Your sales playbook should show its work

Check the drawing too

A tidy flowchart can hide a changed instruction. So a neat picture is a poor test of whether your plan survived.

On October 23, 2026, our work on Mach showed that gap. One sentence Mach could not map could sink an otherwise useful drawing. A note could also lose the line it was meant to explain.

The fix was not a busier picture. Instructions Mach cannot follow now show as notes, with your original words attached. The person checking the plan has something real to compare.

A box that says follow-up tells you little. A box that shows the channel, the person, the wait and the next condition is a promise you can check.

Treat the drawing as Mach's reading of your written plan. It is useful because you can question it before it becomes work.

This is an account of our own October 23 and 24 changes. It is not a study of customer results.

Start with words a colleague could follow

Write a short plan in words before adding branches. Name the kind of company, the person, the reason to write and who checks the draft.

Then say what happens after silence or a reply. "Follow up later" leaves both the wait and the owner open.

In this example we use three business days. That number is made up. It is not a wait we suggest for every buyer.

Ask a colleague to read your plan without seeing the drawing. If they cannot say the next action, the problem starts in the words.

Our guide to checking AI claims covers a similar habit: keep what you know apart from what the software guessed. The same rule belongs in a sales plan. [1]

A made-up plan you can check

Here is a made-up plan for Tarang Labs, a software team in Pune. Jamie Carter owns its deal with Alder Creek Bank in Ohio. Mara Ellis is the bank's chief operating officer.

The team wants one first email built on a dated fact and a matching customer story. Jamie checks the draft before anything goes out.

If Mara does not reply, the plan waits three business days, then drafts a short follow-up. An interested reply goes to Jamie.

The table is a check sheet, not a record of a real campaign. No message was sent and no meeting took place.

Read each row against the real drawing. A missing wait is a missing instruction, even when every box looks neat.

Keep the plan small enough to check in one sitting. A dozen branches make a lost instruction harder to spot.

Add a last column for any instruction the drawing leaves out. For each gap, change the plan or accept a visible note. Then ask Jamie to explain the plan without reading your words. Compare his answer with the check sheet.

A note should quote the missing instruction

Our October 23 checks covered a plan whose useful steps were refused. One sentence did not map cleanly.

That was the wrong place to fail. A whole plan should not vanish because one part needs a person to decide.

The drawing now keeps the usable steps and explains the part it could not follow. A later check made notes quote their source line.

That matters in a long plan. A general warning makes you hunt. A quoted line lets you judge the exact trade-off.

When you check, read the notes before you admire the chart. Ask if a note changes the audience, channel, timing or owner. If it does, decide that before approving any step it touches.

A warning also needs an outcome. Write down whether you accepted the change, changed the plan or removed the step. Otherwise the same doubt returns at the next check.

Failure should not erase your written plan

Another October 23 fix dealt with a failed drawing. The new words had not reached a saved drawing. Refreshing the page brought back older words.

Now Mach keeps the undrawn words in that browser until a drawing works. If time runs out, it says the drawing could not finish.

This copy lives only in that browser. Another computer will not have it. Blocked or full browser storage also limits what survives a refresh.

Whoever checks the plan should know which version they see. New words beside an old drawing are not a finished plan.

When you test a tool, make a drawing fail on purpose. Then refresh. What comes back tells you more than another perfect demo.

For a second check, hide the plan's words and explain the drawing to someone else. Then show the words and compare. This does not prove the plan will work. It tests whether the drawing means what you meant.

Count the time spent checking the plan

The cost includes writing, checking and fixing the plan. A fast first drawing can still mean a slow check.

Here is a made-up budget: 12 minutes to write, 8 to check and 5 to fix. At $60 an hour, 25 minutes costs $25.

That leaves out software fees, research charges and every later message. It is an example for planning, not a measured Mach result.

What we got wrong was treating a good drawing as proof of a faithful plan. Notes and recovery needed their own tests.

Before your next campaign, write one short plan. Compare every instruction with the drawing, and every note with its source line. Then decide which steps are ready to approve.