MM Lite API is Meta's marketing-only send path: you post marketing-category templates to a separate endpoint and Meta applies its own delivery optimisation before the message reaches the user, while the standard WhatsApp Cloud API keeps handling utility, authentication, service replies and every conversational flow. For almost every Indian business in 2026 the correct answer is therefore "run both, split by template category" — not "switch off Cloud API".
This guide covers what MM Lite genuinely changes in delivery, what it provably does not change, the rupee cost model at real Indian volumes, a six-stage campaign lifecycle you can automate, and an honest list of who should stay on plain Cloud API sends.
What MM Lite actually is
Marketing Messages Lite API is an additional send surface on the same WhatsApp Business Account. You keep the same WABA, the same phone number ID, the same access token and the same webhook subscriptions. The difference is the destination: marketing templates are posted to Meta's marketing-messages endpoint instead of the general messages endpoint, and Meta then decides how and when — and in some cases whether — that message is pushed to the recipient.
That last clause is the entire product. On the standard Cloud API, an accepted send is a committed send: barring a hard error, Meta attempts delivery immediately. On MM Lite, acceptance means Meta has taken the message into an optimisation layer. It may deliver it now, hold it for a better moment, or suppress it if its models judge the recipient unlikely to engage or already saturated with marketing that period.
Three things MM Lite does not do
It does not bypass opt-out. A user who blocks you, taps "Stop promotions", or reports the business is still removed from marketing reach, and that signal still feeds your quality score.
It does not bypass template review. Every template still goes through the same approval and category classification. A template rejected for Cloud API is rejected for MM Lite.
It does not exempt you from quality rating or messaging tiers. Your per-day unique-recipient limit and your green, yellow or red rating are account-level and apply to both paths identically.
Anyone selling MM Lite as a ban-proof or filter-proof channel is selling something that does not exist. It is a delivery-optimisation path for messages you were already permitted to send.
MM Lite vs Cloud API: the honest feature split
| Dimension | Cloud API (standard send) | MM Lite API |
|---|---|---|
| Template categories | Marketing, utility, authentication, service messages | Marketing templates only |
| Opt-in requirement | Required | Required — identical standard |
| Template review | Required | Required — same queue, same rejections |
| Quality rating and messaging tiers | Applies | Applies — shared account-level rating |
| 24-hour service window | Opened by a user reply; free session messaging | Same window, opened the same way |
| Inbound replies | Standard webhook | Standard webhook — replies are indistinguishable downstream |
| Send timing control | You control it to the second | Meta may retime or hold the send |
| Delivery of an accepted send | Attempted for every accepted message | Not guaranteed — optimisation may suppress |
| Advanced template formats | Broadest support | Support lags; verify carousels, coupons and limited-time offers before you depend on them |
| Best fit | Everything transactional plus time-critical promos | High-volume, non-time-critical promotional pushes |
Because both paths share the same business account, a badly targeted MM Lite campaign damages your utility and OTP delivery too. That interlock is the single most under-appreciated risk in a migration, and it is why recovering a red quality rating should be finished before you touch MM Lite at all.
What actually changes in delivery, and what does not
The measurable change is the shape of your funnel, not a uniform lift. Fewer messages leave the building; the ones that do tend to land on people more likely to open them. If you report on messages sent, MM Lite will look like a regression. If you report on reads, replies and revenue per thousand attempted, it may look like a win. Both readings come from the same run.
Acceptance is not delivery
Your send worker will receive a success response for messages that never produce a delivered status. Any reconciliation job that treats accepted-with-no-delivery-status as a bug will fire false alarms all night. Classify it as a distinct outcome — suppressed — and count it separately from failed.
Throughput planning changes
Burst pacing loses much of its point when Meta is retiming sends anyway. Your worker pool still needs to respect rate limits and back off cleanly on transient errors, but the elaborate second-by-second scheduling described in campaign throughput engineering matters less on the MM Lite path and just as much on everything still routed through Cloud API.
