Product
Follow ups and booking
The difference between a follow up and spam is almost entirely about when it stops.
Quotes that went quiet
Following up is the highest return activity no one does
In almost every small organisation there is a set of people who asked about something, were sent a price, and were never contacted again. Not because anybody decided not to. Because the week got busy, and by the time someone thought about it the moment had gone. Consonas sends those follow ups on a schedule and stops the moment the person answers.
The money involved is larger than most owners expect, and the reason is that a meaningful proportion of those people did not go elsewhere on price. They went elsewhere because someone else rang them and you did not. Following up is consistently the highest return activity available to a small business and it is consistently the thing that does not happen.
So automation is obviously part of the answer. The difficulty is that most automated follow up is bad in ways that cost more than the follow up gains, and everyone reading this has been on the receiving end of it.
The four ways it goes wrong
It does not stop. You reply on Tuesday and on Thursday you receive a message wondering whether you saw the last one. That single failure tells the recipient precisely how much attention is being paid, and it is worse than not having followed up at all.
It arrives at the wrong time. Three in the morning, or Sunday afternoon, because the schedule counts elapsed hours instead of working ones.
It ignores what someone has agreed to. The list was built three weeks ago and the person withdrew last Tuesday, and the system checked the list rather than the person.
It runs away. Two rules trigger each other, or a sequence has eleven steps because a tool made eleven easy, and a customer receives something absurd. The organisation finds out from the customer.
Every one of those is a design decision rather than an accident, and each is avoidable.
Reply, then silence
Six behaviours, and what each one does
The middle column is the common behaviour rather than a bad tool.
| The behaviour | A typical sequencing tool | Here |
|---|---|---|
| Someone replies | The next step goes out anyway, unless the reply was detected in the same inbox the sequence sends from. | They are removed immediately, because replies and sending share the same record. |
| Someone withdraws consent mid sequence | Applied to the list, which was built before they withdrew. The next step still goes. | Checked again at every step for that individual, so the next step does not. |
| It is Friday at six | The step goes out, because the schedule counts elapsed hours. | Scheduled against working hours, so it goes on Monday morning. |
| A colleague is about to ring them | No way to know what they have already been sent. | Visible on the record: which sequence, which step, what has gone. |
| A customer complains about the emails | Whoever takes the call raises it with whoever set the sequence up. | They stop it themselves, on the record, in one action. |
| Two rules trigger each other | It runs away, and somebody finds out from a customer. | Sequences are bounded and rules cannot chain indefinitely by design. |
Stopping
A reply ends it, and the whole design follows from that
The moment someone answers, they are out. Not at the next scheduled check, not once an integration notices, immediately, because the messages and the replies live on the same record rather than in two systems that have to agree with each other.
This is the behaviour most worth getting right and the one most often got wrong, and the reason is architectural. Where a sequencing tool sends from one place and reply detection depends on watching an inbox somewhere else, there is always a window, and the window is where the damage happens.
Anybody with permission can also stop a sequence by hand at any point. If a customer telephones to complain about the emails, the person who takes that call can end it themselves instead of promising to speak to whoever set it up, which is the difference between a resolved complaint and one that recurs on Thursday.
Timing
Working hours, and an end
Steps are scheduled against working hours rather than elapsed ones. A follow up two days after a Friday enquiry arrives on Tuesday morning, not on Sunday, and a step that carries a delay lands at the hour your organisation opens rather than at three in the morning.
There is one exception and it is deliberate. A step with no delay on it goes the moment it is triggered, whatever the hour and whatever the day. Zero is not a schedule, it is a reaction: somebody enrolled a contact, or a form was filled in, and the first touch belongs to that moment. Holding a Saturday enquiry until Monday morning would be silence rather than courtesy. Every step you gave a delay to is a chase whose timing you chose, and those are the ones that must not land at the weekend.
Sequences are bounded. They have an end, and rules cannot chain into one another indefinitely. This is a deliberate constraint rather than a limitation: a rule that can run away will eventually run away, and it will do so at the worst possible moment, which is usually a bank holiday.
Three or four steps is the right length for almost everyone, and considerably shorter than most tools encourage. Somebody who has not replied to three messages is telling you something, and the fifth message does not change their mind. It changes their opinion of you.
