Most new sellers change their campaigns too early. They see a keyword with clicks and no sales, decide it is a bad keyword, and turn it off. A week later they do the same thing to another keyword. After a month they have changed so many things that nothing in the account can be explained.
This audio is AI-generated with NotebookLM from this article.
The cause is almost never impatience about time. It is impatience about data. Three days is not too short a wait because three days is short; it is too short because three days has not produced enough clicks to tell you anything.
This guide gives you a number: how many clicks a keyword needs before “no sales” is real evidence, where that number comes from, and how to work out your own.
01Why does a keyword with no sales not mean the keyword is bad?
Your ad conversion rate is the share of ad clicks that turn into ad-attributed orders. If your ad conversion rate is 10%, then on average one click in ten becomes a sale. That average is the only thing 10% promises. It does not promise that the tenth click will be a sale.
With a 10% ad conversion rate, a run of clicks with no sale is completely ordinary. Nine clicks in a row with nothing is the expected experience some of the time, not a sign that anything is broken.
You can calculate how ordinary. For each click, the chance of no sale is 90%. For eight clicks in a row with no sale, multiply 0.90 by itself once per click.
- Clicks so far8
- Sales so far0
- Chance of no sale on any one click90%
- Chance of no sale across all 8 clicks43%
A keyword that is performing exactly as expected will show you zero sales after eight clicks roughly four times out of ten. If eight clicks is enough to make you turn a keyword off, you will turn off good keywords about 43% of the time.
02How many clicks do you need before zero sales means something?
You need enough clicks that a run of zeros becomes unlikely rather than ordinary. A workable standard is to wait until the chance of seeing zero sales from a normally performing keyword drops below 10%.
At that point, one of two things is true: either the keyword really is underperforming, or you have been unlucky in a way that happens less than one time in ten. That is a reasonable basis for a decision.
The number of clicks that takes depends entirely on your ad conversion rate. The lower your ad conversion rate, the longer a run of zeros can be before it means anything.
| Your ad conversion rate | Clicks with no sale before zero is meaningful |
|---|---|
| 20% | 11 |
| 15% | 15 |
| 10% | 22 |
| 8% | 28 |
| 5% | 45 |
| 3% | 76 |
If you do not know your ad conversion rate yet, use 22 clicks. It assumes a 10% ad conversion rate, which is a reasonable planning number for a product that is converting normally. Below 22 clicks on a single keyword, you do not have a result. You have a gap in the data.
Notice what the table does to the common advice that 10 clicks is enough. Ten clicks does not clear the threshold at any ad conversion rate in the table: even the 20% row asks for 11. For a product converting at 5%, acting on 10 clicks means acting on roughly a fifth of the evidence you need.
03How do you find your own ad conversion rate?
Your ad conversion rate is ad-attributed orders divided by clicks. It comes from the advertising console, not from Seller Central.
In Campaign Manager, set a date range of at least 30 days and read two columns for the ASIN you are advertising: clicks, and ad-attributed orders. Divide the second by the first. The console also displays a conversion rate column calculated the same way, but column names change over time, so the division is the version that will always work.
- Clicks in the last 30 days640
- Ad-attributed orders in the same period51
- Ad conversion rate, 51 divided by 6408.0%
- Row to use from the table above28 clicks
Do not use unit session percentage from Seller Central Business Reports for this. That figure divides orders by sessions on your listing, and it counts organic visits alongside ad clicks. It answers a different question — how well your listing converts overall. The threshold in this guide needs how well your ads convert, which is a different number on a different denominator.
Take the rate at account or campaign level, not from the keyword you are about to judge. A single keyword does not have enough clicks to produce a reliable rate, which is the reason you are reading this guide. The rate comes from the larger pool of clicks; the threshold then gets applied to the individual keyword.
One caution about the date range. Ad-attributed orders are counted after the click, so the most recent days in any window are still filling in. End your date range about a week before today, or the rate you calculate will come out too low.
If the ASIN has no advertising history at all, you have no measured rate. Use 10% as a planning assumption, take the 22-click row, and replace it with a measured figure once the campaign has a few hundred clicks behind it.
04Why is your campaign total not the number that matters?
