There is a version of market research that produces excellent slides and no customers.
You collect complaints. You tag them. You notice that people find the incumbent expensive, or slow, or bloated. You write a positioning doc about being the affordable, fast, focused alternative. And then you launch into a market that agreed with every word you said and bought nothing.
The problem is not that the complaints were fake. The problem is that a complaint is the cheapest thing a person can produce. It costs nothing to be annoyed on the internet. Most people who complain about a tool on Monday are still logged into it on Friday, and will be next year.
What you actually want is the moment the annoyance converts into motion. Someone opens a thread and writes: we're moving off Asana next quarter, what are people using? That sentence is not a complaint. It is a purchase decision, in progress, in public, with a timeline attached.
We call these switching threads. They are the highest-intent signal available to a founder, they are structurally easy to find, and almost nobody monitors them systematically.
The intent ladder
Not all customer pain is worth the same. It helps to think of public pain as a ladder, where each rung costs the person more to climb.
Rung one: venting. God, this UI. No timeline, no alternatives named, no decision pending. Enormous volume. This is what most Reddit monitoring picks up, and it is the rung where false positives live — echo chambers and loud minorities generate venting almost by definition.
Rung two: comparing. Has anyone used X vs Y? Slightly warmer. The person is curious, but curiosity is often just procrastination with a browser tab open.
Rung three: evaluating. We're trialling three tools this month, here's our shortlist. Now there is a process. There is a budget conversation happening somewhere off-screen.
Rung four: switching. We've decided to leave. What do we move to, and how painful is the migration? Decision made. The only open question is who receives the money.
Most founders monitor rung one because rung one is loud, and loud is easy to find. But rung four is where the market actually reallocates. A single switching thread with eleven replies is worth more than four hundred venting posts, and the reason is simple: the switcher has already done the hardest part of your sales job for you. They have decided that the status quo is unacceptable. You are not selling change anymore. You are just auditioning.
Switching has a vocabulary
Here is the useful structural fact: people announce that they are leaving using a small, predictable set of phrases. Switching intent is linguistically regular in a way that complaining is not.
Complaints can be phrased ten thousand ways. Departures cannot. Watch for:
- alternatives to
- migrating off / moving away from
- finally cancelled
- is anyone still using
- worth switching from
- we outgrew
- replacing X with
- anyone regret leaving
- does anyone have experience migrating
- export my data from
- how do I get my data out of
Those last two are the sharpest of the lot, and the most overlooked. Nobody googles how to export their data from a tool they intend to keep using. A data-export question is a resignation letter with the subject line missing.
The practical implication is that switching intent is monitorable. You do not need sentiment analysis or a model that understands frustration. You need a list of about a dozen phrases and a competitor name, and you need to be watching the places those phrases get typed.
Where switching threads actually live
Reddit and Hacker News are the obvious answer and the incomplete one. Four less obvious sources:
The competitor's own community forum. This is the richest and least contested source in the entire market-intelligence stack. When a paying customer of a tool posts how do I bulk-export three years of records, and a moderator responds with a link to a docs page that does not answer the question, and the thread dies — you have just watched a customer leave, in slow motion, on the incumbent's own property. Almost nobody reads their competitors' support forums. Everybody reads their competitors' landing pages.
Comparison-page traffic. G2, Capterra, and the "X vs Y" pages of every competitor in your category. Reviews on these sites skew incentivised, but the comparison pages are useful for a different reason: they tell you which pairs of products people actually consider substitutes. Sometimes that pairing surprises you and reveals you have been benchmarking against the wrong company.
Job boards, sideways. A company posting for someone with experience in a specific tool is committing to it. A company that has quietly stopped mentioning a tool it named in every posting for two years is drifting away from it. This is the same read described in the hiring tell, applied to churn instead of roadmap.
Migration guides. If a competitor publishes How to import your data from Acme, they have decided Acme's customers are worth hunting. If someone publishes How to migrate from you — congratulations, you are Acme. Migration guides are the most honest competitive document a company produces, because there is no reason to write one except to take someone's customers. Track who is writing them and about whom, and you have a directional map of where the market's money is moving. This is a hard signal in the same family as changelog espionage, just pointed at the customer base rather than the product.
Read the reason, not the noise
Finding the thread is the easy half. The harder half is that switchers routinely misreport why they are switching.
The stated reason is almost never the real one, not because people lie, but because the real reason is unflattering, or complicated, or hard to articulate in a Reddit comment. Three translations worth internalising:
"It got too expensive" usually means the value stopped compounding. Nobody cancels a tool that is saving them thirty hours a month because the price went up twenty dollars. When someone leaves over price, they are telling you that at the moment of the invoice, they could not remember what they were paying for. That is a value-delivery problem wearing a pricing costume. Chasing it with a cheaper plan is how you acquire a customer who will leave you for the same reason in eighteen months.
