Multichannel lift over email only: what 8,714 sends actually show
By Jānis Plūme, Founder, Outbound Pros · 2026-08-06
Quick answer
Multichannel lift is the difference in outcome between an outbound sequence run on one channel and the same motion run across two. In the Outbound Pros fleet snapshot dated 2026-07-06, sequences tagged as a multichannel motion produced a 0.37% positive reply rate across 8,714 emails sent, 7.36x the fleet baseline for that snapshot. The follower sourced motion in the same snapshot ran at 0.14% across 52,786 sends, 2.85x that baseline. That is roughly 2.6 times the rate, measured in one fleet over one period, against a denominator that structurally favours multichannel. It is an observation, not a controlled test.
Segment reporting on a live B2B client fleet, snapshot dated 2026-07-06.
What is multichannel lift?
Multichannel lift is the change in outcome from running a sequence across two channels instead of one, holding the list and the offer constant. That last clause is where almost every published lift number falls apart, because almost nobody holds anything constant.
Underneath it sits the problem this page exists to fix. Lift measured per prospect entered and lift measured per message sent are different quantities, and they can point in different directions on the same campaign. A multichannel sequence sends fewer emails per prospect, because some touches are LinkedIn touches, so a denominator of emails sent moves the rate before anything about the channels compounding has happened. Almost every figure in circulation here is the per send version described as the per prospect version. Ours is too, and we say so below and not in a footnote.
What did the 2026-07-06 fleet snapshot actually measure?
It measured positive replies divided by emails sent, cumulative to that date, across every client sending workspace active in the fleet at the time.
| Segment, 2026-07-06 fleet snapshot | Emails sent | Positive rate | Versus fleet baseline |
|---|---|---|---|
| Multichannel motion | 8,714 | 0.37% | 7.36x |
| Follower sourced motion | 52,786 | 0.14% | 2.85x |
A positive reply means the prospect expressed interest. It does not mean a meeting was booked, attended, or closed. Anyone quoting the number should carry that sentence with it.
"Multichannel motion" is derived from sequence names, because names are the only segment signal in the underlying data. A sequence counts as multichannel because it was named as one, so a multichannel sequence named something else is invisible here and a misnamed one is misclassified. That is a real limitation of the instrument and the biggest single reason this is published as an observation and not a finding.
The two rows are the two motions we publish from that snapshot. Neither is described here as the strongest or the second strongest, because I do not have the full ranking inside that snapshot in front of me, and this site does not claim a position it has not verified.
One check you can run without our data: 0.37 divided by 7.36 and 0.14 divided by 2.85 both put the fleet baseline near 0.05% positive on sends. The two agree to within rounding, which should be true of any published table and often is not. That 0.05% is derived arithmetic and not a figure we measured and are reporting, and this is the only page that shows the derivation.
The row that cuts against this page
The best performing single sequence in the fleet ran at 0.45% positive on 6,674 sends, which is a higher rate than the multichannel segment average above. It is a single sequence rather than a segment, so it does not belong in the same table, and it belongs on this page because a benchmark page that quietly drops the inconvenient number is a brochure.
Whether that sequence ran as a multichannel motion is not something I can state from the snapshot. If it did, this page gets stronger at no cost. If it did not, the argument here is materially weaker and the page has to carry that. I would rather publish the row unexplained than quietly assign it to the segment it would flatter.
I did not go looking for the headline number. We built segment level reporting because we needed to know what to kill, not because we wanted a marketing statistic. It runs as one of 28 standing agent desks covering monitoring, analysis, content and drafting. The multichannel segment came out at the top of the table I was reading, above segments carrying six times the volume, and my first reaction was that the query was broken. It took a week to conclude the query was fine and the denominator was the thing worth writing about.
Why does the denominator flatter multichannel, and by how much?
It flatters multichannel because the denominator is emails sent while the numerator counts positive replies arriving on either channel. A multichannel sequence can produce a positive on LinkedIn that is counted against a denominator LinkedIn never contributed a unit to. An email only sequence cannot. This is the mechanism by which any per send comparison between a one channel and a two channel motion is biased in a known direction before the campaign starts.
Three statements about it, in descending order of confidence.