This is the mistake that survives even after sellers learn the click threshold. They check the campaign, see hundreds of clicks, and conclude they have plenty of data. Then they make decisions about individual keywords using a number that belongs to the campaign.
A campaign total tells you nothing about any keyword inside it. Clicks are never spread evenly. One keyword usually takes a large share of the spend while several others sit on two or three clicks each.
A campaign with 1,200 clicks can easily contain eight keywords that have fewer than 22 clicks each. The campaign has enough data. The keywords do not. Decisions get made at the keyword level, so the count that matters is the keyword count.
Download your search term report and read the click column keyword by keyword. Any keyword below your threshold is not ready for a decision, whatever the campaign total says.
05How long does it take to collect enough clicks?
Long enough that it is worth planning for. Here is the arithmetic for a small launch budget on a campaign with ten keywords.
- Daily ad spend$12.00
- Average CPC$0.60
- Clicks per day across the campaign20
- Keywords in the campaign10
- Average clicks per keyword per day2
- Clicks needed on one keyword22
- Days until one keyword reaches that number11
At $12.00 per day and a $0.60 CPC, the campaign buys 20 clicks per day. Spread across 10 keywords that is 2 clicks per keyword per day, so a single keyword takes 11 days to reach 22 clicks. The even split is a simplification used to keep the example readable; in a real campaign the spread is uneven, which means your slowest keywords take longer than 11 days, not less.
Two things follow from this arithmetic, and both are more useful than any bidding tip.
The first is that fewer keywords per campaign produce usable data faster. The same $12.00 spread across 4 keywords gives each one 5 clicks per day and a decision in 5 days instead of 11. You are not buying more data by narrowing the campaign; you are concentrating it where you can read it.
The second is that cutting your budget when results look bad makes the problem worse. A smaller budget buys fewer clicks, which means a longer wait before you know anything, during which you keep spending.
06What should you do while you wait?
Waiting does not mean doing nothing. It means not changing the thing you are measuring.
- Write down the date you launched each campaign and the click threshold you are using for it.
- Check the search term report weekly, not daily, and look only at the click column to see which keywords are approaching the threshold.
- Act on keywords that have crossed the threshold. Leave the rest alone, including the ones that look terrible.
- Work on the parts of the listing that do not disturb the measurement: supplier questions, images awaiting replacement, stock planning.
- Record every change you do make, with the date. A change you cannot date is a change you cannot evaluate.
There is one exception to leaving campaigns alone. If a keyword has no impressions at all, waiting will not help, because no data is being collected in the first place. Impressions are the number of times your ad was displayed, and zero impressions is a different problem with different causes.
Frequently asked
Does this apply to spend as well as clicks?
Use clicks. Spend is a poor substitute because it mixes two different things: how many chances a keyword had to convert, and what each chance cost. A keyword that has spent $40.00 at a $4.00 CPC has had 10 chances, not 40.
What if a keyword has clicks and sales, but a bad ACOS?
That is a different decision and the threshold does not apply in the same way. Once a keyword has produced sales you are no longer asking whether it can convert; you are asking whether it converts profitably, which is a question about your margin.
Should I use 22 clicks for negative keywords too?
The same evidence standard applies, with one addition: relevance. A search term that is clearly unrelated to your product can be excluded immediately, because you are not making a statistical judgement, you are making a relevance judgement.
My product converts at 25%. Can I decide faster?
Yes. Higher ad conversion rates need fewer clicks, because a run of zeros becomes unlikely sooner. Extend the table using the same standard and you will find 25% needs 9 clicks.
How was the click threshold calculated?
Start by assuming the keyword converts at your ad conversion rate. Each click then has a chance of producing no sale equal to 100% minus that rate, and the chance that a whole run of clicks produces nothing is that figure multiplied by itself once per click. At a 10% ad conversion rate, a run of 21 clicks with no sale happens to a perfectly normal keyword 10.9% of the time, and a run of 22 clicks happens 9.8% of the time. The threshold is the first click count where that figure falls below 10%, which is the point where zero stops being an ordinary result and starts being evidence. Every row of the table is the same calculation at a different rate.
The choice of 10% as the cutoff is a convention, not a law of the platform. A stricter standard — waiting until the chance falls below 5% — would ask for 29 clicks instead of 22 at a 10% ad conversion rate. The looser standard is used here because it is reachable on a small launch budget while still being far more evidence than the eight or ten clicks most sellers act on.
