Search "WhatsApp open rate" and the same figure appears on vendor sites, agency blogs and comparison pages alike: 98%, usually paired with email's supposed 20%. It's repeated so consistently, with such a precise-looking decimal-free round number, that it reads like an industry-standard benchmark everyone has agreed on.
We tried to find the study behind it — the original report, the sample size, the measurement window — the way we did for the "5-minute speed-to-lead rule" and the webinar no-show statistics in our other fact-checks. Nothing traces back. What does exist is more useful: Meta's own (undocumented, inconsistent) metric definitions, and a handful of real, dated benchmarks built from actual delivered-message data across hundreds of brands. Here's what they show.
Key takeaways
- No original study, report or named methodology could be located behind the "98% WhatsApp open rate" figure. The closest traceable origin is an unlinked Mobilesquared estimate of "90–98%" and, separately, vendor material from MessengerPeople (now Sinch Engage) — neither with a published sample size or measurement window.
- Meta's own WhatsApp Business Platform documentation defines a "Read Rate Benchmark" as the 75th percentile of read rates among similar businesses (by country, industry and template category) — a private, relative number shown to each business about itself. Meta does not publish a public average or median read rate anywhere, so no one citing "Meta says 98%" is citing anything Meta has actually said.
- The largest real benchmark we found — a 2026 study of 382–414 active WhatsApp Business accounts totalling 11.8 million delivered messages — puts the median campaign read rate at 72.05% of delivered messages, with a volume-weighted average of 79.6%. Its own stated conclusion: "No account reaches the often-quoted 98% as a campaign average."
- The real number moves enormously with list quality, not with the channel itself: one audit spanning 40 real campaigns found roughly 90–94% read rates on tightly segmented opt-in lists, around 68% on an average opt-in broadcast, and about 38% on a broad, unsegmented send — a roughly 2.4x spread that a single flat "98%" erases completely.
- Even without the inflated number, WhatsApp's real, measured range (roughly 65–80% for well-run campaigns) still comfortably beats email's real range (roughly 20–35%, itself inflated since Apple started pre-fetching images in 2021) — the honest comparison just needs the honest numbers on both sides.
Where does the 98% number actually come from?
We checked the usual suspects: WhatsApp Business Platform vendors, agency "state of WhatsApp" reports, and the handful of research firms that get credited in passing. None of them hold up to a direct check.
The most common secondhand attribution is to Mobilesquared, a UK mobile-messaging research firm, for an estimated range of 90–98%. Multiple current marketing pages repeat this attribution, but none link to an actual Mobilesquared report, and no original document with that figure could be located. A second attribution chain, traced by at least one independent analysis, points to marketing material published years ago by MessengerPeople (acquired by Sinch and now operating as Sinch Engage) — again with no sample size, methodology, or measurement window ever attached to the number.
There's a plausible third explanation, though it's unconfirmed: WhatsApp marketing content may have simply inherited an older claim from SMS marketing, where "98% open rate" has circulated since at least the early 2010s — attributed inconsistently, and just as unverifiably, to Gartner, Forbes and a site called Crazy Egg. SMS has no open-tracking pixel at all, so even the SMS version of this number was always an estimate dressed up as a measurement, not something anyone actually counted.
- Mobilesquared "90–98%" estimate: cited by name across dozens of current pages, linked to an actual report by none of them.
- MessengerPeople / Sinch Engage material: one independent trace credits it as the likely original source, with no public methodology ever attached.
- The parallel "98% SMS open rate" claim: attributed, inconsistently, to Gartner, Forbes or Crazy Egg — none of which could be confirmed as the actual origin, and SMS has no technical way to measure an open at all.
90–98%
the vague, unlinked range a Mobilesquared estimate is most often credited with — the closest thing to an "origin" the 98% figure has, cited by name on dozens of current pages with no published sample size, methodology or date attached to it on any of them.
