Baymard Institute keeps a running tally of every documented study it can find on shopping cart abandonment. Across 50 of them, the average is 70.22 percent.
Seven in ten people who get as far as putting something in a cart leave without buying, and almost none of them say why. They do not open a support ticket. They do not answer a survey. Baymard’s breakdown of the preventable reasons reads like a list of things no one would ever bother to complain about: extra costs at checkout (40 percent), being asked to create an account (18 percent), and a checkout that took too long or got too complicated (17 percent).
That is the problem sitting underneath every customer journey map a small business owner has ever drawn. The customers you can describe in detail are, by definition, the ones who stayed long enough to become describable. The map gets built from them, and the people it was supposed to find never turn up in the research.
Online stores are not worse at this. They are just the ones who can see it
The 70 percent figure is usually read as a fact about ecommerce. Read it the other way and it says something more useful: an online store is not the business with the worst drop-off, it is the business that happens to have instrumentation pointed at the exact moment of loss.
A contractor who sends 30 quotes and books 9 of them has the same shape of loss and no dashboard showing it. A consultant whose discovery calls stop converting learns about it as a slow month, not as an event. The loss is identical. What differs is whether anything recorded it. Most small businesses are running the ecommerce abandonment problem with the lights off, which is why the honest first question is not “what does our journey look like” but “which part of it did we actually watch”.
What a journey map is supposed to contain
Nielsen Norman Group, which has been publishing on this longer than most of the tools have existed, describes the method plainly. A journey map is made by “compiling a series of user goals and actions into a timeline skeleton”, enriching that skeleton with what the person was thinking and feeling, then condensing it into something you can look at. Five things go into it: the point of view (whose journey this is), the scenario (which specific experience is being mapped), the actions and mindsets along the way, the touchpoints and channels where the customer meets your business, and the insights plus who owns fixing them.
Two of their warnings matter more to a two-person shop than to the enterprise teams the article is usually read by. The first is that a map should “result in truthful narratives, not fairy tales”. The second is about that last component: “Without ownership, no one has responsibility or empowerment to change anything”.
Here is where the small-business version diverges, and it is worth being precise about it because the fix is different. In a company with departments, ownership is the component that fails: the map gets drawn, and no one is accountable for any row on it. When you are the whole company, ownership is solved before you start. You own all of it. The component that fails instead is the research foundation, because the cheapest available input is your own memory, and your memory only contains people who talked to you.
Four moves that build the map from evidence instead of recall
Move 1: Pick one customer and one errand, then write both down
“Our customers” is not a point of view and “buying from us” is not a scenario. Both are too wide to be wrong about, which is what makes them comfortable. Narrow it until the sentence is falsifiable: a first-time buyer who found us searching on their phone and wants one specific item by Friday. A repeat client asking for a second project after a nine-month gap. You can map several of these later. You cannot map them at the same time, because they drop out in different places and averaging them hides both.
Move 2: Record before you draw
This is the move that separates a map from a memory, and for a business with a website it costs nothing. Microsoft Clarity is “Free forever” with “No limits on traffic”, and it gives you session recordings and heatmaps: you can watch what people actually did on your pages and see where attention stopped.
Give yourself a fixed budget of attention rather than an open-ended project. Watch twenty recordings of visits that ended without the thing you wanted, whether that is a purchase, a form, or a phone tap, and write one line per session about where it stopped. Twenty is enough to see repetition and small enough to finish in a sitting.
If your business does not run through a website, the equivalent evidence is a phone call. Take the last five people who asked you for a quote and did not book, call them, and ask what they ended up doing instead. That is not a proxy for the data, it is better than the data, and it is the only version of this move that a trades business or a consultancy can run. It is also the part everybody skips, because it is uncomfortable in a way that reading a dashboard is not.
Move 3: Write the stages you observed, not the stages from a template
Most templates hand you awareness, consideration, purchase, retention, advocacy. Those words are fine as a filing system and useless as a diagnosis, because no customer has ever hesitated at “consideration”. They hesitate at a shipping total, a form that wants a password, a page that did not say whether you serve their town.
Name the stages after the observable moments from your twenty sessions or your five calls. Then do the thing the touchpoint column tends to prevent: write down the gaps as well. A touchpoint list records every place the customer met your business. The drop usually happens in the space between two of them, in the four days of silence after the quote went out, and a map with no row for silence will never show you that.
Move 4: Add two columns nobody puts on a journey map
This part is our own addition rather than anything the standard method asks for, and it exists to solve the specific failure above. Next to each stage, add a column for the evidence behind it, and a column for the date you will look again.
The evidence column takes one of three entries: observed (you watched it happen), asked (someone told you directly), or assumed. Writing “assumed” next to four of your seven stages is not a failure of the exercise, it is the most valuable output of it, because it tells you precisely where to point Move 2 next month. The re-check date exists because a journey map with no expiry quietly becomes a description of the business you used to run.
A worked example, and it is a hypothetical
The following scenario is invented for illustration. It is not a customer, a case study, or a result anyone reported. It is here because the shape of the finding is the point, not the numbers.
Suppose a two-person business sells a physical product online and at weekend markets, and the owner believes the problem is discovery: not enough people know about them. Their memory supports it. Every customer they have spoken to found them at a market.
Move 1 narrows the scenario to a first-time buyer who saw the product at a market, went home, and tried to reorder online a few weeks later. Move 2 turns up twenty recordings of visits that ended without a purchase, and eleven of them stop on the same screen. Move 3 names that stage honestly: not “consideration” but “finding out what shipping costs”. The map now has a row that contradicts the owner’s belief, which is the only reason it was worth building. Discovery was working. That is how these people got there.