- The artefact is real and it is not the whole gap. The multichannel segment ran at roughly 2.6 times the other motion in the table, and LinkedIn volume inside these sequences is small relative to email volume, which bounds how much of the gap the artefact can explain.
- I cannot decompose it from this snapshot. That needs positives split by arriving channel, measured against prospects entered instead of emails sent, and the snapshot does not carry the first field. Whether that field is recoverable from the underlying reply data for this period is an open question I have not answered, and it is the cheapest thing anyone could do to make this page better.
- The fix is a controlled split, and we have not run one.
Is 0.37% positive on sends a good rate?
No, taken as an absolute number, and that is a scope difference rather than a contradiction. It sits below the 0.5% mark where we kill a sequence instead of iterating on it, so by that rule the segment average would be killed on sight.
It is not a contradiction because the threshold judges a live sequence in the moment, while these are cumulative aggregates including every angle test ever run under the label, most of which we killed. The multiple tells you the relative story. The absolute number tells you the honest one.
The practical use for a buyer is short: if an agency shows you an aggregate lifetime positive rate substantially above these, ask what is in the denominator before you ask anything else. Usually positives have been divided by replies rather than sends, which is a different quantity. The full threshold ladder and the rate definitions underneath it belong to our sibling property allboundpros.io, which has not published yet, and that is where a reader should go for what a workable positive rate actually is once it does.
How do you check any published multichannel lift claim? The Denominator Test
The Denominator Test is five questions you can put to any lift number, including this one, in about ninety seconds.
| # | Question | What a bad answer looks like |
|---|---|---|
| 1 | What is the numerator a count of? | "Responses." Responses to what, meaning what? |
| 2 | What is the denominator a count of? | Missing, or "contacts", which is not a unit |
| 3 | Are both arms measured against the same denominator? | Positives from two channels counted against one channel's sends. Our own failure, and we say so |
| 4 | Was assignment random, or are you comparing whatever existed? | Almost always the second. Ours is the second |
| 5 | Over what period, and what is the n on the smaller arm? | A percentage with no n behind it |
Scoring our own number: it passes 1, 2 and 5, fails 3 because the arms do not share a denominator, and fails 4 because nothing was randomised. Three passes, two clean fails, published. That is a worse score than most vendors would admit to, and it is a score, which is more than they give you.
What would settle it, and what we have committed to running
A controlled split: same list, same copy, same window, random assignment, one arm email only and one arm email plus LinkedIn, measured on positives per prospect entered instead of per email sent. Changing the denominator to prospects removes the artefact entirely, because a prospect enters the sequence once regardless of how many channels touch them.
Putting the design on record before running it is the part that matters. Pre-registration is standard wherever causal claims are taken seriously, and the conventions are public: the Center for Open Science preregistration guidance covers the form, and the CONSORT statement covers what a randomised trial has to disclose. Nobody in outbound does this, which is why it beats another case study.
On record: the metric is positives per prospect entered, arms are assigned randomly at the prospect level from one list, copy and window are identical, the minimum sample is fixed before the first send, and we publish the result whether or not it favours multichannel. A null gets the same page as a win.
What is not settled is which client account and matched list it runs on, and whether that client will agree to the anonymised result being published including a null. Until both are agreed this is a design on record and not a trial in progress, and describing it as anything else would be the same overclaiming the rest of this page is written against.
What does the LinkedIn side of the same fleet look like?
LinkedIn replies far better per touch and delivers a fraction of the volume, which is the structural fact the cadence argument rests on. On one client programme in professional services we measured 59% connection request acceptance, roughly 9% reply rate on LinkedIn DMs, and roughly 1.5% email reply rate, on the same accounts over the same period.
That dataset belongs to our sibling property and is published in full with its methodology at the same accounts comparison on linkedpros.io. It explains why the channels are not substitutes: you spend the abundant channel finding out who is warm, and the scarce one on them.
How many touches does it take to book a B2B meeting?
No touch count answers this, because the binding constraint is angles rather than touches. Sequences run out of distinct reasons to make contact long before they run out of slots, and touches past that point spend familiarity the earlier ones built. Full working on sequence length and follow up count. How many sends produce a given number of meetings is volume maths, and it belongs to our sibling property allboundpros.io, which owns pipeline arithmetic and has not published yet.