"It's too complicated" usually means one specific workflow broke. Products do not become complicated all at once. Somebody hit a particular wall on a particular Tuesday, and the wall is where the actual product opportunity is. Your job is to reconstruct the wall from the wreckage. Read the replies, not just the post — the person who says ugh, same, especially when you try to do [very specific thing] has handed you the spec.
"We outgrew it" usually means the tool assumed a team size it never revisited. This is a design failure, not a feature failure, and it is the single most common way that good products lose good customers. It is also, from where you are sitting, an opportunity that comes with a free customer list.
The move in all three cases is the same: extract the job that was failing, not the feature that was named. Features are what people ask for. Jobs are what they pay for. A switching thread is one of the very few places you can watch someone describe both in the same paragraph.
Now do nothing (for a moment)
Here is where founders lose the thread, sometimes literally.
You have found someone announcing they are leaving your competitor. Every instinct says: reply immediately, name your product, offer a discount, get there before anyone else does. And you can watch this instinct destroy the opportunity in real time on any sufficiently active subreddit, where the first three replies to any switching thread are founders pitching, and the fourth is a moderator removing them.
A switching thread is not a lead. It is a person, mid-decision, who has just told a community they trust that they need help. If you arrive as a vendor, you are noise. If you arrive as someone who understands their problem better than they have articulated it, you are the answer they were hoping the thread would produce.
The mechanics of doing this without sounding like a spam bot are covered in the reply gap, and they apply with double force here, because the stakes are higher and the audience is watching. The short version: answer the question they asked, including when the honest answer is a competitor. Disclose who you are. Do not link on the first reply. Be genuinely useful to a person who might never buy from you, in a public place, where the eleven other people reading the thread are also mid-decision and have not said so.
The switcher may not become your customer. The lurkers frequently do.
There is an apparent contradiction here, and it is worth naming before it does damage. The next section argues that switching signals expire fast. This section argues for patience. Both are true, because the speed belongs at a different point in the chain than most founders put it.
Speed belongs at detection. Patience belongs at response. If you find the thread on day two, you can spend two days understanding the person's actual problem and still arrive while the decision is open. If you find it on day twenty-six, you will be rushed into exactly the pitch that gets you removed. Founders who move too fast in the reply are almost always compensating for having moved too slowly in the finding.
Switching signals rot faster than anything else
Every signal decays, but they decay at wildly different rates, and this is the part most research workflows get wrong.
A competitor's pricing change stays true for months. A complaint about a slow UI stays true until they fix the UI. A structural market gap can sit unexploited for years.
A switching thread is dead in thirty days. Sometimes seven. The person announced they were leaving, twelve people gave them suggestions, they picked one, they migrated, and now they are somebody else's customer with switching costs freshly rebuilt around them. The window is not a metaphor. It closes.
This is why switching intent punishes the standard research pattern so brutally. If your market research lives in a document you review quarterly, you will find every switching thread in your category approximately three weeks after it stopped mattering. You will read it, recognise it as a perfect fit, feel a small internal collapse, and move on to the next dead lead.
The only workflow that works for perishable signals is a continuous one. Watch the phrases, watch the forums, get told when they fire, respond in days rather than months. Not because speed is inherently virtuous, but because this particular signal type has a half-life measured in the same unit as a sprint.
What to do on Monday
Concretely, if you want to start:
- List your five real competitors. Not the ones in your pitch deck — the ones customers actually name when they compare. The comparison pages will tell you who these are, and you may not like the answer.
- Take the eleven phrases above and pair each one with each competitor name. That is your monitoring surface. It is smaller than it sounds and it is almost entirely unwatched.
- Find each competitor's community forum or support subreddit, and read the last month of unanswered questions. Unanswered is the operative word.
- Log what you find with the reason attached, translated from stated to actual using the three patterns above. A switching signal without a diagnosed reason is just a name you cannot use.
- Score it for perishability, not just for strength. A weak signal that stays true for a year may be worth less than a strong one that expires on Friday — or considerably more, depending on what you can actually ship. Urgency, evidence, and a defensible next action: that is what the opportunity scoring framework exists to make legible, and perishability is simply urgency with a clock attached.
None of this is difficult. It is just relentless. Five competitors, eleven phrases, four surfaces, every day, forever, with the diagnosis written down while it is still fresh — that is the part that quietly stops happening in week three, and it stops happening to disciplined people. Whatever you build to carry it, build the carrying, not the enthusiasm.
Complaints tell you a market is unhappy. That is useful, eventually, in aggregate, for positioning.
Switching threads tell you someone is standing in a room with a credit card, asking a group of strangers who to give it to.
The difference is not subtle, and it is not a matter of degree. One is evidence about a market. The other is a market, moving, right now, whether or not you were watching.