Shut Down or Double Down? Kit's Concentrated Bet on a Stalled Product
Short answer: Shut down a stalled SaaS product when no identifiable customer group repeatedly pays, stays and gets a measurable result, or when another test would put your household or viable business at risk. Double down only when retained customers prove there is a valuable core, you can name one plausible constraint holding growth back, and you can fund a 12-week test with a fixed loss limit. More money is not the decision. A concentrated, time-boxed attempt to disprove your diagnosis is.
Founders often treat closing and persevering as judgements about character. They are capital allocation decisions. Past effort is a sunk cost, while flat monthly revenue can conceal customer evidence that has not been properly tested.
Kit, then called ConvertKit, faced that distinction early. Its account says Nathan Barry founded it in January 2013, aimed for $5,000 in monthly recurring revenue within six months while investing $5,000, and reached about $2,000 before growth stalled. In late 2014, the choice became explicit: close it or commit properly. Kit's published history sets out the figures. They are company-reported, not independently audited.
Use the Double-Down Decision Ledger
The Double-Down Decision Ledger separates evidence from hope. Complete all six entries before committing another pound.
| Ledger entry | Evidence to record | Double-down signal | Shutdown signal | |---|---|---|---| | Customer value | Retention, repeated use and specific outcomes | One segment keeps paying and depends on the product | Users praise it but rarely use or renew | | Constraint | Lost-sale notes, onboarding failures and channel results | One bottleneck explains most stalled growth | Several unrelated failures need simultaneous repair | | Concentration | Named segment, promise, acquisition route and core job | You can narrow all four for one test | The plan remains "improve marketing" or "add features" | | Cash exposure | Test cost, operating burn and protected reserve | The loss is affordable if the test fails | Failure threatens payroll, tax, debt payments or household bills | | Leading evidence | Conversations, migrations, activation and paid conversions | Weekly measures can change before revenue catches up | Only a distant revenue goal can show progress | | Stop rule | Date and minimum result agreed in advance | A decision can be made within 12 weeks | The deadline moves whenever results disappoint |
My position is strict: do not double down on the whole business. Double down on one falsifiable explanation for the stall. Expanding the product, audience and marketing activity together is a more expensive form of uncertainty.
Start with the paying behaviour, not total sign-ups
A small group of retained customers can justify another test. A large group of inactive accounts cannot. Split customers by who they are, why they bought and when they joined. Check activation, continued payment and use of the promised core job.
Use your own billing and product records: opening customers, cancellations, retained revenue, repeated core actions and support conversations. Compare at least three recent monthly cohorts where your sales cycle permits. Annual contracts need a longer window and active-use evidence before renewal.
Look for concentration. Ten professional photographers using the product weekly may be better evidence than 200 mixed free users. If no segment repeatedly uses and pays, revisit the problem before financing more development. [Why people want an idea but will not pay](/business-ideas/why-people-want-your-idea-but-will-not-pay/) separates polite interest from an economic problem.
Diagnose one constraint that could explain the stall
Stalled revenue is an outcome, not a diagnosis. Trace the customer path from qualified prospect to continued payment and locate the sharpest break.
| Observed pattern | Likely constraint to test | First evidence to collect | |---|---|---| | Qualified prospects decline meetings | Segment or promise | Ten refusal conversations and message variants | | Prospects attend but do not buy | Value, price or trust | Decision notes from the next 15 opportunities | | Buyers pay but never reach first value | Onboarding | Time to core action for every new account | | Activated customers cancel quickly | Product outcome or fit | Cancellation reason and usage before cancellation | | Customers stay but acquisition is slow | Distribution | One manual channel tested to a fixed prospect list |
Kit's stated response was concentrated. The company says Barry invested $50,000, focused on professional bloggers, hired an internal team and began direct sales. It also manually moved new customers through concierge migrations. The company account reports growth from $1,300 to $1,600 MRR in the first month, then to $2,000 the following month. That sequence did not merely add money. It narrowed the audience, changed the sales motion and removed switching friction.
The public account omits prospect volume, conversion, churn, founder living costs and the $50,000 allocation. Do not borrow the budget. Borrow the logic: identify a customer and bottleneck, then spend against that bottleneck.
Turn the diagnosis into a capped bet
A useful double-down budget fixes cash, time and scope before activity begins.
Protect committed tax, payroll, supplier bills, debt, contractual liabilities and a household reserve. Then calculate the most you can lose. A [13-week cash flow forecast](/business-finance/build-a-13-week-cash-flow-forecast/) shows whether cash runs out before evidence arrives.
Choose a period covering one normal buying and activation cycle. Twelve weeks is a useful guide for low-priced SaaS sold directly to small businesses. Six-month procurement needs a longer deadline.
Forbid unrelated work. Fixing onboarding while redesigning the website and testing advertising makes the result uninterpretable. One segment, route, promise and product outcome are enough.
If you hire staff, change contracts or invest through a company, employment, tax and insolvency duties vary by country and company position. Take advice from a qualified local accountant or solicitor before committing money that may be owed to employees, creditors or the tax authority.
