“My high-ticket offer stopped converting” — how to find what actually changed

Four ratios sit between your traffic and a signed client: enquiry rate, qualified share, calls actually held, and close rate. A 20% fall in each is a 59% fall in sales (0.8 × 0.8 × 0.8 × 0.8 = 0.41), which looks identical on the bank statement to one thing breaking outright. Pull all four before you touch the offer.

• Pull four numbers, not one. Enquiry rate per visit, qualified share of enquiries, share of qualified leads who hold a call, close rate on held calls.
• Compare the same four to a period when it worked — same length, same season, not “last month vs this month”.
• Drift and breakage produce the same top-line. Four ratios each down 20% and one ratio down 59% both leave you 41% of the sales.
• The offer is checked last, behind traffic mix, follow-up speed and deliverability, because that is the honest order of frequency.
• Cost to run it yourself: one spreadsheet, four columns, an afternoon — if your CRM timestamps contact attempts. If it does not, that is the first repair.

The Four-Ratio Split: why “everything is slightly worse” looks like “one thing broke”

The Four-Ratio Split is the diagnostic: express last quarter as four multiplied ratios, express the good quarter the same way, and compare the ratios rather than the sales count. It matters because sales are a product, not a sum. Here it is on 4,000 monthly visits, with every ratio dropping the same modest 20%.

Now the same fall with three ratios untouched: hold enquiry rate, qualified share and call volume exactly where they were, and drop close rate alone from 20% to 8.2%. You also land on 2.46 sales. Two businesses with identical revenue charts need opposite fixes, and the revenue chart cannot tell them apart. The decision rule we use on the split: if all four ratios are down and none is down more than about 25%, treat it as drift and repair the cheapest ratio first; if one ratio is down 40% or more while the others sit within 10% of their old values, it is a break — fix that one and change nothing else, or you will never know what worked. The forward version of this arithmetic, where the same compounding works in your favour, is set out in the compounding conversion maths behind small stage gains.

Is it a real drop, or just a quiet quarter?

Before diagnosing, check the drop is bigger than the noise, because at high-ticket volumes it often is not. If you average six sales a month and absolutely nothing has changed, a Poisson distribution with a mean of 6 puts three sales or fewer at 15.1% — roughly one month in seven. Two of those in a row is unremarkable. At six sales a month, a single bad month is not evidence of anything; two quiet weeks is noise, six is a trend. The practical rule: do not diagnose on fewer than about 25 held calls per period, and compare rolling 90-day windows rather than calendar months, which vary in working days and in whether they contain a public holiday. If the drop survives that test, it is real. The same problem in its close-rate-only form, including the confidence-interval method for small deal counts, is worked through on why a sales close rate is low and the test for each cause.

Which symptom is which? The one metric that separates them

Each symptom below is distinguished by exactly one metric. Pull that metric, and four indistinguishable stories become one.

Did my traffic mix change without me changing anything?

This is the most common answer and the least intuitive one, because you changed nothing. Two mechanisms do it on their own. First, auction price: Meta reported that average price per ad increased by 12% year-over-year in the quarter ended 30 June 2026, with ad impressions up 14% — a flat budget in a rising auction simply buys you a different, usually smaller, slice of the same audience. Second, matching: Google’s own documentation states that broad match may also take into account the user’s recent search activities, the content of the landing pages and assets, and other keywords in an ad group when deciding which searches to serve. Change a landing page and the query mix moves underneath you. A traffic-mix change is invisible in your ad account’s headline metrics and obvious in a qualified-share-by-source table. Build that table before you conclude the market has softened.

Is it lead quality, or is my follow-up not landing?