Final thoughts
The question “how long should I wait?” has no answer, because time is not what you are collecting. Clicks are. Two campaigns launched on the same day can be four days apart in evidence, depending on budget and keyword count.
Replacing the question with “how many clicks does this keyword have?” changes what you do every week. It tells you which decisions are ready to make and which are not yet decisions at all. It also tells you something uncomfortable but useful early on: that a very small budget spread across many keywords does not buy slow learning, it buys no learning.
Episode transcript
Read the full transcript
Transcribed automatically from the recording. Speakers are not labelled.
Why is the calendar the wrong way to judge a campaign?
[00:00]You know, in almost every area of business, time is just kind of the ultimate undisputed metric of success, right? Absolutely. Right. Yeah, that's how we measure everything. Right. Like you launch a new initiative, maybe a new marketing push, and you just wait. You give it a week, maybe a month, and you judge the results based on what happened during that elapsed time.
Right. It's a clean, comforting way to measure progress. You look at the calendar and, well, the calendar tells you if you're winning or losing. Yeah, because we are entirely wired to track the passage of time and just expect a proportional amount of progress. Like if I work out for a month, I expect to see results.
Exactly. Or if I run a billboard ad for three weeks, I expect to see foot traffic. It feels universally intuitive. But then you step into the world of Amazon PPC advertising and suddenly that calendar becomes your absolute worst enemy. Oh, it really does. So whether you're an experienced Amazon seller prepping for your Q4 strategy meeting or maybe you're a beginner just trying your hardest not to burn through your entire launch budget, today's mission is to stop you from making the number one mistake in PPC advertising.
Because the mistake almost everyone makes is measuring their campaigns with a completely wrong instrument. Right. They are using a stopwatch when they should be using a click counter. A stopwatch instead of a click counter. I love that. So today we are doing a deep dive into this incredibly eye-opening article from AmazeBase. It's titled "How much data do you need before you change a campaign?" It's a great read. Yeah.
Is a keyword with zero sales a broken keyword?
Super practical. Yeah. And, okay, let's unpack this. The central problem the article highlights is that sellers are notoriously impatient. Very. You launch a campaign, you wait three days, you see absolutely zero sales, and you just panic. You think the sky is falling. Right, you go into the dashboard keywords to stop the bleeding. But the issue isn't that three days is a short amount of time. The issue is that three days hasn't produced enough clicks.
[01:54]Exactly. We are constantly tracking the calendar when we really should be tracking the data volume. And that is the core illusion of pay-per-click advertising. Time doesn't buy you data. Only clicks. Yeah, a keyword with zero sales isn't necessarily a broken keyword. In fact, zero sales is often the exact mathematical expectation for a perfectly good keyword that just hasn't been given enough chances yet. Okay, so to really wrap our heads around why we have to completely ignore the calendar, we first have to understand how normal statistical variation is basically tricking our brains into acting way too early.
How often does a good keyword show zero sales by chance?
Right. The math behind it is fascinating. Let's talk about the math behind a perfectly normal, healthy campaign. Okay. So let's assume you have an ad conversion rate of 10%. A solid 10%. Right. Human intuition says that if you have a 10% conversion rate, then out of 10 clicks, one will inevitably be a sale. Because one in 10 is 10%.
Exactly. And that's true on a macro, long-term, average level. But it guarantees absolutely nothing about the next sequential click. Right. A 10% conversion rate simply means that for every single individual click, you have a 90% chance of getting nothing. I always like to think about this, like rolling a 10-sided die. Oh, that's a good analogy. Yeah, imagine you have a die on your desk with 10 sides and only one side says sale. The other nine sides say no sale. Every time a shopper clicks your ad, you are rolling that die.
Yep. Now, if I roll that die eight times in a row, getting no sale every single time wouldn't even be that weird, right? It wouldn't be weird at all. It is completely ordinary. The AmazeBase article actually breaks down the exact probability on this.