Worked example: CreatorPost Email
CreatorPost Email is a fictional SaaS business for independent publishers. It has 40 paying customers at £30 a month, producing £1,200 MRR. Hosting, payment fees and support consume £6 per account, so monthly contribution is:
40 customers x (£30 price - £6 variable cost) = £960
The founder has £24,000 in cash, but £12,000 is reserved for household costs and an orderly closure. Another £3,600 covers three months of current operating costs. That leaves no more than £8,400 for a controlled test:
| Test use | Amount | |---|---:| | Product work on publisher migration | £3,600 | | Temporary migration support | £2,400 | | Direct outreach to 150 qualified publishers | £1,200 | | Contingency | £1,200 | | Total maximum loss | £8,400 |
The diagnosis is that qualified publishers value the product but avoid switching because moving subscriber tags and sequences is risky. The 12-week bet therefore offers a paid £40 plan with a managed migration. The success threshold is 65 retained customers at £40, each costing £8 a month to serve:
65 customers x (£40 - £8) = £2,080 monthly contribution
That is £1,120 more monthly contribution than the current £960. If sustained, simple payback on the £8,400 test cost is:
£8,400 / £1,120 = 7.5 months
The surprising result is that reaching £2,600 MRR does not instantly recover the bet. It only creates a plausible 7.5-month payback before allowing for churn, tax or the founder's salary. CreatorPost therefore adds a week-six stop rule: fewer than ten qualified sales calls or fewer than five agreed migrations means the switching-friction diagnosis is weak, so remaining discretionary spending stops.
Judge the mechanism before the headline revenue
Revenue lags during a short test. Inspect the chain creating it. If outreach produces conversations, paid commitments, completed migrations and core use, the diagnosis is gaining support. If the first link fails, more product work is unlikely to help.
Kit reports reaching $15,000 MRR by July 2015 and $98,000 by the end of that year. Its history attributes the later acceleration partly to customer referrals and an affiliate programme. Those later results do not prove that every stalled SaaS product should persist. They show that the earlier change produced observable movement quickly: $1,300 to $1,600, then $2,000. Your test also needs early evidence before the full revenue target.
Track results weekly by cohort. Keep customers acquired before the test separate from those won through the new route. Compare acquisition effort, activation, support and retention. Use [cash flow versus profit](/business-finance/cash-flow-versus-profit/) so improved recurring revenue does not hide a short-term cash gap.
Make the decision on the date you set
Use the next 12 weeks in this order:
- In days 1 to 3, protect taxes, liabilities and your household or closure reserve. Write the maximum remaining loss in pounds.
- By day 7, identify the retained customer segment and collect evidence for one constraint from billing, usage, lost sales and cancellation conversations.
- By day 10, define one segment, promise, route and core outcome. Record week-six and week-12 thresholds before spending.
- During weeks 2 to 6, run the manual version first. Sell directly, help customers switch and observe where progress stops.
- At week 6, end spending if the leading threshold fails. Do not rewrite it after seeing the result.
- At week 12, compare new-cohort contribution and retention with the budget. Continue only if the mechanism works and your [business can make money](/business-finance/can-this-business-make-money/) after founder pay and overhead.
If the test fails, close or pause the product cleanly: notify customers, honour refunds and data commitments, settle liabilities and preserve records required in your jurisdiction. A disciplined closure releases cash and attention. That is an operating decision, not a personal verdict.
Related guides
Frequently asked questions
How much money should I risk on a double-down test?
Risk only cash left after protecting tax, payroll, creditors, contractual commitments, closure costs and a realistic household reserve. The right amount is therefore determined by your downside, not by another founder's investment. Build a 13-week cash forecast, assume the test returns nothing within that period, and check the lowest closing cash balance. If that balance breaches a protected obligation, reduce the test. Include your own minimum living requirement as a real cost. A £5,000 experiment can be reckless with £7,000 available and conservative with £100,000 available.
What if customers like the product but growth is still slow?
Treat compliments as weak evidence and retained payment as stronger evidence. Identify whether one customer group uses the core function repeatedly, renews and would experience a real cost if the product disappeared. If so, slow growth may reflect positioning, switching friction or distribution rather than an unwanted product. Test one of those constraints manually. If customers rarely use the product, need reminders to return or will not pay a sustainable price, "liking" it does not justify further investment. Ask about recent behaviour and outcomes, not hypothetical future use.
Can existing recurring revenue justify keeping the product alive?
Yes, but only if the revenue exceeds the work and cash required to serve it, or the product can be maintained deliberately as a small asset. Calculate contribution after hosting, payment fees, support, refunds and any founder time needed each month. Then compare that return with the best alternative use of your time. A stable £2,000 MRR product requiring two hours a week is different from one demanding daily support and emergency fixes. Keeping it small, selling it or winding it down are all legitimate choices. Growth is not the only acceptable outcome.
Should I change the niche, pricing and product at the same time?
No. Change only what your diagnosis requires, because simultaneous changes prevent you learning why the result moved. You may need a linked adjustment, such as focusing on one segment and packaging the product around that segment's job, but hold the acquisition route and success measure steady where possible. Start with direct customer conversations and a manual offer before engineering a specialised feature. If the niche responds but price prevents profitable delivery, test pricing next. Keep a decision log so evidence from one round determines the following round rather than being lost in activity.
How long should I give a stalled SaaS product before closing it?
Give the specific test one complete buying and activation cycle, not an arbitrary number of years. Twelve weeks is a useful working guide for founder-led, low-priced SaaS sold to small organisations, because it permits outreach, sales, onboarding and an early retention observation. Enterprise procurement or seasonal buying may require longer. Set an earlier leading threshold so you do not spend for 12 weeks when nobody will even take a qualified call. A deadline only works when the budget and minimum result are written before the test begins.
How do I close a SaaS product without damaging customer trust?
Tell paying customers early, give a clear service end date and explain what will happen to billing, exports, stored data and support. Stop taking annual payments that extend beyond the closure date. Provide usable data exports and reasonable migration time, then issue refunds required by your terms and local consumer or contract law. Securely retain or delete data according to applicable obligations, and keep enough operational access to resolve final issues. Requirements differ by country and customer type, so ask a qualified local solicitor or data protection adviser about contracts, notices and records.
Comments (0)