Follow-Up Cadence — The Published Numbers Are Right and Still Wrong for You

The follow-up literature has converged on a number. Eight to twelve touchpoints over fourteen to twenty-one days, across phone, email and LinkedIn. Longer sequences of twelve to sixteen touches report roughly double the contact rate of short ones. Multi-channel sequences report dramatically higher reply rates than email-only.

Alongside this sits the most-quoted statistic in sales: around 80% of sales require five or more follow-ups, and nearly half of reps stop after one attempt.

All of this is well-evidenced, and copying it directly into a local-market outbound motion will still give you the wrong cadence. Not because the research is bad, but because the recommended shape scales with deal size, and the published guidance is written around deals considerably larger than a €400-a-month retainer with a hairdresser.

The Variable Everyone Reports and Nobody Applies

The same sources that give the 8–12 headline also break it down, and the breakdown is where the useful information is.

Low-value deals: roughly 6–8 touches over two to three weeks. This is the band most local and SMB outbound actually lives in.

Mid-market, in the €10k–50k range: 8–12 touches over four to six weeks. This is the band the headline number describes.

Enterprise, above €50k: 12–18 touches over six to twelve weeks. This is where the impressive long-sequence statistics come from.

The 12–16 touch sequences reporting double the contact rate are measuring enterprise motions. Running sixteen touches at a two-person plumbing firm over a €300 offer does not double your contact rate. It makes you the company that will not stop calling, and in a local market where businesses talk to each other, that has a cost the datasets do not measure.

The rule that survives contact: cadence length should scale with the size of the decision you are asking for. A small ask that needs sixteen touches is not under-followed-up. It is mis-targeted.

What the Channel Data Does and Does Not Say

Multi-channel outperforms single-channel by a wide margin — reply rates reported at nearly three times email-only. This is one of the more robust findings in the literature, and it is frequently misapplied in local markets.

The channel mix in the research is not available to you. Those numbers lean heavily on LinkedIn, where a profile visit plus a message reportedly pulls a higher reply rate than any email follow-up. A self-employed roofer does not have a meaningfully active LinkedIn presence. The finding is real; the channel is not there.

In local markets the third channel is usually physical or telephonic. Phone, email, and then either a second phone attempt at a different hour or something genuinely local. The lesson from the multi-channel data is not "use LinkedIn." It is that the same message through a second channel substantially outperforms repeating the first — and you should pick your second and third channels from what your market actually uses.

Cold email carries a legal weight that cold calling does not, and it differs by country. In Germany the email path is materially tighter than the phone path. A cadence blueprint that alternates email and call touches is making a compliance decision, silently, in a sequencing tool. That decision should be explicit and per-country.

Spacing beats volume. Touches two to three business days apart outperform the same number spread thinner. This finding does transfer to local markets, and it is free to implement.

Where Cadences Break in Practice

Nobody defines what ends a sequence. Most teams specify the number of touches and never specify the exit. A prospect who replies "not now" needs to leave the sequence and enter a dated callback. If the tool keeps sending, you have automated the destruction of a warm lead — and this is the single most common cadence defect.

The last touch is treated as an ending rather than a handoff. Sequence completion without a reply is not "lost." It is a lead whose timing was wrong, and the right destination is a long-cycle list revisited in a quarter. Teams that delete at sequence end throw away the majority of their eventual pipeline.

Every touch says the same thing in different words. Six touches that each say "just following up" are one touch repeated six times, and the prospect experiences it as pressure without information. Each touch needs its own reason to exist — a new observation, a re-run audit, a relevant change. This is where the audit-based opener keeps paying: it regenerates.

Cadences run on the sender's calendar. A sequence that fires touch four on the Friday before Christmas fires it because a counter incremented. Business-day awareness and local holidays are unglamorous and prevent a specific, recurring category of self-harm.

Reply detection is weaker than teams assume. Without open and click tracking — which many EU-facing setups deliberately avoid — reply detection is threading-based, and it misses replies from a different address, forwarded threads, and phone callbacks. A prospect who called back and got sequenced anyway is a real and avoidable failure. Any human contact through any channel must be able to halt the sequence.

Designing One That Fits

Start from the ask, not from a template. Write down what you are asking for and its value. That number sets the band: small asks get six to eight touches over two to three weeks. Then design within the band rather than copying a blueprint written for a different one.

Give every touch a distinct job, in writing. Touch one is the observation. Touch two is a different channel, same observation. Touch three introduces a new finding. Touch four is the direct ask. If you cannot write the job, delete the touch — a shorter sequence where every touch has a reason outperforms a longer one padded with check-ins.

Define every exit condition explicitly. Reply, any inbound contact on any channel, an explicit no, a dated constraint, and sequence completion each need a defined destination. Sequence completion should mean "long-cycle list, revisit in a quarter," not "delete."

Enforce suppression at send time, not only at enrolment. Someone who opts out mid-sequence must stop receiving the remaining touches. Checking only at enrolment is how opt-outs get ignored for another eleven days, which is worse than never having had a suppression list.

Measure per-touch, not per-sequence. Track which touch number produces replies. Most teams find a cliff — a point after which additional touches produce almost nothing but do produce complaints. That cliff is your real cadence length, and it is knowable from a few hundred sequences of your own data.

Re-measure when the offer changes. Cadence length is a property of the offer and the market, not of your team. A new offer at a different price point needs its own band, and inheriting the old sequence is how a well-tuned cadence quietly becomes wrong.

The statistic about half of reps stopping after one attempt is quoted to argue for persistence, and persistence is right. But the failure mode it describes and the failure mode of a sixteen-touch sequence aimed at a €300 deal are both cadence-design failures — one from too little thought, one from copying the answer to a different question. The band comes from the size of your ask. Everything else is design within it.