Okay. If the chance of no sale on a single click is 90% and you get eight clicks in a row, you just multiply 0.90 by itself eight times. And what's the result? The result is roughly 43%. Wait, really? So if I have a perfectly healthy keyword that converts at a solid 10% and it gets eight clicks, there is a 43% chance it will show zero sales. Yeah, four times out of ten, a totally normal keyword will look like a complete failure at eight clicks. Wow. And here is the massive danger in that. If you are the type of seller who logs into campaign manager, sees a keyword with eight clicks and zero sales and you aggressively turn it off.
[04:16]Which happens all the time. Right. You are actively killing a perfectly good, normally functioning keyword almost half the time. Just throwing it away. You are sabotaging your own account because you are reacting to completely normal statistical variation. Exactly. You are reacting to noise thinking it's a signal. And this happens because digital marketing gives us real-time dashboards.
How many clicks do I need before zero sales means something?
We love our dashboards. We do. We see a cost accumulating in real time and we want real time validation. But probability doesn't care about your anxiety. So what does this all mean? Knowing that eight clicks is just normal bad luck doesn't really give me a game plan, though. I mean, I need to know where the actual safe zone is.
Sure. If a 43% chance of making a mistake is too high, where is the line? Do I have to wait for 15 clicks, 20 clicks? Well, the standard the article sets until the chance of seeing zero sales from a normal keyword drops below 10%.
Okay, 10%. Right, once that probability crosses below that 10% line, you finally have a solid foundation for a business decision. Okay, because at that point, you're pretty sure. Exactly, because at that point, one of two things is true. Either the keyword really is bad, or you are experiencing statistically rare bad luck. Like something that happens less than one time in 10. Right. Okay, I can live with the less than one in 10 chance of making the wrong call. But that threshold has to depend entirely on what your actual ad conversion rate is, right? Oh, absolutely. Like a product that converts at 5% is going to need a lot more clicks to prove it's broken than a product that converts at 20%. Yeah, the lower your ad conversion rate, the longer a run of zeros can be before it actually means anything. Makes sense. AmazeBase lays out a specific threshold table based on this math that honestly every seller should have sticky noted to their monitor. I love a good cheat sheet. Let's hear the numbers. Okay, so if your product has an amazing ad conversion rate of 20%, a run of zeros becomes statistically meaningful after 11 clicks. 11 clicks. Okay. If your conversion rate is 15%, you need 15 clicks.
[06:21]Right. If it's 10%, you need 22 clicks. 22? At an 8% conversion rate, it jumps to 28 clicks. Wow, it really scales up. It does. If you are selling a high ticket item, and your conversion rate is only 5%, you have to wait for 45 clicks. 45 clicks with zero sales.
Is the 10-click rule enough?
That takes some nerve. It does. And at a 3% conversion rate, you need a whopping 76 clicks before zero sales is actual evidence of a problem. Okay, I have to point out something crazy here. Yeah. If you spend any time on Amazon's seller forums or if you watch those internet guru videos on YouTube, the absolute most common piece of advice you hear is just kill a keyword if it hits 10 clicks and no sales. Oh, I hear that constantly. But looking at this math, 10 clicks doesn't clear the evidence threshold for any of the standard rates. Not a single one.
Not even 20 percent. No. Even at a stellar 20 percent conversion rate, 10 clicks isn't enough evidence. You need 11. That's wild. And if your product converts at 5 percent and you kill a keyword at 10 clicks, you're acting on roughly one-fifth of the evidence you actually need to make a statistically sound decision. You are basically just guessing and hoping you got lucky.
What if I have a new product with no conversion data?
Pretty much. But wait, what if you have a brand new product, like you just launched, you have zero historical data so you have absolutely no idea what your ad conversion rate is going to be. Well, what's fascinating here is that the math still protects you if you use a baseline assumption.
Okay, what's the baseline? The rule of thumb for a new launch is to assume a 10% conversion rate, which is a very reasonable average for a standard Amazon product. Okay, 10%. And based on that 10%, you use 22 clicks as your hard baseline. So 22 clicks is the magic number for a launch.
Which conversion rate should I use?
[08:12]Exactly. Until a keyword hits 22 clicks, anything less isn't a bad result. It's just a gap in your data. It's a gap in the data. I love that framing. So we have our threshold table. We know we need 22 clicks for a 10% converter Right, but applying this rule to an existing product requires you to know your true ad conversion rate Yes, it does and finding that number is where a lot of sellers completely derail themselves It's maddening because Amazon puts a metric called unit session percentage right there on the main business reports dashboard Oh, yeah, it literally says conversion rate right there.
