Business Club Varna2 years, 140+ members, zero ad spend
19 families signed in one fall season, more than double the year before on a smaller budget
Beacon Hill Admissions Advisors in Boston took 105 enquiries between September and November 2024 on $6,900 of ads, held 17 first consultations, and signed 9 families. A year later, on $5,550 of ads over the same 3 months, it signed 19. The team had long explained the quiet enquiries as families shopping around and not serious. What we nearly did about that was cut the enquiry form to a name and an email, and that version was built and scheduled for a Monday when the mail logs came back on the Friday before.
About the client
Beacon Hill Admissions Advisors is a 6-person college-admissions and test-prep consultancy on Charles Street in Boston, founded by Erin Sullivan in 2016, selling SAT and ACT prep from $1,800 and application coaching from $2,800 to $6,500 to families in Brookline, Newton, Wellesley and the city, at an average of about $4,200 per signed family. Google Search and Meta had run since 2022 at $2,300 a month in ad spend (our fee is not included in any figure here), with the fall window the larger of 2 busy seasons. A January 2024 move from GoDaddy email to Google Workspace broke nothing visibly; that October 2 parents mentioned a reply found in spam or never received at all, and Erin suspected the email for about 18 months without anyone checking, because HubSpot said every message was sent.
College-admissions coaching and SAT/ACT prep
Boston, Massachusetts
Six people
2016
Our analysis of the situation
The ads were the first thing struck off: $65.71 per enquiry, a fair price for education, a clean search-terms report, and Google and Meta enquiries booking consultations at 17 and 15 percent, too close for either channel to be sending junk. Response speed was the second: HubSpot showed the automated reply inside 2 minutes and the advisor's follow-up at a median of 5 hours, but HubSpot records the handover to the mail server, not the arrival.
Google Workspace's Email Log Search and a Postmaster Tools account nobody had set up showed about 9 percent of follow-ups bouncing, SPF failing on nearly every message, and 12 of 20 seed messages in spam, so roughly 22 percent of the sequence never reached an inbox. The SPF record still named GoDaddy's servers, there was no DKIM key and no DMARC policy.
With 9 of 17 consultations signing, a healthy 53 percent, the leak was between enquiry and consultation, about one in 6. We chose the enquiry-to-consultation rate and cost per signed family, and the strategy was to get the email to arrive and then give a family comparing 3 consultancies a reason to answer it: a named person, a price on the page and a clear next step.
Supporting research
WordStream's Google Ads benchmarks put education and instruction at about $77 a lead [1], so Beacon Hill's $65.71 was buying enquiries at a fair price. The Oldroyd and InsideSales study in Harvard Business Review found that contacting a web lead within an hour makes qualifying it nearly 7 times as likely [2], the standard the HubSpot timeline was checked against. EAB's 2019 case study of an adult degree-completion program found enrolment stalled for the same reason, the institution's own follow-up email was not arriving [3].
Our strategy to solve the problem
We fixed the DNS before we touched anything a family could see. The SPF record was rewritten to name Google's servers, a 2048-bit DKIM key was generated in the Google Workspace admin console and published, and a DMARC record went up in monitoring mode, moved to quarantine 6 weeks later. We paused the HubSpot sequences and spent 2 weeks re-warming the domain with ordinary one-to-one mail from the advisors before switching the automation back on. By mid-July, Postmaster Tools showed delivery at about 97 percent, up from about 78 percent. Nothing else we did would have worked without this, and on its own it did very little. The consultation rate in July and August moved a few points and no more, because a follow-up that arrives is still a follow-up from info@ with nothing in it a family had asked for.
This was also the fortnight of the 2-field form. Before the log search came back, our working theory was friction: the enquiry form asked for a name, an email, a phone number and a message, and we had rebuilt it in Fluent Forms as name and email only, with a Monday go-live date. On the Friday we went back into HubSpot and sorted the fall 2024 enquiries by how much they had typed. The families who skipped the phone number and wrote one line had booked a consultation 6 percent of the time. The ones who wrote a paragraph booked at 24 percent. The short form would have produced more of the first kind. We deleted it and built the opposite.
Beacon Hill's part in these weeks was access: the GoDaddy account, where the domain was still registered, the Google Workspace super-admin login, and HubSpot admin. No agency can fix deliverability without the first 2, and we have been asked to try.
The new form is 5 questions in Fluent Forms Pro with conditional logic, and it takes under 2 minutes. Graduation year. Whether the family wants test prep, application coaching or both. The kind of schools they have in mind, as a range rather than a list. A budget band. Whether a parent will be on the first call. On submission the family gets back a checklist of what to bring to that call, and a senior's family also gets a deadline calendar for the coming application round, which settled the "what do you need from us" exchange that used to take 3 emails.
