Volume went out, almost nothing came back — and a fifth of what came back was hostile.
Open tracking was off (plain-text sends), so replies are the only signal we have.
| Campaign | Copy angle | Unique sent | Replies | Positive | Not interested | Do not contact |
|---|---|---|---|---|---|---|
| ESOS CEO/MD — A | "Your ESOS action plan lists solar…" | 588 | 11 | 1 ✓ | 1 | 0 |
| ESOS CEO/MD — B | "Your ESOS shows £X savings…" | 300 | 15 | 0 | 3 | 5 |
| Eng-Mfg · One Pager | Renewal Trap + LSN case | 361 | 9 | 0 | 1 | 3 |
| Eng-Mfg · Audit | Peer Who Moved / Reframe | 386 | 7 | 0 | 3 | 0 |
| Role-based | info@ / sales@ routing ask | 290 | 8 | 0 | 3 | 3 |
| Simple copy | "A cost you can't negotiate" (live 2 days) | 231 | 6 | 0 | 1 | 1 |
Note: ESOS-B — the most personalized copy, quoting the company's own £ savings figure at the CEO — produced the most replies and the worst quality: 5 of the 12 angry removals, zero positives. Personalization amplified the spam signal instead of relevance.
Every substantive reply, grouped by what it tells us.
"Jo passed on your email. I'd be happy to have a call… suggest a few dates and I'll get a video call set up."
"We still have a year left on contract. So not interested time being."
"Signed 5/6 weeks ago with British Gas a two year contract."
"Not something we're looking to act on at this point in time; I'll retain your details."
"Stop emailing me please."
"Please stop spamming me." · "Do not email us again." · "Just leave please." · "Please unsubscribe."
"Coppermill has permanently closed…" · "The company is no longer on a site" · "I am no longer working full time as Managing Director"
"No thanks" (SPS CEO) · "This is not of interest" (Caddick MD) · "We're not interested" (Acrefine, AMC)
UK businesses are carpet-bombed by broker/TPI email. Every opener here — "contract comes up for renewal… rates climbed… network charge up 62%… fix your price" — is the exact linguistic fingerprint of that category. Wattstor's genuinely different offer (funded solar + battery, zero capex) sits in paragraph 2–3, phrased in the same cadence. Readers classify and delete in two seconds; the differentiation never gets read.
Initially flagged the generic lookalike domains as a trust risk. Discounted on evidence: 3M+ sends across the agency show domain-name semantics don't move performance, and zero replies in this dataset reacted to the sender domain. What does remain (now under H6): the same prospect receiving the same email from two different "Steve Bingham" addresses because they sat in two campaigns — a cross-campaign dedupe issue, regardless of what the domains are called.
ESOS filers are by definition large enterprises. The list sampled: Tesco, IKEA, AstraZeneca, Aviva, Accenture, Pearson, Jet2, Wetherspoon, the Royal Albert Hall, Guide Dogs, private schools, care groups, a football investment firm. These orgs have energy procurement teams and existing PPAs; cold email to the group CEO gets routed (Travelodge — the one win) or binned (everyone else). Many are physically wrong fits: leased premises, offices, charities — no owned roof, no industrial load.
Onsite generation is an ops-heavy, multi-stakeholder decision and nothing in the copy reduces that weight. The replies show it as inertia ("not a priority", "just renewed") rather than stated fear of project work. The twist: the headline benefit — "fixed for 25 years" — actually adds weight. To an owner-manager that reads as lock-in risk, not security. The offer asks for the psychology of a marriage with the CTA of a flyer ("want the one-pager?"). People feel the gap and take the zero-effort path: ignore.
LSN Diffusion is named in every manufacturing sequence but never with a result — no % saved, no £ figure, no payback, no quote. For an offer that sounds too good to be true (free kit, you only save), missing proof confirms the scam prior. The instinct to lead with real savings is correct: it's the single highest-leverage missing asset.
Duplicate leads across campaigns (Branston, AstraZeneca in two each), dead companies, retired contacts, generic inboxes feeding ticket systems. The role-based info@/sales@ campaign produced literally zero value — gatekeepers, auto-responders and anger.
All 18 emails (every step and A/B variant) pulled from SmartLead and read in full. Full text in sequences_fulltext.md.
It's designed to look like internal mail. The trick works — and then the reader opens a cold pitch and feels played. Deception in the subject line poisons everything after it, and it's the kind of pattern that turns a delete into a "stop spamming me." The other subjects ("{{first_name}}, renewal", "a cost you can't negotiate") are pure broker-speak. None of the 6 campaigns has an honest, specific subject.
