How to build a follow-up cadence that actually gets replies
Front-load it. Put most of your touches in the first few days while the customer is still shopping, mix text with email so you are not stuck in one inbox, and send at hours a person is awake and holding their phone. Then look at the shape of the whole thing before you turn it on.
Why does the shape of a cadence matter more than the number of touches?
Because intent decays fast. A customer who filled out a form on Saturday morning is a different person by Wednesday. A cadence with five touches in the first week will beat one with twelve spread evenly across a month, even though the second one looks like more effort on paper.
The trouble is that most cadence tools make the shape impossible to see. Every step renders as a big expanded card mixing timing controls you set once with message text you edit constantly. A seven-touch cadence becomes thousands of pixels of scrolling, and nobody can tell you whether it is front-loaded or evenly spread without counting.
So people build cadences they have never actually read. That is the real failure. Not a bad strategy, an invisible one.
How do you read a cadence at a glance?
Collapse it. In bkd.ai's cadence builder each step is now one row carrying its delay, timing, channels, and resolved send time, and it expands in place only when you need to edit the message. Above the list, a timeline plots one dot per day that carries a touch, so front-loading is visible in a second.
Timing reads as a sentence rather than a form: wait two days, at morning, on weekdays. That sounds like a cosmetic change and it is not. When the setting reads like the instruction you would give a rep, you catch the mistakes.
The mistakes are usually the same three. Touches bunched on days nobody replies. A step set to send at an hour your store is closed. And a gap in the middle where the cadence quietly goes dark for nine days and the lead cools all the way off.
How should you personalize each message without writing every one?
Use merge variables for the detail, and AI for the touches where the conversation so far should shape the wording. In bkd.ai you type a forward slash in any message field and get a keyboard-driven picker of the real variables your account supports, so the first name, the vehicle of interest, or the trade-in details fill in at send time.
That last point matters more than it looks. A variable name that seems right but is not real substitutes into nothing at send time, and your customer gets a text with a blank where their name should be. bkd.ai's picker only offers fields that actually exist in your account, so that class of mistake is off the table.
Then mark individual steps as AI-personalized rather than the whole cadence. Write the touches where the exact wording matters. Let bkd.ai's Sidekick adapt the ones where knowing what the customer already said is worth more than a polished script.
Frequently asked questions
How many touches should a car dealership follow-up cadence have?
Most working dealership cadences run three to eight touches. What matters more than the count is where they sit. A cadence with five touches in the first week beats one with twelve spread evenly across a month, because buying intent decays fastest in the first few days.
What are merge variables in a follow-up message?
Merge variables are placeholders that fill in real customer detail at send time, like first name, vehicle of interest, or trade-in details. In bkd.ai you insert them by typing a forward slash in any message field, which opens a picker of the real fields your account supports.
Should follow-up messages be sent by AI or written in advance?
Both, per step. Write the touches where the wording matters, such as a first hello or a price follow-up, and mark the ones where context helps as AI-personalized so bkd.ai's Sidekick adapts them to the conversation so far. Mixing the two per step is usually stronger than choosing one for the whole cadence.