The answers decide who replies. A 2026 senior with engineering in the school range is shown the name, photo and HubSpot Meetings link of Tom Feeney, who handles STEM applicants, and the first personal reply comes from tom@ rather than the shared inbox. 2 of the 3 advisors agreed to be named and photographed and to answer within a day. The third said no, and we did not push; her enquiries route to Erin.
The answers also branch the HubSpot sequences. A senior applying this cycle gets 5 short emails over 12 days with a booking link in each, because the deadlines are weeks away. A sophomore's family gets one email a month through the research phase. The old sequence had treated everyone the same and stopped writing after the third unanswered message, which for a family 18 months from a decision is about the point they start paying attention.
Completion of the form fell from 91 percent to 74 percent once it had 5 questions. Erin asked about that in the July review, and we asked her to wait for September.
The prices went on the site on 25 August, and that was the hardest sign-off of the engagement. The old page said "contact us to discuss your family's needs". The new one shows the package prices as ranges by grade, the test-prep block price, and a plain line that no admission outcome is promised. 2 better-known consultancies in Newton and Lexington do not publish theirs, and Erin took until 18 August to agree, on the argument that a family who cannot see a price assumes the highest one and moves on.
The same week we cut the budget. Meta's broad-interest prospecting campaign, $450 a month, had produced 22 enquiries in fall 2024 and one consultation, so it went, and the retargeting campaign stayed. That took the spend from $2,300 to $1,850 a month, and the enquiry count for the fall window fell with it, from 105 to 82. Nothing in the result comes from more traffic.
In October a referral offer went live: a $150 credit for both families, with a referral code field on the form that writes to a HubSpot contact property. It ran for 7 weeks of the window and 3 of the 19 signed families arrived with a code.
Somewhere in October, Tom mentioned a father from Quincy who had booked a call through the new page and did not want to talk about packages. His nephew was a sophomore at Northeastern, and he wanted to know whether the co-op program was as good as the nephew said, and whether Tom happened to know anyone in the housing office there. Tom did not. They talked for 40 minutes anyway. The daughter's family signed a junior-year package in November. HubSpot shows every email we sent them as delivered and unopened.
Before and after
Same fall window, one year apart.
| Metric | Before (1 Sep to 30 Nov 2024) | After (1 Sep to 30 Nov 2025) | Change |
|---|---|---|---|
| Monthly ad spend (Google Search + Meta) | $2,300 | $1,850 | −20% |
| Enquiries | 105 | 82 | −22% |
| Cost per enquiry | $65.71 | $67.68 | +3% |
| First consultations held | 17 | 36 | +112% |
| Enquiry to consultation | 16.2% | 43.9% | +27.7 points |
| Families signed | 9 | 19 | +111% |
| Consultation to signed | 53% | 53% | flat |
| Cost per signed family | $767 | $292 | −62% |
| Follow-up email delivered | ~78% | ~97% | +19 points |
| Form completion (of those who started it) | 91% (4 fields) | 74% (5 questions) | −17 points |
The consultation-to-signed rate held at 53 percent, and we left it alone on purpose. The advisors were already signing about half the families they sat down with, and the whole change is in how many families sat down. Enquiries fell by a fifth and cost per enquiry went up 3 percent, both because the broad Meta campaign was cut, and neither number cost the firm a family.
Before, $6,900 of fall advertising bought 9 signed families, about $37,800 of packages, and 3 advisors spent October chasing 88 people who had, as far as anyone knew, decided not to answer. Now $5,550 buys 19, about $79,800 of packages, and the follow-up work is answering questions from families who already know the price and who they will be talking to.
Summary
5 of every 6 families who enquired never reached a first consultation, and for 18 months the firm had blamed the families.
The HubSpot timeline showed replies going out in minutes, which cleared response speed and pointed at delivery, and Google's mail logs showed why: a January 2024 move from GoDaddy to Google Workspace had left the domain's authentication records pointing at the old host, so about a fifth of every follow-up sequence never reached an inbox.
We rewrote the DNS records, replaced a 4-field form with 5 screening questions that route each family to a named advisor, branched the follow-up by graduation year, published the prices, and cut the one Meta campaign whose enquiries never booked.
Fall consultations went from 17 to 36 and signed families from 9 to 19, at $292 per signed family instead of $767.
References
- WordStream, Google Ads benchmarks by industry. wordstream.com
- James B. Oldroyd, Kristina McElheran and David Elkington, "The Short Life of Online Sales Leads", Harvard Business Review (March 2011). hbr.org
- EAB, "The graduate marketing strategy that tripled leads" (April 2019). eab.com
Final word from the client
The mail was broken for a year and a half and the fix was one afternoon of DNS. I'd suspected it since that mother in Newton told me, and I still had my advisors rewriting their follow-up emails to sound warmer. So, relieved. And a little annoyed at myself, honestly, because I could have checked.