How do you decide whether to keep or kill a new bet?
Decide before you start. The most reliable way to judge a new product line, a new market, a new channel, or a pivot is to agree on what failure looks like on day one, write it down, and measure against it on fixed dates. Once a team has spent three months and a budget on something, every weak number starts to look like "it's still early." Criteria written in advance are the only ones nobody can quietly move.
Good kill criteria share a few traits:
- They name a number and a date. "Traction" isn't a criterion. "200 trial signups from the new market by day 45" is.
- They match what you can know at that point. Revenue from a new product rarely shows up in the first month, so early checkpoints use signs of interest, such as visits, signups, replies, and booked calls. Later checkpoints use the results that matter, such as paying customers, deal size, and cost per customer.
- They define a fair test. A bet that never got its landing page, its campaign, or its outreach didn't fail. It wasn't tried. Your criteria should say how much work and spend the bet gets before its numbers count.
- They say what happens next. Each threshold leads to keep going, change course, or stop. Without that, a missed target becomes a discussion instead of a decision.
Ninety days is a common window for marketing bets because it's long enough to ship the work and see real results, and short enough that a failing bet doesn't eat a year. A bet with a long sales cycle, such as an enterprise product, may need a longer window, while a new paid channel can often show its answer in six weeks.
The hard part isn't writing the scorecard. It's keeping it honest week after week, while the people closest to the bet want it to work. That steady, unbiased checking is a good job for an AI agent, because it doesn't get attached to the idea and it doesn't forget to look.
Kite is an AI marketer. It researches your business, builds a growth strategy, and ships the work that brings you new customers. Here's how it helps you judge a new bet on the evidence.
How Kite helps you decide whether to keep or kill a new bet
Kite treats a new bet as a goal with a deadline and a clear test. It starts with what it already knows about your business from your team's knowledge base: your products, your customers, what a lead or customer is worth to you, and how your existing channels perform. From that, it proposes the goal, the kill criteria, and the first pieces of work, and you approve them before anything starts.
Once the bet is running, Kite builds a live scorecard, checks it every week, and reports on what moved. At each checkpoint it brings you a recommendation to keep going, change course, or stop. Whatever you decide, Kite carries it out and records the result, so the lesson outlasts the bet.
- Set the kill criteria before you start
- Build a 90-day scorecard that updates itself
- Check progress every week and name what's behind
- Make the call at each checkpoint
- Record what you learned
Set the kill criteria before you start
Kite proposes kill criteria from your own numbers rather than generic benchmarks. If you've told Kite that a new customer is worth $4,000 in the first year, it can work out how much you can afford to spend to win one and set the bet's cost-per-customer limit from that. If your current website turns 3 percent of visitors into trials, it can use that rate to judge whether a new market's 0.5 percent is a slow start or a real problem.
Here's what a proposal might look like for a project-management company launching a version of its product for construction firms:
- Day 30: At least 1,500 visits to the new construction landing page and 60 trial signups. Missing both means the message or the targeting needs to change before more money goes in.
- Day 60: At least 15 sales calls booked with construction firms. Fewer than 8 means stop.
- Day 90: At least 10 paying construction customers at a cost below $1,200 each. Fewer than 5 means stop, and between 5 and 9 means change course and run one more 45-day round.
Kite explains where each number came from, so you can push back on any of them. When you agree, it saves the bet as a goal, with the outcome as its title and the criteria in its description. The reasoning behind the criteria goes into your knowledge base, where anyone on the team can read it later. Kite also proposes the first tasks for the bet, such as building the landing page, writing the launch emails, and preparing the ads. You approve the goal with the Approve button in the conversation or by replying, and the work starts.
Related use cases
- Turn a growth goal into a marketing plan Coming soon Turn a business result you want into a goal, a strategy, and the first tasks.
- Research a new market before entering it: Size up the buyers, competitors, and channels in a market before you commit to it.
Build a 90-day scorecard that updates itself
A scorecard only works if someone looks at it, so Kite builds it where the bet already lives. Each goal in Kite has its own dashboard, and Kite fills it with live charts, tables, and headline numbers for every criterion you agreed on. The numbers refresh from their sources, so nobody has to rebuild a spreadsheet before the weekly meeting.
Kite pulls the numbers from wherever they're recorded. On a website Kite hosts, its built-in analytics track visits, form submissions, and conversions on the new pages. For a site measured elsewhere, Kite reads the same figures from Google Analytics or PostHog. Revenue and paying customers come from Stripe, and pipeline comes from your CRM, such as HubSpot or Salesforce. If the bet runs on paid ads, Kite reads spend and results from the ad account.
Kite keeps the bet's numbers separate from the rest of your business. A new market's signups are counted on their own, not mixed with the signups your main product would have earned anyway. Otherwise a healthy core business can make a failing bet look fine.
Related use cases
- Build a marketing dashboard: Put your marketing numbers on one page that updates from your connected tools.
Check progress every week and name what's behind
Kite sets up a weekly check on the bet as soon as the work starts. Each week it reads the scorecard and the status of the work behind the bet, then adds a short progress note to the goal. The note says what moved, what didn't, and the specific reason the bet isn't where it should be.
That reason is often the most useful part of the note, because a bet can fall behind for two very different reasons. Sometimes the work shipped and the market didn't respond. Other times the work never shipped at all: the landing page is still waiting for approval, a campaign stalled, or a task failed. A note that says "trial signups are 40 percent behind, and the construction ad campaign hasn't launched yet" points to a different decision than one that says "the campaign has run for four weeks and signups are still 40 percent behind." The first bet hasn't had a fair test. The second one has.