Source: Secondary attribution across current WhatsApp marketing content (checked October 2026); no primary Mobilesquared report could be located
Does Meta itself actually publish a 98% figure?
No — and this is worth stating plainly, because "Meta says" is exactly how the number gets laundered into credibility on vendor pages. Meta's WhatsApp Business Platform developer documentation does define a read-rate metric, but it isn't a public average at all: it's the 75th percentile of read rates among a business's own peer group, grouped by country or region, industry, and template category (marketing templates compared only against other marketing templates), recalculated daily from the trailing 30 days. It's shown privately, to each business, about its own account — never published as one worldwide number.
Meta's own documentation isn't even fully consistent on the base metric underneath that benchmark. The example response in the current developer docs lists a template with 265 messages marked "read" and a reported read rate of 481.818182% — a figure over 100%, with no explanation anywhere on the page for how a read count can exceed the delivered count it's supposedly a share of. If the platform's own reference example doesn't reconcile cleanly, a clean, memorable "98%" was never going to come from this source.
| What you'll see cited | What Meta's documentation actually defines |
|---|---|
| "WhatsApp's open rate is 98% (Meta)" | Meta publishes no public average or median read rate anywhere in its developer documentation or business help center. |
| "Read Rate Benchmark" | The 75th percentile of read rates among similar businesses (same country/region, industry, template category), recalculated daily over a trailing 30-day window — a private comparison, not a headline figure. |
| "Read Rate" (the underlying number) | Defined as messages read as a share of messages delivered — but the platform's own current documentation example reports a rate above 100%, unexplained. |
One more gap worth knowing: a recipient who has turned off read receipts in their own WhatsApp settings is counted as not having read the message at all, even if they opened it — meaning every read-rate figure on this page, Meta's own benchmark included, is a floor, not a ceiling.
So what does real, measured data actually show?
This is the part with real evidence behind it — not a borrowed estimate, but dated studies built from actual delivered-message logs, with the sample size and method disclosed.
The largest one we found tracked 382 to 414 active WhatsApp Business accounts (the count depends on the exact cut of the data) from June 2025 onward, then switched its own measurement method in March 2026 to count reads against delivered messages instead of sent messages — a change the study discloses, and one that alone lifted the reported numbers by double digits with zero change in actual recipient behavior. The headline figures below are the delivered-basis ones, covering the roughly three months after that change: 11.8 million delivered campaign messages across the 382-brand cohort. Its own published conclusion, stated directly: no account in the dataset reaches the widely quoted 98% as a campaign average, at any point in the measurement window.
| Source | Sample | What it measured | Figure |
|---|---|---|---|
| 2026 cross-brand WhatsApp benchmark study | 382–414 brands, 11.8M delivered messages (delivered-basis window, Mar–Jun 2026) | Campaign messages read ÷ delivered | Median 72.05% · volume-weighted average 79.6% |
| Independent 40-campaign audit (2025–2026) | 40 real WhatsApp campaigns, cross-checked against platform-reported averages | Read rate by list-segmentation tier | 90–94% (tight opt-in) · ~68% (average opt-in) · ~38% (broad, unsegmented) |
| Business-account anecdote sourced to Meta data | Not independently disclosed | Read rate among "strongly engaged" accounts | 65–80% |
The spread matters more than any single average: a business sending to a narrow, warm, well-consented list can genuinely land in the low-90s — which is likely part of why the 98% myth persists at all. A business blasting a broad, stale list can fall to the high-30s. "WhatsApp open rate" isn't one number; it's a range that tracks list quality, not the channel.
Does WhatsApp still beat email, without the inflated number?
Yes — the real numbers on both sides still favor WhatsApp by a wide margin, they're just smaller and messier than the clean "98% vs 20%" headline suggests. Our own email open-rate research puts a healthy marketing email open rate at roughly 20–35%, and flags that even that figure runs inflated: since Apple's Mail Privacy Protection shipped in 2021, Apple Mail pre-fetches tracking pixels automatically, so a meaningful share of "opens" never involved a human looking at the email at all.