Can I just use that? Nope, please don't. You should never use the unit session percentage from Seller Central business reports for your PPC decisions. Never. Never. It is answering an entirely different question about your business. Okay, how so? Unit session percentage divides your total orders by your total sessions. So it mixes your organic free traffic with your paid ad clicks.
Ah, I see. It exists to tell you how well your listing converts holistically, which is great for overall listing health or SEO, but it is fatal for PPC math. Right. Because it inflates the denominator with free traffic, making your ad performance look completely different than it actually is.
Exactly. Right. You have to pull this metric directly from campaign manager. Okay. Campaign manager. The formula you need is ad attributed orders divided by ad clicks. And you want to look at a period of at least 30 days to get a reliable average.
Okay. 30 days. And you must use the conversion rate at the campaign or account level, not the keyword level. Well, obviously not the keyword level because the individual keyword doesn't have enough data yet to produce a reliable rate. Exactly. I mean, that's the whole reason we are doing this exercise in the first place.
Why should my date range end a week before today?
Precisely. You have to look higher up. Right, we are using the macro data of the whole campaign data of the specific keyword. But wait, when I pull that 30 day report in Campaign Manager, I know there's a huge trap regarding the dates you select. Yes. This raises an important question about the reality of how Amazon attributes sales and why they delay presenting that data to you. The attribution lag. Right. Ad attributed orders are counted after the click happens, and there is a significant attribution lag.
[10:29]Like how long? Well, shoppers might click your ad on their phone on a Tuesday during their commute, leave the item in their cart, and finally check out on their laptop on a Saturday. Yeah, I do that all the time. We all do. Furthermore, Amazon has to clear the payment processing, which takes time. So if you include yesterday and the day before in your 30-day window, your calculated conversion rate is going to look artificially terrible.
Oh, it'll look awful. Because the clicks registered immediately, and you paid for them immediately, but the sales from those clicks haven't finished baking in the system yet. Exactly. To calculate a true ad conversion rate, you have to end your date range about a week before today. So cut off the last seven days. Right. A seven-day look-back window gives the data time to settle, ensuring your orders and your clicks are actually matched up. Okay. So let's say you've done the math correctly. You avoided the business reports trap. You accounted for the seven-day attribution lag. and you know your true ad conversion rate is 10%.
Should I count clicks per campaign or per keyword?
Great. So your magic number is 22 clicks. You are now armed and ready to dive into your campaigns and make some structural decisions. You'd think so. But knowing that 22-click rule actually creates a dangerous, false sense of security if you apply it to the wrong dashboard. This is the mistake that survives even after sellers learn the 22-click rule.
Right. They look at the wrong level of data entirely. Here's where it gets really interesting. Okay. Imagine a basketball coach looking at the post-game stat sheet and seeing the team has collectively missed 50 shots. Okay. He's furious about the lack of efficiency, so he storms over and benches a rookie who just stepped onto the court two minutes ago and hasn't even touched the ball yet. That's entirely unfair to the rookie. Exactly. He's punishing the individual player for the exhaustion and failure of the whole team. And that is a perfect representation of what sellers do in Campaign Manager.
They really do. They will open a campaign, see a total of 1200 clicks, and conclude that they have plenty of data to start optimizing and pausing things. 1200 clicks seems like a lot. It does. But clicks are rarely spread evenly across a campaign, because Amazon's algorithm aggressively funnels traffic to broad, historically proven search terms. Oh, so one or two massive keywords might eat up like 1150 of those clicks.
[12:43]Exactly. While the other eight long tail keywords are just sitting there starving with maybe three or four clicks each. Yeah, a campaign with 1,200 clicks can easily hide eight different keywords that don't have anywhere near enough data to judge. Wow. And because you are making the business decision to pause or adjust bids at the keyword level, the statistical sample size must be judged at the keyword level. You cannot fire the rookie for what the team did.
How do I reach the click threshold faster?