What this number does not prove, and who should ignore it
It does not prove that adding LinkedIn to your sequence will multiply your reply rate. It shows that in one fleet, over one period, a coordinated multichannel motion measured better than the other motion published from the same snapshot, against a denominator that helped it. Worth knowing. Not the same claim. Four groups should read this and do nothing.
Teams whose problem is the offer. If three genuinely different angles have produced near silence across a few thousand sends, no channel structure fixes it. You get the same silence on two surfaces, at higher cost.
Teams with one LinkedIn seat and a very large list. Divide prospect count by daily LinkedIn ceiling. If the answer in working days exceeds your window, LinkedIn cannot open the sequence and probably cannot serve the list at all. That is the first check in the capacity planning guide.
Teams that need clean channel attribution for a board slide. You cannot get it, neither can we, and the number handed to you elsewhere is last touch reporting wearing a causal claim.
Anyone below roughly a $10K deal size, or running B2C or ecommerce. The economics do not work and a self serve motion beats us on the same budget.
One timing point most vendor content skips: nothing in that table happened in week one. Fresh infrastructure needs weeks of warm up before it carries volume, onboarding takes about 21 days, and a client approval round sits between the copy and the first send. The ramp arithmetic is worked through on the capacity page, and the infrastructure discipline underneath it is the parent's subject rather than ours.
What to actually do with this
Settle the arithmetic before the strategy, because the arithmetic is the part with an answer. Take your prospect count, your daily LinkedIn ceiling and your window, and find out whether a two channel sequence is available on this list at all. If it is, the shape is our WideNET motion: email opens across the full market, LinkedIn escalates onto the slice that showed a signal. If the list is small, high value and signal triggered, the shape is Spearhead and LinkedIn opens. The free Sequence Cadence Builder does that division and returns a day by day plan, ungated.
If you would rather hand the motion over, the group's agency runs both channels as one sequence for 36 active B2B clients, and the list building sits inside the same engagement: the managed programme behind those sequences. The strategic case for running both, and the parent's own worked fourteen day example, sits with the parent's writing on multichannel sequencing. Read that for the argument and this page for whether the number behind it holds up. The wider service picture is on the parent agency's main site, and the methodology caveats are on the results and methodology note.
Free, no signup, and nothing you type leaves your browser. It answers whether a two channel sequence is available on your list before you design a touch.
Frequently asked questions
Does adding LinkedIn to a cold email sequence actually increase replies?
In our fleet snapshot dated 2026-07-06, a coordinated multichannel motion produced 0.37% positive on 8,714 emails sent at 7.36x the fleet baseline, while the follower sourced motion ran at 0.14% on 52,786 sends at 2.85x that baseline. That is an observation from one fleet over one period, not a controlled test, and part of the gap is the denominator artefact described above. Our data says probably yes and cannot say how much.
What reply rate should I expect from multichannel versus email only?
Do not use these figures as a target. They are cumulative aggregates including every angle test we killed, so they sit below what a live scaled sequence produces. What counts as a workable positive rate on sends is a rate definition question and it belongs to our sibling property allboundpros.io, which has not published yet, rather than here.
Is 8,714 sends a big enough sample, and why is it smaller than the other segment?
It is a real sample for a segment comparison and a small one for a claim about an industry. It is smaller than the 52,786 send segment because multichannel sequences run on narrower, better qualified lists, which is itself a confound: list quality cannot be separated from channel structure here.
Why publish a number that flatters you and then explain why it flatters you?
Because a rate whose denominator is hidden is decoration rather than evidence, and because this site's whole claim is that it knows how to sequence channels. If the arithmetic does not survive examination, there is no site.
Do you know which channel the positive replies arrived on?
Not split cleanly in this snapshot, and that is the field that would let us decompose the gap. It is named on the page as an open item rather than smoothed over, and recovering it would make a controlled split cheaper to run as well as strengthening the analysis that already exists.
MultichannelPros is part of the Outbound Pros group, operated by Jānis Plūme.
Last updated: 2026-08-06