Most agencies pick one channel and defend it like a religion. The email people think LinkedIn is a time sink for consultants who enjoy the sound of their own voice. The LinkedIn people think cold email is spam with extra steps. Both are half right. Both are also leaving money on the table, because the merchants you want to reach are not consistent about where they pay attention, and you do not get to choose for them.
The double-touch sequence is not multichannel for the sake of a slide. It is a division of labour. LinkedIn buys recognition. Email carries substance. Run them in the wrong order, or ask either one to do the other’s job, and the whole thing collapses into noise that costs twice as much to produce.
The problem with one channel is arithmetic, not taste
A pure LinkedIn campaign has a hard ceiling: you cannot message a person who has not accepted your connection request. Every profile that ignores you is not a soft no, it is a permanent no — you never got the chance to say anything at all. On a list of a few hundred ecommerce founders, that quietly removes a large slice of your addressable market before you have written a single word of copy.
A pure email campaign has the opposite failure. You can reach almost everyone, but you arrive as a stranger in an inbox that already receives a dozen agency pitches a week. The merchant’s first filter is not “is this relevant?” — it is “do I recognise this?” You lose on the second question before the first one is ever asked.
Put them together and the two failures cancel out. The LinkedIn touch does not need to convert; it needs to make your name and your face familiar. The email touch does not need to build rapport; it needs to deliver one specific, verifiable observation about their store and a reason to reply. Familiarity plus specificity is a much higher bar than most of your competitors are clearing.
Start from the store, not from the person
Standard LinkedIn prospecting starts with a title filter: founder, ecommerce manager, head of digital. That is how everyone builds a list, which is precisely why everyone ends up contacting the same few thousand people. Start from the other end. Build the list from the stores, then go and find the humans attached to them.
Working from a store database, you filter on things a job title can never tell you: platform, country, catalogue size, monthly organic traffic, detected apps, whether a Google Business profile exists and what its rating looks like. Out of 153,515 verified stores across 60 platforms, a serious ICP filter — Shopify stores in one country, catalogue between 80 and 400 products, organic traffic trending down over the last six months — usually lands somewhere between 150 and 600 domains. That is a week of real outreach, not a year of spray.
Then you go and find the person. The merchant name and a contact email are already in the record; the domain gives you the company page; from there LinkedIn gives you the founder or whoever runs acquisition. You are doing the reverse of what everybody else does, and the difference shows up in the first line of every message you send.
Touch one: LinkedIn, and do not pitch
The first touch has exactly one job: get accepted, and be remembered. That means no offer, no calendar link, no “I help ecommerce brands scale”. The connection notes that get accepted are short, specific, and ask for nothing at all.
The specificity comes from the store data you already hold. You are not writing “love what you’re doing”. You are writing something only somebody who actually looked would know: that they run a 300-product catalogue on a platform most of their competitors abandoned, that their organic curve turned in the autumn, that they carry three payment methods where their category usually carries five. One observation. No conclusion. No ask.
Before the request, spend thirty seconds warming the profile: view it, react to a recent post, leave one comment that is not flattery. This is unglamorous and it works, because the acceptance decision is made in roughly two seconds and prior exposure is one of the very few things that moves it. If you are running this at volume, batch the warming for a whole cohort on one day and send the requests the next.
Do not send a follow-up message the second they accept. That is the most common way agencies burn a warm connection: the acceptance is read as permission, the pitch arrives, and the relationship is over before it started. Let the acceptance sit.
Touch two: email, and bring the artifact
Two or three days after the connection lands, the email goes out. Now you have permission to be substantive, and the merchant has a face to attach to the name — they have seen your profile, possibly your comment, and they let you in. The email is no longer cold. It is the second contact from somebody they already accepted.
This is where the store record earns its keep. A useful email has four moves. One: the observation, stated as fact, with the number attached. Two: the implication, stated as a business consequence rather than a technical one. Three: the artifact — the thing you already made for them, attached or linked, not promised. Four: a small ask.
An illustrative shape, deliberately generic: “Your organic traffic has been flat for eleven months while your product count grew by roughly a third — you are adding catalogue faster than you are adding search surface. I pulled your top thirty organic keywords and mapped which category pages are missing. Two pages, attached. If the mapping is wrong I would genuinely like to know.” Nothing in that requires a call, a demo or a pitch. It requires a reply.
The artifact is the whole game. Anybody can write a clever email; almost nobody sends work. A one-page teardown built from real fields — traffic trend, organic keyword footprint, detected app stack, Google Business rating and review count — costs you an API call and a template. It costs your competitor an afternoon, which is exactly why they never send one.
The calendar
Keep it boring and finite. Day 0: profile view, one reaction, one comment. Day 1: connection request with the single observation. Day 3 or 4, for accepts only: email one, with the artifact. Day 8: email two, one paragraph, a second angle on the same store — usually the local reputation data if the first email was about traffic. Day 12: a two-line LinkedIn message referencing the email without reproaching them for the silence. Day 18: a one-line close-the-loop email that gives them an easy exit.
Six touches, eighteen days, then out. The people who never accepted the connection move to a pure email track with a longer artifact, because you have nothing left to lose with them. The people who accepted but never replied go to a quarterly re-touch, not to the bin — a merchant who ignored you in April because they were mid-migration is a completely different prospect in September.
What to measure, and in what order
Three numbers, and only three, at least at first. Connection acceptance rate tells you whether your profile and your observation are working. Reply rate on email one tells you whether the artifact is any good. Reply-to-call rate tells you whether your offer matches the segment you picked. If acceptance is healthy and replies are dead, your artifact is generic. If replies are healthy and calls are dead, you targeted stores that cannot afford you.
Resist the urge to optimise the subject line before you have fixed the segment. Most double-touch sequences that underperform are not copy problems; they are list problems wearing a copy costume. Narrow the filter until every single store on the list makes you think “I know exactly what I would do for them”, and the copy largely writes itself.
Where this breaks
It breaks when you automate the LinkedIn side. Connection notes generated at volume read like connection notes generated at volume, and the platform is unforgiving about it. The warming and the note are the two things to do by hand, or with a human firmly in the loop; everything upstream — list building, data pulls, artifact generation — is where automation belongs and pays.
It breaks when you run it against a segment too broad to share a problem. A double-touch sequence aimed at “ecommerce stores” is a worse version of a single-channel campaign, because you have doubled the effort per prospect without increasing the relevance of anything you say. Aimed at PrestaShop stores in one country with 500-plus products, declining organic positions and no review-response activity on their Google profile, it works, because every message can say the same true thing.
And it breaks when the second touch does not carry an artifact. The entire justification for the LinkedIn step is that it buys you the right to send something substantial. If email one is a pitch, you spent the recognition you built on nothing at all, and you would have been better off staying single-channel and cheap.