No, you can't. You have to ignore the campaign total entirely. You really do. The only way to do this correctly is to download your search term report and check the click column, row by row keyword by keyword. Row by row. Yep. If a specific keyword is below your 22-click threshold, it is not ready for a decision, no matter what the campaign total is at the top of the screen says. Okay, so if I'm listening to this right now and I realize I have to wait for 22 clicks on every single keyword before I can pause the losers, my natural follow-up question is about speed. is everything. Right? Waiting is incredibly expensive in the Amazon game. How do I get to 22 clicks faster? Do I just have to blindly double my daily budget to force the data through? Not necessarily. You can manipulate the speed of learning just by changing how you structure your campaigns. Okay, how so? Well the AmazeBase article walks through a fantastic mathematical breakdown of how budget, keyword count, and time all interact.
Let's run the numbers. Let's assume I have a pretty restrictive launch budget, say a daily ad spend of $12. Okay, $12. And let's say my average cost per click, my CPC is 60 cents. That means my $12 buys me exactly 20 clicks a day across that entire campaign. Right. And if you have 10 keywords inside that campaign those 20 clicks get divided up. Okay. That's an average of two clicks per keyword per day. Two clicks a day. Yeah. And if your threshold is 22 clicks and you are getting two clicks a day, it is going to take 11 agonizing days for a single keyword to reach the threshold where you could finally make a decision.
[14:52]And as we just talked about with the basketball analogy, the spread is never perfectly even. Never. An average of two clicks a day really means one dominant keyword gets 10 clicks and the others get 1 or 0. Exactly. Which means some of those keywords are going to take a month to hit the 22-click mark at that budget. But if we connect this to the bigger picture, you can speed up your learning without spending a single penny more by simply concentrating your budget.
Concentrating it, like how? Take that exact same $12 daily budget, buying the exact same 20 clicks. But instead of spreading it across 10 keywords, you launch a campaign with just four highly targeted keywords. Oh, I see. Now those 20 clicks are divided by four. Each keyword gets five clicks a day.
Five clicks a day. Right, and at five clicks a day, you hit your 22-click threshold in just five days instead of 11. Wow. You literally cut the waiting time in half without raising your budget, just by narrowing your focus. Exactly. You are buying the exact same amount of data, but you are concentrating it so you can read it faster. And understanding this mechanic also reveals the hidden danger of cutting your budgets in a panic. Which is such a psychological trap.
Why does cutting my budget slow down my learning?
It really is. When a launch starts slow and the results look terrible on day three, a seller's instinct is to slash the daily budget to stop the bleeding. Oh, I see it all the time. They say, "I'm losing money. Let me drop my daily budget from $20 down to $5 until things improve. But all that does is buy fewer clicks per day. Right. It stretches out that painful waiting period. The Amazon ad auction doesn't care that you're nervous. No, it doesn't. If you cut your budget so much that you are only buying one click a day for a keyword, you now have to wait 22 days before you have actual evidence to optimize it. 22 days.
What should I do while I wait for clicks?
[16:52]You're extending the bleeding out over weeks instead of ripping off the band aid to get statistical significance. So we've established that waiting for this evidence threshold is mathematically mandatory. Yes. We know we can concentrate our budget to make those clicks accumulate faster, but we still have to endure the wait.
You do. Practically speaking, what exactly should you be doing while those campaigns are cooking? Because sitting on your hands when your money is being spent is incredibly difficult. Well, you aren't doing nothing. You're exercising the discipline of measurement. Okay. What are the rules of engagement? The very first rule of engagement is to write down your launch date and write down the specific click threshold you calculated for that campaign. Just document it. Put it in a spreadsheet, put it on your whiteboard, whatever. Because if you don't anchor yourself to that number, the day-to-day emotions of spending money will convince you to break your own rules. Exactly. Second, record every single change you make to your campaigns along with a date. Always use a date. A change you can't date is a change you can't evaluate.
This is basic scientific method. You cannot isolate variables if you change bids, tweak your title, and update your main image all on different undocumented days. If you tweak a bid but don't know when you did it, you won't know which date belongs to the old bid and which belongs to the new one.
How often should I check the search term report?
