk services.
Resellers, CPaaS platforms and regional suppliers: buying routes, adding backup routes.
Get the critical message to the right people on time — and prove it.
A technical test, a commercial comparison and a cross-team review; tuning today’s routes while opening new markets.
Stated needs, and the ones nobody writes down.
The right destination and operator; latency, success criteria and peak load tested for each use case.
API or SMPP, status callbacks, error codes, reports and message tracing.
Quote validity, per-segment billing, incident handling, invoices, advance notice of route changes.
Fewer abandoned logins, missed notices and failed deliveries; for marketing, real extra sales.
No complaints, misfires, budget overruns or failed switch-overs; less dependence on one supplier.
One contact for technical, commercial, upstream and support; when something breaks, someone owns it.
Where buyers and suppliers first find each other.
Provider websites, developer documentation, solution enquiries and online sales chat.
CRM, SaaS and cloud marketplaces; messaging built into systems the buyer already runs.
Formal tenders, industry trade shows, introductions by operators and integrators, customer referrals.
Conversation → needs brief → technical test → multi-role review. Not a price list by email.
Operators, aggregators and regional suppliers; check country, operator, use case and limits.
A confidentiality agreement (NDA) and customer identity checks (KYC) → commercial and technical onboarding → test and acceptance.
Who has a say, and in what order.
Define the use case, the customer experience and what success means.
Check integration, stability, permissions and the audit trail — and can veto.
Price, contract liability, credit and payment; the budget holder approves.
A new market, an outage, a contract ending or cost pressure; agree the use case and acceptance criteria.
Build a shortlist; test with the same sample and the same time window.
Contract and integration signed off separately; review after a staged rollout, then renew.
What wins the first order, and what wins the renewal.
Delivery reports, handset spot checks, latency and failure reasons — never a single ‘success rate’.
Documentation, sender-ID and template registration, permissions, incident plans.
Segment billing, charges for failed messages, minimum commitments, payment terms, disputes.
Is quality sustained? Are incidents found, fixed and reviewed quickly?
Can invoices be checked? Are route changes announced in advance? Is support reliable?
Can new countries, use cases and volumes be proven? Can a switch be rolled back?
The trade in one lineCustomers don’t buy message counts. They buy delivery they can verify, risk they can control, and a total cost they can sustain.
Three kinds of message, three definitions of success — and the objection each one answers.
Try: compare Passcodes with Opt-in marketing.
01 · The business outcome
Less waiting and drop-off in sign-up, login and payment checks, judged on real verification sessions.
02 · Metric and denominator
Resends do not create new sessions; abnormal requests and pre-send failures are listed separately. This is not the whole sign-up funnel. Delivery-report latency and handset arrival latency are measured separately.
03 · Objection → evidence
“Your success rate is so high — so why can’t our users log in?”
Line up session IDs, message IDs, submit and delivery-report times and verification results; split failures into network delivery, timeouts, code entry and the app’s own errors.
01 · The business outcome
Orders, flight changes, appointments and payment alerts: fewer missed, duplicated or chased-up notices.
02 · Metric and denominator
Count each ‘business event × intended recipient’ pair once; retries and segments never count twice; unknown or timed-out pairs stay in the denominator, listed separately. With handset sampling, both come from the same sample, and results do not extrapolate to all traffic or to ‘read’.
03 · Objection → evidence
“The invoice says everything was sent — so why is customer service getting complaints about missing notices?”
Reconcile send logs, recipient mapping, templates, segment completeness and timing against business events; add delivery reports and handset spot checks. ‘Submitted’ does not close the case.
01 · The business outcome
Within audiences who properly opted in, measure the extra conversions a campaign causes, while watching opt-outs, complaints and message fatigue.
02 · Metric and denominator
Both rates use everyone originally assigned to each group; undelivered messages stay in their group; report the difference in percentage points. Agree audience, window and exclusions in advance. Without a control group, report association only — not lift.
03 · Objection → evidence
“Cheaper per message and more clicks — so we should renew, right?”
Check incremental conversion, segment and other agreed fees, opt-outs and complaints together. Do not credit SMS with conversions that would have happened anyway, or with every last click.
Suggested definitions; no target values are given. Before a test, agree the sample, time window, exclusions and denominator, and keep supplier delivery reports and business events as separate evidence.
Seven steps, each ending in something you can show. Gold marks the first purchase; renewal loops back to a new trigger.
Try: tap a step, or use ← → to walk the path.
01
A new market, a new product, an unstable route, or a contract about to expire.
OutputWhy now
02
Confirm destinations, use case, peaks, success criteria and budget.
OutputNeeds brief
03
Align credentials, sender IDs, route limits and contract terms.
OutputShortlist and comparison
04
Same use case, same time window, side by side; keep delivery reports and handset evidence.
OutputTest report
05First purchase
Engineering, legal, procurement and finance each sign off.
OutputContract and connection order
06
Increase traffic in agreed stages; be able to switch back.
OutputGo-live acceptance record
07
Review quality, cost, reconciliation and service; decide on renewal and growth.
OutputReview and growth plan
↺ Renewal starts the next trigger
Before the price
A useful first conversation settles these first. The checklist carries the buyer and the kind of message you picked above.
A new market, an outage, a contract ending, or cost pressure?
Destination countries and operators, message type, language and sender.
Daily volume, peak hours, throughput and the go-live window.
Agree submission, delivery reports, handset checks and business results separately.
Business, engineering, security, procurement, legal and the budget holder.
Sustained quality, total cost, responsiveness, reconciliation and room to grow.
A framework, not a survey. Buyer types, needs and decision roles are analytical hypotheses drawn from common industry practice — not interview findings, a customer list or anyone’s revenue mix.
The page names no clients or prices, and makes no claim about coverage, quality or service levels. The first version was drawn in September 2026 as a working map for a messaging sales team; this version keeps its logic and removes everything company-specific.
Version 2 · redrawn 30 September 2026 · first drawn 11 September 2026
Public documentation, cited for definitions — not a list of suppliers.