Market observation → LSN / data point → one-line "Want it?" CTA. Every single email ends in a yes/no interest question; not one delivers value directly (the one-pager is always gated behind a reply, never just given). Once a reader has seen one, they've seen all six campaigns — which matters because leads overlap across campaigns.
Same body, different CTA ("free audit" vs "one-pager") and different sender domain. With duplicated leads (Branston, AstraZeneca confirmed in two campaigns each), the same person can receive the identical email twice from two different "Steve Binghams" — the strongest possible spam signal.
"I have put a free read of your site's numbers together. Want it?" — nothing was put together. If anyone says yes, there's nothing to send, and "I prepared a report for you" is a known spam trope. Same family: the CTA changes mid-thread (step 1 offers an audit, step 2 offers a one-pager), which reads as template, not conversation.
ESOS-A inserts {{Action Plan}} — a free-text field from a government dataset — directly into a sentence, which produces broken grammar for many rows. ESOS-B's £{{GPB}} token is a typo waiting to send "£{{GPB}}" literally. The merges that did fire correctly (Sport Republic's "£430,473") landed on the wrong kind of company anyway — precision aimed at the wrong target.
On every step 1. It explicitly invites the negative reply, and 12 people accepted the invitation. A "stop" reply trains spam filters and burns the domain for everyone else on it. Soft opt-out language belongs in the breakup email, not the opener.
Audit step 2A (the qualification email): "If you own your site, have roof or land spare, and electricity runs into six figures, the numbers usually work… I'd say so rather than waste your time." Honest, self-disqualifying, specific — this thinking should open a sequence, not hide in step 2. The standing-charge education email (simple-copy/one-pager step 2) teaches something real and checkable on the reader's own bill. Both survive into the new copy drafts in section 07.
Three angles built from the learnings. Bracketed placeholders need real numbers from Wattstor before launch — the proof is the angle, do not launch without them.
Hi {{first_name}},
Last year we built a solar-and-battery system at LSN Diffusion's plant in South Wales. We paid for all of it. They now buy their power [X]% below the grid rate, saved around £[X] in the first year, and their price is capped — it cannot rise — for the life of the deal.
Their side of the work was [two meetings and site access]. No capex, no downtime, and if the system underperforms, that's our loss, not theirs.
If {{company_name}} owns its site and spends six figures a year on electricity, the same maths likely works. Want me to run your number?
{{first_name}} — the reason this isn't too good to be true: we make our money selling you the power the system generates, at a rate set below the grid. If it generates less than promised, we earn less. You can't lose the bet; we can.
The usual worry is the hassle — planning, the DNO, install. We handle all of it; LSN's team spent roughly [N hours] on the whole project.
Worth a look at your site's number?
{{first_name}} — last one from me. If energy sits with someone else at {{company_name}}, a name is all I need. If the timing's wrong, tell me and I'll leave it.
Either way: when your next renewal letter lands, you'll know there was another option on the table.
Hi {{first_name}},
Everyone assumes putting solar and a battery on site is a year of their life: planning, the DNO queue, contractors all over the yard. That's why most plans for it die in a drawer.
Here's what it actually took from LSN Diffusion's team in South Wales: [two meetings, a site walk, and sign-off]. We did the rest — design, planning, grid paperwork, build — and we paid for all of it. Production didn't stop for a day.
They now pay [X]% less than grid for power, capped so it can't climb.
Want the one-page version of what your site's involvement would look like?
{{first_name}} — quick filter so I don't waste your time: this works on sites you own (or hold a long lease on), with roof or land going spare, spending six figures a year on power. If that's not {{company_name}}, say so and I'm gone.
If it is — the site survey is free, takes [an hour], and you get the numbers whether or not you do anything with them.
{{first_name}} — if this is more the MD's call or the FD's, point me their way and I'll stop landing in your inbox. Otherwise I'll close the file. Thanks for bearing with me.
Hi {{first_name}},
You get ten emails a week about switching energy supplier. This isn't one — I'm not a broker and there's nothing to renew.
We build solar and batteries on factory sites at our own cost, then sell the site its own power below the grid rate. LSN Diffusion in South Wales did it: £[X] saved in year one, price capped for good, and they paid nothing to build it.
It only fits sites you own with real power spend — if that's {{company_name}}, I'll send the one-pager. If not, ignore me and I won't follow up twice.