Exactly. Okay, but what about checking the reports? If I am concentrating my budget and expecting a decision in five days, am I checking this every morning with my coffee? No, check the search term report weekly, not daily. Weekly. Daily checks just invite the temptation to tinker. When you do open the report, you look solely at the click column and you only act on the keywords that have explicitly crossed the threshold. And what about the exceptionally ugly ones? Like the keywords that have 15 clicks, zero sales and just look like a total disaster.
Can I pause a keyword because it spent too much?
Leave the ugly ones alone. Really? Yes. Until they hit that 22-click mark, they are statistically meaningless. They might feel like a disaster, but the math says they are just incomplete. Okay, let me push back on this, because I know exactly what sellers are screaming at their speakers right now. What if a keyword has only 10 clicks, so it hasn't hit the 22-click threshold, but my CPC is unusually high, say $4? Those 10 clicks just cost me $40, and I have zero sales to show for it. $40 feels like a lot of evidence. Surely I can pause that keyword because it's spending too much money. This is perhaps the most common misconception in all of Amazon advertising.
[19:16]Oh, really? Spend is a terrible substitute for clicks when you are judging performance. How so? Spend mixes two completely different things together. How many chances a keyword had to convert and what each of those chances cost? Ah, so the probability is based on the number of attempts, not the cost of the attempts.
Exactly. A keyword that spent $40 at a $4 CPC has only had 10 chances to make a sale. Right. 10 chances is still only 10 chances whether you paid $4 for them or 40 cents for them. Wow, yeah. The statistical probability of getting zero sales on 10 chances doesn't change just because the clicks were painfully expensive. It's still just 10 rolls of the die.
When should I act without waiting?
Right. The math of conversion simply does not care about your wallet. That is a harsh reality, but completely logical. It is. Is there ever a time when you shouldn't wait, like any exceptions where you break the rule and act immediately? The one major exception is zero impressions.
Zero impressions. Impressions are the number of times your ad was actually displayed on a shopper's screen. If a keyword is sitting there with zero impressions, waiting won't help you, because no data is being collected at all. Because you can't get clicks if you don't get impressions.
Exactly. You aren't even in the game. Zero impressions is a relevance issue, meaning Amazon doesn't think your product matches the keyword or it's a bid issue where your bid is too low to win the auction. Okay, so that requires immediate action. Yes. But if you are getting impressions and you are accumulating clicks, you have to wait for the threshold. Man, this entire approach is a massive paradigm shift. It really changes how you look at the console. Yeah, like if you are listening to this, the question you ask yourself when you open your advertising console is no longer, "How long should I wait to change a campaign?" Time is entirely irrelevant.
What is the one question that matters?
[21:09]Completely irrelevant. The only question that matters is, "How many clicks does this keyword have?" Accumulated evidence is the only metric of truth. And as we learned with the budget math, a small budget spread too thin across way too many keywords doesn't buy you slow learning. It buys you no learning Zero learning you are just spinning your wheels bleeding budget slowly and constantly resetting your campaigns before they ever have a chance to tell you the truth Yep, but now you have the statistical armor to defend your PPC decisions Whether you are running a massive seven figure agency or you are nervously monitoring your very first product launch.
Could my own changes be resetting Amazon's algorithm?
It gives you the confidence to let the math work for you rather than letting panic work against you. Exactly. And honestly, it actually leaves us with a really fascinating final thought to ponder about the Amazon ecosystem as a whole. Oh, where does this lead us? Well, if we as human sellers require a specific statistical volume of clicks rather than elapsed time to accurately judge a keyword's performance, what does that mean for Amazon's own A9 algorithm?
Oh wow. Right. The algorithm needs data, just like we do. It's a machine learning algorithm. It needs volume. Exactly. So every time we get impatient and tweak a comma in our listings title, or we shuffle our back end search terms just because it's been a few days and we haven't seen a jump in organic rank. Wait, are we constantly resetting the algorithm's evidence collection before it even had a chance to learn what our product is? We might be throwing away 50 clicks of Amazon's own machine learning data just because we felt like we needed to do something.
Wow. If the machine needs clicks to learn and we keep changing the variables before the clicks arrive, we are our own worst enemy. We are sabotaging our own organic rank by being impatient. That is such a huge takeaway. The calendar is a liar, only the clicks tell the truth. That's the bottom line. Keep accumulating that evidence. Until next time.