Line up the honestly-sourced numbers for both channels side by side and WhatsApp still wins clearly on both the read/open side and on what happens after someone reads it.
Read / open rate
Click / response rate
Both channels' real numbers are lower than their respective marketing-blog folklore — email's "98% never happened" problem has a mirror in its own inflated opens post-2021. The honest version of this comparison still says "use WhatsApp for anything time-sensitive," it just doesn't need an invented decimal to make the point.
If 98% isn't real, why do some businesses swear they see it?
Because for a narrow enough audience, something close to it genuinely happens — just not as a campaign-wide average, and not forever. The 90–94% range in the segmented-list audit above belongs to businesses messaging people who already know them: recent customers, confirmed appointments, an active support thread. Meta's own benchmark system quietly encodes the same idea by comparing utility and authentication templates separately from marketing templates, because a one-time passcode or an order-confirmation message gets read at a completely different rate than a promotional broadcast ever will.
The myth isn't that a high number is impossible. It's that marketing pages take the number from the best-case, narrowest-audience scenario and present it as the channel's baseline — the same way a single study's best subgroup result gets repeated as if it were the whole finding.
This is also the most actionable fact in this entire piece: the lever that moves your real number from the high-30s to the low-90s isn't the channel, it's how deliberately you build the list you're sending to.
What should you actually track instead of chasing 98%?
- 1Track delivered, read and clicked/replied as three separate numbers. Collapsing them into one "open rate" is exactly how a borrowed estimate replaces a measurement in the first place.
- 2Before comparing your number to anyone else's, ask whether it's calculated on a delivered or a sent basis — the same underlying data can report roughly 15 percentage points higher just by changing the denominator, with zero change in actual recipient behavior.
- 3Separate your utility and transactional templates from your marketing broadcasts when you benchmark. They answer different needs and were never going to read at the same rate.
- 4Benchmark against a defensible range — roughly 65–80% delivered-to-read for a well-segmented marketing campaign — rather than a single number nobody can source.
In oToK, a WhatsApp campaign is sent to an audience built on top of per-channel consent — dynamic or static, with the recipient estimate matching the actual send — and the report builder's campaign-logs dataset keeps delivered, read and clicked as separate, queryable numbers instead of one blended figure. The point isn't to manufacture a bigger number; it's to see the real one.
Frequently asked questions
- Is the "98% WhatsApp open rate" statistic real?
- No primary study, sample size or methodology could be located behind it. The closest traceable origin is an unlinked Mobilesquared estimate of "90–98%" and, separately, older marketing material from MessengerPeople (now Sinch Engage) — neither with a published source. Meta itself publishes no public average read rate at all.
- What WhatsApp open or read rate should I actually expect?
- The largest real benchmark we found (382–414 brands, 11.8 million delivered messages) puts the median at 72.05% of delivered messages, with a volume-weighted average of 79.6%. Expect lower (around 38%) on a broad, unsegmented send and higher (90%+) on a tightly segmented, warm opt-in list — the range depends on your list quality far more than on the channel itself.
- Does Meta publish an official WhatsApp open rate benchmark?
- Meta's WhatsApp Business Platform documentation defines a "Read Rate Benchmark," but it's the 75th percentile of read rates among a business's own peer group (by country, industry and template category), shown privately to that business — not a public average. No figure Meta publishes anywhere supports the commonly cited "98%."
- Does WhatsApp still beat email on open rates once you use the real numbers?
- Yes. Using verified figures instead of the mythical ones, WhatsApp's real median read rate (around 72%) still comfortably beats a healthy email open rate (roughly 20–35%, itself inflated since Apple Mail began pre-fetching tracking pixels in 2021). The honest comparison still favors WhatsApp — it just doesn't need an invented number to make the case.
