Operated function · abandoned purchase
The cart that reached payment and stopped, the quote that was opened four times and never signed, the application half-completed on a Tuesday — all the same failure, and in most businesses all of it is answered with one reminder email sent an hour later.
The reporting says how many started and how many finished. It rarely says where in between the motion stopped, which means the gap is a single number covering half a dozen unrelated problems — a shipping cost that appeared late, a required field nobody can complete on a phone, a payment method the customer does not have, a form that lost its state on a network blip, a price that needed an approval the buyer did not have.
Each of those has a different fix and only one of them is addressed by a reminder. A person who abandoned because the delivery cost doubled at the last step does not need to be reminded; they need the cost shown earlier, and reminding them repeatedly is an irritation applied to a problem you created.
On the higher-value side the same failure wears a suit. A quote goes out, is opened several times, and goes quiet. Nobody records that it was opened, nobody notices the pattern, and the follow-up happens whenever the salesperson has a gap — which is systematically later for the smaller deals and for the reps with the fullest pipelines.
And the window is short. Intent decays fast: the person who stopped at payment this morning is reachable this morning and largely gone by Thursday. Most follow-up operates on a cadence designed around when a marketing tool sends, rather than around when a human being is still in the moment.
Meanwhile the one channel used for all of it is email, which is the channel with the worst reach to a person who is mid-purchase on a phone.
Abandonment is measured as a single gap between started and finished, so the causes cannot be separated — and the majority are structural failures that follow-up does not address at all.
Splitting the gap by step is the first move and it is mostly free, because the step data already exists in the surface. Once it is split, a large part of the loss usually concentrates in one or two steps, and those become an engineering fix rather than a messaging one — permanent, applying to everybody, and not requiring anybody to be contacted.
What remains after the structural fixes is the genuinely recoverable group: people who were interrupted, who wanted to check something, who needed a second person to approve. That group is small relative to the raw abandonment number and it converts far better, because the intent was real and nothing about the product stopped them.
For that group, timing dominates everything else. Reaching somebody while the intent is still live is worth more than any amount of message optimisation, and it argues for a channel that reaches a phone rather than one that reaches an inbox they will open on Thursday.
And the higher-value version — the quiet quote — is the same operation with a longer clock and a person on the end of it. Recording that a document was opened, noticing when opening stops, and putting that in front of the owner while it still means something is the entire mechanism.
Where purchases actually stop — measured by drop-off by step, against a baseline that had only a single start-to-finish figure.
Loss attributable to a structural cause — measured by abandonment concentrated at a specific step, before and after that step is changed.
Recovery within the genuinely recoverable group — measured by completions from the interruption and approval segments specifically, not blended with structural loss.
Time from stopping to being reached — measured by elapsed minutes to first contact, and the share reached inside your own defined window.
Quotes that go quiet without follow-up — measured by documents opened and then untouched past your threshold, surfaced rather than discovered at forecast review.
Follow-up evenness across deal size — measured by contact rate by value band, which is where the smallest deals are usually invisible.
that following up recovers structural loss. If people leave because a cost appears at the last step, contacting them is the wrong instrument and doing it repeatedly is worse than doing nothing. Most of the value on this page is the diagnosis, and a meaningful share of the fix belongs to your own surface rather than to any operated contact.
Step and order state come from the checkout, quoting or application surface you already operate. Nothing is rebuilt, no purchase flow is replaced, and the structural fixes this identifies are made in your surface by your team — which is the point, because a fix that lives in a vendor’s layer disappears when the vendor does.
For quotes and proposals, open and view state comes from the document or CRM system already sending them. Where that system does not record opening, the limitation is reported rather than substituted with an inference.
Contact goes out on channels you own, from your domain and your number, so a person mid-purchase receives a message from the business they were buying from.
The person who abandoned may never have completed a purchase, which means the basis for contacting them is narrower than for a customer and is governed by what they consented to when they entered their details. That consent is read from your system of record, never inferred from the fact that they typed an address into a form.
Frequency is capped by your rule and the default position is once. A second message to somebody whose blocker was structural is an irritation applied to a problem you have not fixed, and a third is a complaint.
Every contact carries a receipt naming the step they stopped at, the classified cause, the message and the outcome, so a complaint is answered from a record.
Operational access is not permission to train. Cart, quote and buyer data does not become material improving anything serving another organisation, including one selling the same category.
The consent question is the sharp one, because the subject is a prospect rather than a customer and the rules differ by jurisdiction and by how the details were captured. That is settled at scoping, in writing, before any message is designed.
The team that owns the purchase surface should be at the table from the first conversation, because most of what the diagnosis produces is work for them. A follow-up operation that ships while the structural cause stays unfixed will underperform and will be blamed for it.
Where an obligation attaches through a jurisdiction or a capture source, it is marked applicability-gated rather than presented as standing.
One purchase surface over one trailing period — read-only, nothing sent and no flow changed — with abandonment split by step and classified by cause.
The first phase produces the step breakdown, and in most businesses it reallocates the problem immediately. Where a single step holds a disproportionate share of the loss, the correct next action is an engineering change in your own surface, not a messaging programme — and it will help everybody rather than only the people who can be contacted.
That is a legitimate stopping point and a common one. A business that fixes its last-step cost disclosure and stops there has taken the larger win and bought no ongoing operation.
Where a genuinely recoverable segment remains after the structural work, follow-up starts there: one segment, one message, inside your window, capped at a single contact until there is evidence for more.
Almost everybody does, and the question worth answering is what share of your abandonment those emails can possibly address. If most of your loss is at a step where a cost or a requirement appears late, the email is being sent to people with an unresolved structural blocker — it will underperform for a reason that has nothing to do with the message. The step split settles that in a week and changes nothing while it runs.
Then the relevant object is the quote rather than the cart, and it is the same failure with a longer clock: a document opened several times and then quiet, with follow-up happening whenever the owner has a gap — systematically latest on the smallest deals. If your CRM cannot currently show you which open quotes were viewed and then went silent past a threshold, that is the same blind spot in a different shape.
It is, past the first message, and it is why the default here is one contact and why most of the population is not contacted at all. The larger part of the value is the diagnosis, which requires contacting nobody. If your read of your own market is that no outbound follow-up is appropriate, the step split is still worth having and the structural fixes it produces are the ones that scale anyway.
Then you have the first half and the useful question is whether anything acts on it. Step drop-off in an analytics tool is a chart somebody looks at monthly; the gap this addresses is that the cause is not classified and nothing routes the recoverable cases to anyone while the intent is still live. If your team is already fixing the top drop-off step each cycle and following up inside the window, you do not need this.