Treat vendor uplift figures as unverified
Meta and several partners have published encouraging early-adopter numbers for read rate and conversion. Those are vendor-reported, drawn from self-selected large senders, and there is no independent India-specific dataset in the public domain as of 2026. Do not budget against them. Run a holdout instead: split one audience eighty-twenty, send the same template in the same window, and compare reads, replies and attributed orders per thousand attempted. Twenty thousand messages per arm over two weeks is enough to see a real difference.
The six-stage marketing lifecycle, mapped to automation
1. Campaign build
Draft the template, confirm Meta will classify it as marketing (a promotional line inside a utility template gets reclassified and repriced), and submit for review. Automate the approval poll so a campaign cannot be scheduled against a pending template.
2. Audience assembly
Pull the segment, then subtract: unsubscribed users, users messaged in the last N days, and anyone without a recorded opt-in source and timestamp. Deduplicate on the WhatsApp ID, not the CRM row — the same person often exists three times. This stage kills more bad campaigns than the send stage ever will.
3. Send
Route by category at the queue level: marketing templates to MM Lite, utility and authentication to Cloud API, session replies to Cloud API. One routing function, one config flag, so a rollback is a flag flip and not a deploy. Respect your daily unique-recipient cap — see WhatsApp broadcast limits and tier upgrades for how those caps step up.
4. Delivery outcome
Persist every status callback and reconcile after a fixed window — six hours is a practical cutoff. Buckets: delivered, read, failed with an error code, and suppressed. Dashboards should show attempted, delivered and read side by side, because delivered-over-attempted is the number that moves the moment you enable MM Lite.
5. Reply handling
A reply opens the free 24-hour service window regardless of which path sent the original message. Auto-route it: clear intent to a bot flow, anything ambiguous to a human queue. This is where the campaign actually earns money, and where most Indian teams under-invest — a two-minute first response converts several times better than a two-hour one.
6. Re-engagement
Non-responders go into a cooldown, not a resend loop. Suppressed recipients should be cooled longer, not chased harder — Meta already told you they are saturated. Re-approach through a genuinely transactional utility message when there is a real event such as an order, a renewal or an appointment, never a marketing message dressed as one.
Get a 1-minute BSP audit on WhatsApp
Drop your WhatsApp number — we line-item your current invoice against Meta India rates in under 60 seconds. India-hosted, DPDP-compliant.
Cost and ops trade-offs in rupees
MM Lite is not a discount. Marketing messages delivered through it are billed at Meta's standard India marketing per-message rate, the same as a Cloud API marketing send. The only structural saving is that a suppressed message is never delivered and therefore never billed — real money at scale, but a side effect of Meta filtering your audience, not a lower price.
Meta's own India rates move periodically and should be read from WhatsApp Manager on the day you budget, not from any blog. What is fixed is our side of the bill:
RichAutomate pricing, usage-only:
- ₹0 setup — onboarding, business account connection and number migration cost nothing.
- ₹0 monthly — no seat licence, no minimum commitment, no platform subscription.
- Client Pay: ₹0.10 per message — platform fee only. You hold the Meta billing relationship and pay Meta directly at their rate.
- SaaS Pay: ₹1.20 per marketing message, ₹0.30 per utility message — all-inclusive, Meta cost plus platform in one line item.
- 24-hour service window: free — unlimited session messaging once a user replies, on either model.
At Indian volumes the billing model choice dominates the routing choice. Three worked scenarios, platform side only:
| Monthly marketing messages | Client Pay platform cost at ₹0.10 | SaaS Pay all-in at ₹1.20 | Typical profile |
|---|---|---|---|
| 50,000 | ₹5,000 plus Meta billed direct | ₹60,000 | Single-city D2C brand, weekly drop |
| 2,00,000 | ₹20,000 plus Meta billed direct | ₹2,40,000 | Regional retail chain or edtech cohort push |
| 10,00,000 | ₹1,00,000 plus Meta billed direct | ₹12,00,000 | National marketplace or large BFSI campaign engine |
Above roughly two lakh marketing messages a month, Client Pay is almost always cheaper because you buy Meta capacity at Meta's rate with no markup; below that, SaaS Pay removes an entire finance workflow for a modest premium. The full decision tree is in Client Pay vs SaaS Pay billing.