Booking
Real availability, and the appointment lands on the record
A link somebody outside can use to book time. It offers availability from the same calendar your colleagues are already watching, so an accepted time is a time that genuinely exists rather than one drawn from a separate availability configuration that has drifted.
The booking creates the appointment against the relationship, so it appears in the customer history like any other. There is no scheduling system holding a parallel account of the week that somebody has to reconcile.
It is deliberately simple. A gap either side and a minimum notice are the whole of the scheduling logic. No round robin across a team, no payment at the point of booking, no routing based on answers to questions. If scheduling at that level of sophistication is the centre of your business, a specialist product will serve you better and we would rather say so here than have you find out.
Behaviour by behaviour
Following up without becoming a nuisance
Each behaviour, and the reason it exists.
- Stops on reply
- The moment someone answers, they are out of the sequence. This is the single behaviour that separates a follow up from an automated nuisance, and it is the one most often got wrong.
- Bounded
- A sequence has an end. Rules cannot chain into each other indefinitely, because a rule that can run away will eventually run away at the worst possible moment.
- Working hours aware
- A step with a delay on it is counted in working days and lands at the hour you open, so nothing you scheduled arrives at three in the morning and no reply is expected on a Sunday. A step with no delay is the exception and goes at once, because that one is a reaction rather than a schedule.
- Consent checked at send
- Every step checks consent again for that individual, so somebody who withdrew after the sequence started does not receive the next one.
- Visible on the record
- Which sequence someone is in, which step they are at and what has already been sent, on their timeline, so a colleague ringing them knows what they have received.
- Stoppable by hand
- Anybody with permission can take somebody out at any point. If a customer telephones to complain, the person who takes the call can stop the sequence themselves instead of raising it with whoever built it.
- Booking page
- A link offering real availability from the same calendar your colleagues already watch, so an accepted time is a time that exists.
- Booking creates the appointment
- Against the relationship, appearing in the customer history like any other, rather than in a separate scheduling system holding a parallel account of the week.
A follow up that arrives after somebody has already answered is the fastest way to tell a customer exactly how much attention is being paid to them.
Which is why replies and sending share a record instead of living in two systems that have to agree.
Three refusals
No purchased lists, no open rates, no round robin
No cold outbound
No contact database. No email address finding. No automated prospecting across channels. No way to import a purchased list and begin sending to it.
That is a position rather than a gap. The tooling that makes cold outbound easy at volume is the tooling that makes it easy to do badly at volume, and we would rather not build it. The rules on unsolicited electronic marketing in the United Kingdom sit in the Privacy and Electronic Communications Regulations, and they are worth reading before anyone buys a list. If that is how you grow, this is the wrong product.
No open rates or click tracking
Partly because those numbers have become substantially less reliable as mail clients began fetching images on the recipient's behalf, which makes an open a measurement of someone's mail provider rather than of a person. A metric that is quietly wrong is worse than no metric, because decisions get made with it.
What you can see is who replied, which is the outcome that matters and the one that cannot be faked by a proxy server.
No sophisticated scheduling
The booking page is a link, a calendar and an appointment. Round robin, buffers, routing and payment do not exist. That is the right level for most small organisations and clearly not enough for a business whose operations are scheduling.
The first sequence
Start with the chase you already fail to do
The single behaviour that separates a follow up from a nuisance is stopping the moment someone replies. Everything else on this page is detail.
Start with the follow up you already fail to do
Not a nurture programme. The specific thing that already happens in your business and already gets missed: the quote that was sent and never chased, the enquiry that went quiet in week two, the customer who said try me after the summer.
Automating something you already do badly is a much smaller step than inventing a programme, and it produces a measurable result in a fortnight rather than an argument about tone.
Use it for the boring things too
Most organisations think of sequences as sales tooling. The higher return use is frequently operational: chasing a document, reminding someone about a renewal, confirming an appointment. Those are the follow ups that get forgotten and they are the ones customers are actively pleased to receive.
Three steps, not eleven
A first follow up a few days after, a second a week or so later, and a last one that says plainly that you will stop unless they say otherwise. That third message is the one people leave out and it is the one that gets the most replies, because it gives someone a reason to answer now.