These are separate ratios and they need separate tests. Deliverability first, because it is binary and cheap to check: Google’s sender guidelines require bulk senders to keep spam rates reported in Postmaster Tools below 0.3%, and define bulk as more than 5,000 messages a day to Gmail accounts. A nurture sequence that quietly moved to spam produces exactly the symptoms of a dead offer: same enquiries, no replies, no calls. Then speed. In our own client work, moving speed to lead from hours to minutes is the single lever that moves contact rate most, and doubling contact rate roughly doubles the calls held. The honest wrinkle: those two do not multiply. Faster first contact is a large part of how contact rate improves in the first place, so stacking “3× from speed” on “2× from contact rate” double-counts the same mechanism. Treat them as one lever measured two ways. The mechanics of that lever are set out on how to increase speed-to-lead conversion rate.

Why the offer is the last thing you should change

The objection to this page is fair: “you are going to tell me to change my offer, and the offer is fine.” Usually it is. In our own experience of being handed underperforming accounts, offer fatigue — the same promise to the same audience for long enough that the audience has heard it — ranks behind traffic mix, follow-up speed and deliverability in frequency, and it is the most expensive of the four to test, because a rewritten offer needs a fresh cohort of leads before it means anything. That ranking is ours from the accounts we have run, not a study. Rewrite the offer only after the other three ratios have been measured and cleared, because an offer change re-bases every ratio at once and destroys your ability to attribute the recovery. One genuine exception: if qualified share has held up, calls are being held, and the loss reasons on held calls have shifted from “wrong time” to “this is not what I thought I was booking”, the promise and the call have drifted apart, and that is an offer problem.

What running this diagnostic actually costs

Honestly: a spreadsheet and an afternoon, if your data is instrumented. Six columns — period, source, enquiries, qualified, calls held, sales — one row per source per month, twelve months back. Compute the four ratios per row, then chart each ratio, not the sales count. Three or four hours the first time, twenty minutes a month afterwards.

What makes it expensive is the data, not the analysis. The diagnostic needs a timestamp on every enquiry and every first contact attempt, a qualification decision recorded against the lead rather than remembered, and held calls distinguished from booked calls. Most CRMs will do all three and most are not configured to. If yours is not, the first repair is instrumentation, and the honest threshold is this: below roughly 30 enquiries a month, do this by hand in a spreadsheet and do not buy anything — the sample is too small to support a decision either way. Above a few hundred enquiries a month, the manual version breaks down, because the ratios have to be cut by source and by week to be readable, and that is a data job rather than an afternoon. What that looks like when the follow-up itself is run for you is described on lead generation for high-ticket service businesses; when the diagnosis lands on the last ratio, the levers are on how to increase sales close rate.

Frequently asked questions

Why did my offer stop working when nothing changed?

Usually because something outside your control changed instead. Media auctions reprice continuously — Meta’s own results release reported average price per ad up 12% year-over-year for the quarter ended 30 June 2026 — and match types, audiences and inboxes all shift underneath a campaign that has not been edited. Compare the four ratios against a period when it worked; the one that moved names the cause.

How do I tell offer fatigue from a lead quality problem?

Qualified share separates them. If the share of enquiries that qualify has fallen, the problem is upstream of the offer: your traffic mix changed and you are talking to different people. If qualified share is unchanged and close rate on held calls has fallen, the same people are hearing the same promise and buying less of it, which is the fatigue case.

How far back should I compare?

Use rolling 90-day windows and compare like season with like season. Calendar months differ in working days and in holiday coverage, which alone can move held-call volume by a fifth. If you have fewer than about 25 held calls in a window, widen the window rather than accept a noisier number.

My sales are down 30% — how much of that could just be luck?

More than most operators expect at high-ticket volumes. At an average of six sales a month, three or fewer occurs 15.1% of the time from ordinary variation alone under a Poisson distribution with mean 6. That is one month in seven with nothing wrong. Judge the drop across a quarter, not a month.

Why are my emails and texts getting no replies at all?

Check deliverability before copy. Google’s email sender guidelines require senders of more than 5,000 messages a day to Gmail accounts to authenticate with SPF, DKIM and DMARC and to keep spam rates reported in Postmaster Tools below 0.3%. A sequence that has crossed that line still reports as sent and lands nowhere.