Delivery funnel: observed ranges, not a study
The table below reflects ranges Indian operators commonly see on opted-in marketing broadcasts. It is field observation across mixed verticals, not a controlled study, and your list quality will move every number in it.
| Funnel step | Cloud API marketing broadcast | MM Lite marketing broadcast |
|---|---|---|
| Delivered as a share of accepted | 88-97% | Lower and more variable — optimisation suppresses part of the audience |
| Read as a share of delivered | 25-45% | Often higher, since low-propensity recipients were filtered out |
| Click or reply as a share of delivered | 1-6% | Similar to somewhat higher; the honest answer is to measure it on your own list |
| Cost per delivered message | Meta marketing rate | Same rate; total spend falls only because fewer messages are delivered |
Who should switch, and who should not
Switch marketing traffic to MM Lite if you send more than about one lakh marketing messages a month, your opt-in provenance is clean and auditable, your promotions are not time-critical to the minute, your reporting layer can carry a fourth outcome bucket, and your quality rating is currently green.
Stay on standard Cloud API sends if you send under ten thousand marketing messages a month, because the operational cost of a second path exceeds the gain; if your volume is mostly utility or authentication; if your promotions must land at an exact moment such as an 8pm flash sale or an event reminder; if your quality rating is yellow or red; or if your billing reconciliation cannot yet distinguish suppressed from failed. Fix the reporting first — the endpoint will still be there.
One more group deserves a mention: businesses whose marketing messages are really transactional. If your template is an order update with a coupon bolted on, split it. The transactional half belongs on the utility path at ₹0.30 under SaaS Pay, and the promotional half belongs in a separate marketing send. Merging them costs more and reads worse, and it puts a transactional message behind an optimisation layer that was never meant to gate it.
Migration playbook
Migrate one phone number at a time, and keep the Cloud API path live behind a config flag for the whole trial. Do not attempt to migrate utility or authentication traffic — it is not eligible, and trying only generates errors. Reuse existing approved templates rather than resubmitting; the approval carries over. Dual-write your reporting for at least two weeks so you can compare like with like, and watch the quality rating daily during that window rather than weekly.
Keep the rollback trivial. If read rate does not improve, or the rating dips, flip the flag and every marketing send returns to the standard endpoint on the next queue tick. If you are also changing providers in the same quarter, sequence it properly — the checklist for migrating your WhatsApp BSP assumes you are not simultaneously rewriting your send path, and for good reason.
Compliance: the rules MM Lite does not rewrite
Under India's Digital Personal Data Protection Act, consent must be free, specific, informed and withdrawable, and you must be able to show when and how it was obtained. Store the opt-in source, the timestamp and the exact wording shown to the user against every contact. Honour opt-outs within the same business day, across every channel, and never re-add an unsubscribed number through a fresh spreadsheet import.
Meta's commerce and business policies apply unchanged: no restricted goods, no misleading claims, no messaging users who never opted in. The enforcement mechanism is identical on both paths — blocks and reports drag your quality rating down, the rating drags your messaging tier down, and a lower tier caps every campaign you run afterwards. MM Lite optimises delivery inside those rules; it does not soften them.
Running both paths without doubling your ops load
The clean architecture is one campaign engine, one contact store, one consent ledger and one reporting pipeline, with a routing function that picks the endpoint from the template category. That is the design RichAutomate ships: marketing traffic can be pointed at the optimised path while utility, authentication and every service-window reply continue over Cloud API, with a single dashboard reconciling attempted, delivered, read, replied and suppressed across both. The engineering cost of the second path is one routing function and one extra status bucket — everything else, from consent records to agent inbox, is shared.
Billing stays usage-only — ₹0 setup, ₹0 monthly, ₹0.10 per message on Client Pay or ₹1.20 marketing and ₹0.30 utility all-in on SaaS Pay, with the 24-hour service window free on either model. No seat licences, no minimum spend, and no extra charge for running a second send path.
Ready to route marketing through the optimised path while keeping transactional messaging exactly where it is? Create your RichAutomate account and connect your business number — setup is free, billing is usage-only, and you can run an MM Lite holdout against your existing Cloud API sends from the very first campaign. Everything lives at richautomate.in.