Write them as though you had written them individually
Short, specific to what was actually discussed, and signed by a person. The test is whether the recipient could tell it was automated. If they could, it is doing your reputation more harm than the follow up is doing good.
This is why long sequences fail. No one can write eleven messages that each sound individual, so by the fourth they are generic, and generic is the thing people unsubscribe from.
Count the gaps in working days
A sequence that fires on a Sunday afternoon is announcing itself. The same clock the rest of the product uses applies here, and the effect is that follow ups land when a person would have sent them.
Then look at what stopped
After a month, the useful number is not how many were sent. It is how many stopped early because somebody replied. That is the count of conversations that would not otherwise have happened, and it is the entire return on the exercise.
One calendar only
A booking page that does not create a second calendar
It writes into the calendar your colleagues already read
Which is the whole requirement. A booking tool with its own calendar produces the exact problem it was bought to solve, one step later: somebody is double booked while both systems looked clear.
Decide your availability once, honestly
The common mistake is offering more than you want. Somebody who publishes every hour of every day gets booked at eight in the morning by someone in another time zone, and then adds rules, and then stops using it.
Offer the windows you would genuinely accept. A booking page with four good slots a day is used more than one with twenty reluctant ones.
The appointment lands against the record
So the two minutes before the meeting are spent reading what has happened rather than looking for it. For a first meeting arranged by somebody else in your business, this is the difference between arriving informed and arriving apologetic.
Clashes are reported rather than prevented
Deliberately, everywhere in the product. The person booking usually knows something the calendar does not, and a system that refuses is a system people work around by not recording things in it.
Three tables
A sequence, a step and an enrolment are three different things
Three tables, kept apart on purpose, and what each one is answerable for.
The sequence itself holds almost nothing
A sequence has a name, a description if someone bothered to write one, an owner, a purpose, and a flag saying whether it is running. That is very nearly all of it. The plan is deliberately thin, because the interesting state belongs to the people on it rather than to the thing they are on.
The purpose is the field people read past and it is the one that governs the whole thing. It is the word consent is checked against. A sequence whose purpose is sales asks whether this particular person agreed to hear from you about sales, not whether they agreed to hear from you at all. Two sequences with two different purposes will therefore reach two different sets of people out of the same hundred contacts, and that is the intended behaviour rather than a quirk of it.
Steps are an ordered list, and there are two kinds of them
Each step carries its position in the order, a delay in days, and a kind. The delay is held as a number of days rather than as a date, because a sequence is written once and then run against people who join it on different mornings. What the delay resolves to is stored on the enrolment, which is the only place it can honestly live.
One kind of step sends an email built from a template. The other raises a task. The task step is the half most people ignore and it is frequently the more valuable of the two: making the third step of a chase sequence a task titled "Ring them" puts a piece of work on someone's list on the right morning, and a telephone call on day eight is worth a great deal more than a third email on day eight.
A step that sends an email must name a template that exists and that the person writing the sequence is allowed to read, and both are checked when the sequence is saved rather than when it runs. A step pointing at a template that is not there would otherwise fail three weeks later, against one person, quietly, and that is the worst of all the available moments to find out.
A sequence may hold at most twenty steps, and any one delay may be at most 365 days. Neither ceiling will trouble anyone writing the three steps we suggest. They exist so that a figure typed with an extra nought on the end is refused in front of the person who typed it.
The enrolment is one person, part way through
An enrolment names the sequence and the person, and then carries the state that actually moves: whether it is active, stopped or finished, which step the person has reached, when the next step falls due, why it stopped if it stopped, when it started and when it ended. That is the row a colleague is really looking at when they ask what this customer has been sent.
The reason for stopping is written down as words rather than as a code, and it appears on the record. In practice it reads as one of five things: they replied, there is no consent for this purpose, they are restricted from processing, there is no address to write to, or somebody stopped it by hand and said why. A sequence that went quiet without saying which of those happened would be useless to the person holding the telephone.
The enrolment also carries the owner and the sensitivity of the contact it is about, copied across at the moment it is created. Correspondence with a sensitive relationship is itself sensitive, and without that copy the list of who is currently being chased would name somebody a reader is refused everywhere else in the product.