Notice what the map does not do. It does not tell you whether to change the shipping price, absorb it, or state it earlier on the page. It tells you which one decision is worth your week. Choosing among the three is still yours, and if the bottleneck turns out to be capacity rather than conversion, the counting method in How to Scale a Business: Count What Is Waiting is the better tool for that job.
What this costs, and the limit that actually binds
For watching behavior, three options are worth knowing, and two of them are genuinely free at small-business volume.
Microsoft Clarity is “Free forever” with “No limits on traffic”, covering heatmaps and session recordings. PostHog gives you 5,000 session recordings a month on its free tier and states plainly that “Your free allowance renews every month.” Contentsquare, which is where Hotjar now lives (“Hotjar is now part of Contentsquare. The two platforms have merged into a single, more powerful experience intelligence platform.”), has a Free plan at $0 and a Growth plan at $49 per month, or $39 per month paid yearly.
The comparison inside that last one is the part worth slowing down for, because it inverts what you would expect. Contentsquare’s free plan covers up to 200,000 monthly sessions. Its paid Growth plan starts from 7,000. The paid tier does not begin where the free tier runs out of traffic, because traffic is not what you are buying. What Growth adds is 13 months of data access against the free plan’s 1 month, and 3 projects against 1.
For a business doing a few thousand sessions a month, that reframes the decision entirely. You will never hit the free ceiling. What you will hit is the memory limit, and you hit it the first time you want to ask whether this quarter is worse than last spring. If the question you care about is seasonal, you are paying for the ability to look backwards, not for capacity.
For drawing the thing, Miro has a Free plan at $0 with 3 editable boards, a Starter plan at “$8/month per member, billed yearly”, and Business at “$20/month per member, billed yearly”. Figma offers a free Starter plan with unlimited drafts, and a Collab seat at “$3/mo” that includes FigJam. A spreadsheet also works, and for a seven-row map with two extra columns it works about as well. The canvas is the cheapest part of this and the least important; if you find yourself comparing whiteboard tools before you have watched a single recording, you have started at the wrong end.
Where an AI prompt actually helps, and where it cannot
The Customer Journey Mapper prompt on our sibling site does the structuring pass: you hand it your segments, channels, touchpoints and known friction, and it returns a stage-by-stage map with opportunities attached. It is a Pro prompt built as a five-step chain, which you can find at businessprompter.com/prompt/customer-journey-mapper alongside the rest of the library at BusinessPrompter.com.
What it is good at is the work an owner has no appetite for at 9pm: turning scattered notes into an ordered document, spotting stages you left out, and proposing what to measure. What it cannot do is Move 2. It has no access to your recordings, your customers, or your unreturned calls, and if you feed it the memory map it will produce a beautifully organized version of the same fairy tale.
That division is the honest case for using it. The analysis pass is the part a small business would otherwise have to buy from a consultant or skip entirely, and now it costs a subscription and twenty minutes. The watching stays with you, and it should, because the judgment about what a hesitation meant is the part you are actually good at and the model is not.
The part we would argue for hardest
If you only take one thing from this, make it the evidence column, and make it honest.
Journey maps have a bad reputation in small business circles, and the reputation is deserved, but the usual diagnosis is wrong. The complaint is that they are too corporate, too slow, too much process for a company of three. The real problem is that an unsourced map is indistinguishable from a confident opinion with boxes around it, and it is far more persuasive than the opinion was on its own. You will act on it. You will spend money against it. The formatting does the convincing.
Writing “assumed” in a column is a small act and it removes that entire failure mode, because a map that shows its own gaps cannot be mistaken for a finding.
The related work is worth naming so you can pick the right tool. If you already know where customers hesitate and the question is what to say to them in the first month, Strategic Communication Plan: Stop Early Customer Churn covers that cadence. If the gap is earlier and the question is how new people find you at all, How to Get Customers Without Starting Over Each Month is the one. And if what you actually need is not a journey but a clear statement of what you offer and to whom, the six lists in Value Proposition Canvas: Six Lists, No Template Needed get there faster than any map will.
Start with the twenty recordings or the five phone calls. Everything else on this page is optional, and the map you draw before you have done either of those is a picture of your own assumptions in a nicer font.
Common questions
How long does a first customer journey map take?
The drawing takes about an hour. The evidence takes longer and is the part worth budgeting: twenty session recordings is roughly a sitting, and five follow-up calls to people who did not book will take you a couple of days to actually complete. If someone promises you a finished map in ninety minutes, what they are promising is the drawing.
Do I need analytics skills to do this?
No. Move 2 is watching recordings of real visits and writing down where they stopped, which needs attention rather than statistics. Microsoft Clarity is “Free forever” with “No limits on traffic” and shows you heatmaps and recordings without any query writing.
My business has no website worth measuring. Is this still useful?
Yes, and the phone-call version of Move 2 is the more valuable one. Five calls to people who asked for a quote and did not book will tell you more about your journey than any dashboard, because a service business loses people in conversations, not on pages.
Should I pay for a behavior tool or stay on the free tier?
At small-business volume the free tiers are not the constraint. Contentsquare’s free plan covers up to 200,000 monthly sessions while its paid Growth plan starts from 7,000, because what the paid tier adds is 13 months of data access against 1 month. Pay when you need to compare this season against last one, not when you run out of traffic.