This weekly check is also how you know whether a pivot is working. A pivot changes who you sell to or what you sell, so its early signs show up in who arrives, not just how many. Kite watches whether the visitors, signups, and sales calls coming in match the new customer you're aiming for, and it calls out when the old audience still makes up most of the activity.
When the bet hits its targets early, Kite marks the goal as achieved on its own and tells you what moved.
Related use cases
- Find out which marketing channels are working Coming soon See which channels bring in customers and which only bring in traffic.
Make the call at each checkpoint
At each checkpoint, Kite compares the results with the criteria you agreed on and recommends one of three calls. It shows which criteria the bet met, which it missed, and by how much. It also tells you whether the bet got the fair test the criteria called for.
- Keep going when the bet meets its criteria. Kite continues the work and moves on to the next checkpoint.
- Change course when the evidence points to one fixable problem, such as strong trial signups but weak conversion to paid. Kite proposes the specific change, such as a new pricing offer or a different target role, along with revised criteria and a short deadline. Once you agree, it updates the goal and the weekly check so both measure the new plan.
- Stop when the bet misses the thresholds you set for stopping. Kite says so plainly, even when the bet has been running for months and the team likes it.
You make the decision. You're accountable for where the budget goes, and you may know things the numbers don't, such as a large customer who asked for exactly this product. If you stop the bet, Kite turns off its weekly check and any other scheduled work that belongs to it. It records the bet as stopped with the reason, rather than marking it achieved, so your goals list stays honest about what worked.
Related use cases
- Choose which marketing channels to invest in: Decide where to put your marketing time and money, including when a channel isn't worth it yet.
Record what you learned
A stopped bet still pays for itself if the lesson survives. Kite writes a short summary into your team's knowledge base: what you tried, the criteria, the results at each checkpoint, the call you made, and why. If the bet changed course, the summary says what changed and what happened after.
Kite reads that history before later work. If someone suggests a construction product again a year later, Kite can show what happened last time and which conditions would need to be different. If a lesson from the bet applies elsewhere, such as an ad message that worked well with a new audience, Kite can use it in the next campaign.
Related use cases
- Keep a record of what works in your marketing Coming soon Keep the results of campaigns and experiments where the next piece of work can use them.
- Reposition your company after a pivot Coming soon Rewrite your positioning when a pivot changes who you sell to or what you sell.
What you get
Kill criteria in writing
Each checkpoint has a number, a date, and the call that follows, along with the reasoning behind every threshold.
A live 90-day scorecard
A dashboard on the bet's goal shows every criterion with current numbers from your analytics, payments, CRM, and ad accounts.
Weekly progress notes
Each note says what moved and names the specific reason the bet is behind, including work that hasn't shipped yet.
A recommendation at each checkpoint
Kite recommends keeping the bet, changing course, or stopping it, and shows the evidence for that call.
A record of the result
Kite saves what you tried, what happened, and what you decided in your knowledge base, so later work can build on it.
How to get started
Tell Kite about the bet in plain language. You might say, "We're launching a version of our product for construction firms next month. Help us decide in 90 days whether to keep it," or "We moved into the UK in July. Is it working?" You can work with Kite in the Kite web app or in Slack.
Kite proposes the goal, the kill criteria for each checkpoint, and the first tasks. Change any threshold you disagree with, then approve. If the bet has already started, Kite can set criteria from today and use the results so far as the baseline.
Kite measures traffic and conversions on websites it hosts without any setup. For anything measured elsewhere, connect the products that hold the numbers, such as your analytics product, your payment processor, your CRM, or your ad accounts. Kite asks before it publishes pages, sends email, or spends money on the bet's behalf.
Integrations used to decide whether to keep or kill a new bet
Kite's built-in analytics cover the websites it hosts, and its research covers the market around the bet. Connect your own products when the numbers that decide the bet live there.
Measure interest and usage
Measure revenue and pipeline
Kite reads the customers, subscriptions, and charges that tell you whether the bet is bringing in paying customers.
Kite reads the contacts and deals the bet creates, so the scorecard can count booked calls and pipeline.
Kite reads the leads and opportunities that come from the bet when your sales team works in Salesforce.
Common questions
Can an AI marketing agent decide whether to keep or kill a new bet?
An AI marketing agent can do the measuring, the weekly checking, and the recommendation, and Kite does all three. The final call belongs to the person accountable for the budget. Kite holds the bet to the criteria you agreed on, tells you plainly when it has missed them, and explains the evidence. You decide, and Kite carries out the decision.
How long should I give a new bet before killing it?
Long enough for the bet to get a fair test, and no longer. For most marketing bets that's around 90 days, with checkpoints at 30 and 60 days. A new paid channel can often show its answer in four to six weeks, while a product with a long sales cycle may need two full sales cycles before revenue numbers mean anything. Set the window before you start, based on how long your buyers take to decide.
What's the difference between a goal and a kill criterion?
A goal says what success looks like. A kill criterion says what failure looks like and when you'll stop. You need both, because a bet can miss its goal without clearly failing. "Reach 10 paying customers by day 90" is a goal. "Fewer than 5 paying customers by day 90 means we stop" is a kill criterion. The space between them is where changing course makes sense.
What if the bet misses its kill criteria but the team still believes in it?
Ask what the team knows that the numbers don't show, and write it down. If there's a specific reason to expect a different result, such as a partnership about to launch, change course with new criteria and a short deadline. If the only reason is how much has already gone into it, that's the reason to stop. Kite records either decision with its reasoning, so the next review can see whether the belief was borne out.
Can I set kill criteria for a bet that's already running?
Yes. Kite reads the results so far, uses them as the baseline, and proposes criteria for the checkpoints that remain. You'll have less room to change course than if you'd set criteria on day one, but you'll still make the next decision on evidence.