One person may hold exactly one enrolment on any one sequence. That is a unique index in the database instead of a check in the code, because chasing the same person twice down the same sequence is the failure a sequence is most often blamed for, and it should not be possible to arrange it by accident from two different directions.
And the booking page, field by field
A booking page holds the last part of its own address, a name, a description, an owner, the length of the meeting in minutes, the least notice someone may give, how far ahead the page will look, a location if the meeting has one, a time zone written as a proper zone name from the IANA time zone database, and whether the page is open. Thirty minutes, a day of notice and a month of horizon are what you get without typing anything.
Availability underneath it is a set of windows, and a window is a weekday and two times expressed as minutes past local midnight. The default is nine to five, Monday to Friday. Times are stored that way, against a named zone, rather than as instants, because nine in the morning is a standing intention rather than a moment: the instant it falls on in March is not the instant it falls on in July, and a booking page that confuses the two is wrong for half of every year.
One quote, followed
One quote, one reply, one meeting, and what the records say afterwards
Hypothetical. There is no customer behind this and none of it is a case study.
Consider a seven person architectural practice
An enquiry arrives through the form on their website on a Monday. It becomes a lead, with the consent the form recorded attached to the person who filled it in, including what it said above the box they ticked. On Wednesday the lead is converted, a contact exists, and a fee proposal goes out that afternoon.
Somebody then puts that contact on the practice's quote chasing sequence. It has three steps: an email three days later, a second email five days after that, and a task on day fourteen titled "Ring them or close it". The enrolment is created immediately, pointing at step one, and because a delay is counted in working days rather than elapsed ones its next run is the Monday: Thursday and Friday count, the weekend does not, and the third working day falls on the Monday morning.
Monday, and then the Monday after it
The first step comes due on the Monday, at the hour the practice opens. Consent is checked again at that moment, for that person, for that purpose, on that channel, rather than being taken as settled at the point they were enrolled. It is still given, the contact still has an address, and nothing is restricted, so the template is filled in with their first name and the practice's telephone number and the message goes.
The enrolment moves to step two and its next run is set for five days later. A line appears on the person's timeline saying a sequence message was sent. No one in the practice did anything, and the record shows what happened anyway, which is the only arrangement under which a colleague can be trusted to know.
Thursday, when they reply
The customer answers, saying the fee is fine but they want to talk about the programme. That reply arrives as an ordinary email. It is matched to the contact by the address it came from, joins the conversation, and every active enrolment that person holds is stopped, with the reason recorded as that they replied. Not the one sequence that sent the message. Every one of them, including the newsletter chase they happened to be on as well, because a person who has just written to you is not someone to keep chasing about anything.
The second email is therefore never sent, and the task on day fourteen is never raised. That is the whole of the mechanism, and it is worth being plain that it is not clever. It is a query on a single record, in a single database, run at the moment the message lands.
The following week, the meeting
Someone sends the customer the address of the practice's booking page. What the customer sees is the name of the meeting, what it is about, how long it is, and where it happens, and nothing else. No list of who works there, and no way to tell whether a missing time is a meeting already in the diary or an hour the practice does not work.
Times appear in the customer's own zone with the practice's hours named beside them, which matters more than it sounds: a client in Spain booking with a firm in Sheffield needs to know which ten o'clock is being offered. They take Thursday morning, give their name and address, and add a note saying they will bring the survey drawings.
What exists afterwards
Because the address matches the contact created back in that first week, the booking does not make a second record of the same person. It makes an appointment against the existing relationship, titled with the page name and the customer's name, carrying the note and the location. The customer is recorded as an attendee who has accepted. A line goes on the timeline saying they booked through the booking page. An entry goes in the audit trail attributed to a visitor rather than to any colleague, naming the page and the address used, because nobody with an account did this.
Two things are worth saying about the end of that story, since neither is obvious. The product does not send the customer a confirmation email: the page tells them they will hear from the practice directly, and if the practice wants to confirm in writing then a person confirms in writing. And a contact created by a booking has no consent record of any kind, so putting that new person straight onto a sequence would stop at the first step with no consent for this purpose. Both are deliberate. Both surprise people in their first fortnight.
Eight situations
A reply from an address you have never seen, and the hour that does not exist
Every one of these is a real situation rather than a hypothetical objection.
They reply from an address you have never seen
A reply is matched to a person by the address it came from. Someone who enquired from their work address and answers from their telephone using a personal one is not recognised as the same person, so their conversation arrives unattached and their sequence keeps running. This is the honest limit of the mechanism and it is worth knowing before it happens rather than afterwards.
What fixes it is adding the second address to the contact, which is worth doing the first time you see it. What does not fix it is a cleverer guess: a system that decided two addresses were the same person because the names looked alike would eventually attach one customer's correspondence to another customer's record, and that is a far worse failure than a missed match.
They withdraw consent on the Tuesday, and a step falls due on the Friday
The Friday check governs. Consent is asked again for that individual immediately before every message, and the enrolment stops rather than sends. The same is true if they were restricted from processing in the meantime, and true again if the address was removed from their record: rather than failing, the enrolment stops and says which of the three it was.
The message could not be delivered
The step is still counted as taken. That is the uncomfortable choice of the two available, and it is made deliberately: repeating a step because a send failed would chase somebody twice, and being chased twice is worse for the relationship than missing one message. A failed send is a fact about the sequence instead of an instruction to try harder.
You want to put the same person through the same sequence again next year
You cannot. One enrolment per person per sequence, and a finished one counts. The reasoning is that the common cause of somebody being enrolled a second time is not a considered decision twelve months later, it is two colleagues doing the same thing in the same week, or an automation quietly repeating itself. Refusing all of them is cruder than distinguishing between them, and it fails in the direction that does not embarrass anyone.
The same reasoning applies to editing. A sequence's steps are written once. If the wording turns out to be wrong, you write another sequence and enrol people on that, which also leaves the old enrolments telling the truth about what was actually sent to the people who received it.
The morning the clocks go forward
There is an hour that does not happen. A booking page whose windows run to two in the morning would, on that one day, offer half past one, and the instant behind half past one that morning is really half past two. A slot whose local time is not the local time intended did not exist, so it is left out of the list rather than offered and then argued about.
The same arithmetic problem is why availability is walked by calendar date rather than by adding twenty four hours at a time. Adding a day to an instant is not adding a day to a date, and doing it the easy way loses a whole working day from the listing once a year, depending only on what time the range happened to start.
Two people choose the same slot within the same second
One of them gets it, and the other is told that the time has just been taken and asked to choose again. The time is recomputed at the moment of booking rather than trusted from what the browser sent back, because a time someone's browser returns is not the same thing as a time you offered. The organisation's own database handles one request at a time, which is what makes that check dependable rather than merely likely.
A time that was never offered at all is refused on the same path, whether it was outside working hours, inside somebody's existing meeting, too soon, or simply typed into the request by someone curious.
The person booking is a sensitive relationship
The appointment is marked sensitive too, and stays off the calendar of anyone barred from that record. The title of a meeting carries the customer's name, so without that the marking could be read straight off the calendar by the very people it exists to keep it from.
They ask to be erased
The enrolments go with them, along with their consent records, their timeline and their attendance at meetings, because none of it has a life of its own once the person is gone. Appointments and conversations are the organisation's own records, so those stay and stop naming anyone.
Two different rights
Writing the words and choosing the person are two different rights
Five permissions, and why enrolling is one of them
There are three permissions for sequences and two for booking. Sequences separate seeing them, writing them, and putting somebody on one. Booking separates seeing the pages from publishing them.
Splitting the writing from the enrolling is the decision worth explaining. Composing what the organisation says to its customers under its own name is an editorial act, and deciding that this particular customer should receive it is an operational one. They are held by different people in most organisations of this size, they go wrong in different ways, and a single permission covering both would mean that letting somebody chase their own quotes also let them rewrite what everybody sends.
What each starting role can do
A manager writes sequences and enrols anyone. Someone on the standard role writes sequences of their own and may enrol any contact they can see, which is the shape most salespeople need: their own wording for their own chasing, without touching what anyone else wrote. Someone on the restricted role writes none at all and may enrol only people whose records belong to them. A read only role sees which sequences exist and who is on them, and can put nobody on anything.
Stopping is held with enrolling rather than with writing, on purpose. The person who takes the complaint is usually the person who was allowed to start the chase in the first place, and requiring them to find whoever wrote the sequence is how a resolved complaint turns into a repeated one on the Thursday.
These are starting points rather than fixed positions. Roles are the organisation's to change, and the scope attached to each permission is what actually decides reach.
A booking page in a colleague's name is a claim on their diary
Anyone may publish a booking page for themselves. Naming a colleague as its owner needs the booking permission at full reach, and the reason is not administrative tidiness. A page publishes when its owner is free, and when somebody is free is the exact inverse of when they are busy. Anyone who could publish a page in a colleague's name could therefore read that colleague's diary off the internet without being allowed to see a single appointment in it. Bookings also land on the named person's calendar rather than on the calendar of whoever made the page, which is the second reason.
Closing a page is held to the same standard as publishing one, so a colleague's page cannot be quietly taken down by someone else. A closed page and a page that never existed give a visitor exactly the same answer, so the addresses cannot be walked to find out who works for you.
What the plans do to this
Follow up sequences arrive on Standard, as the pricing table says, along with the shared inbox they depend on for knowing that somebody replied. If a plan lapses, sequences stop advancing instead of failing: nothing further is sent, the enrolments sit where they are, and nothing is lost or quietly deleted.
The obligations around all of this are on every plan, including the free one, permanently. Every sequence created, every person enrolled, every enrolment stopped and the reason given, every booking page opened or closed, and every booking made by someone outside the organisation goes into the audit trail nobody can tidy up. Consent records, subject access tooling and complete export cover the same ground. None of it is something you buy your way up to, because a record of what you sent to whom, and on what basis, is not a feature.
The trades where chasing is the whole job
Follow ups earn their keep fastest where a price goes out and nothing comes back. Recruitment runs two populations who both need chasing, property and accommodation chases viewings and renewals on a calendar someone else controls, local service businesses lose most of their unconverted quotes to silence rather than to price, and travel holds enquiries that go quiet for months and then book. What a sequence should say differs in each, and those pages say how.
Asked about chasing
Asked about follow ups and booking
Is this cold outbound?
No, and deliberately not. There is no contact database, no email address finding, no automated prospecting across channels and no way to import a purchased list and start sending. This is for following up with people who have already been in touch with you. If cold outbound at volume is your growth model, this is the wrong product and we would rather say so.
How many steps should a sequence have?
Three or four, and considerably fewer than most tools encourage. The evidence that somebody is not interested is that they have not replied to three messages, and the fifth message does not change their mind, it changes their opinion of you. Long sequences exist because a tool made them easy rather than because they work.
What counts as a reply?
A message back from that person. Because sending and receiving share the same record, the detection is not dependent on a separate inbox integration noticing something. An out of office does not count as engagement and does not stop the sequence, which is the behaviour most people want.
Can I put someone into a sequence without them having consented?
Consent is checked at every step, so a person who has not agreed to that channel will not receive the messages. Whether a particular follow up to somebody who enquired requires consent in your jurisdiction is a question for you rather than for us, and the product records what they agreed to and when so that you can answer it.
Can two people be in different sequences at the same organisation?
Yes. Sequences run against people rather than organisations, because consent and replies belong to individuals. Two contacts at one customer can be at different stages of different sequences, and both are visible on the organisation's record.
Does the booking page do round robin across a team?
No. It offers availability from a calendar and creates an appointment, with a gap either side for travel and a minimum notice you set. There is no round robin, no payment at the point of booking and no routing based on answers to questions. If scheduling at that level is central to your business, a specialist tool will serve you better.
What happens if the calendar is wrong?
The booking page offers availability from the same calendar your colleagues are already watching, so it is as right as that calendar is. There is no separate availability configuration that can drift out of step with reality, which is the usual cause of a double booking in scheduling tools.
Can I see how a sequence performed?
You can see who is in it, who replied and who did not, on the records. There is no separate analytics view with open rates and click rates, partly because those numbers have become substantially less reliable as mail clients started fetching images on the recipient's behalf, and a metric that is quietly wrong is worse than no metric.
Follow up on one month of quotes
Three steps, written as though to one person, stopping the moment anybody replies. That alone will tell you what following up is worth in your business.
Three people, a thousand relationships, no card and no time limit.