Compounding

00:00 Hello, and welcome back to How To Get Discovered. I'm Maya. And I'm Tom. This is episode 9, and it's called Compounding. It's about what actually happens to a back catalogue once people can find it, over weeks, then months, then years, and why some episodes keep earning forever while others are finished in a week. It's also the episode where Maya pays up. Reluctantly. Last week, at the end of all that business about crawlers and C-names, I set her a challenge. Show me, honestly, what the first month looks like for an ordinary show. Not the success story, not the one somebody screenshots.

00:41 The ordinary one. And I said yes, which I've regretted slightly ever since. Good. So that's the first half. The second half is why the curve looks the way it does and why most podcasters give up right before it gets interesting. Go on then. Ordinary show. I've invented one, so nobody's numbers get dragged into it. It's a weekly show about growing your own veg. Two friends who share an allotment, three years in, about 140 episodes. Audience? A few hundred regulars. Lovely people, very loyal, mostly in their 50s, a lot of opinions about slugs.

01:22 I'd listen to that. You would, and you'd never have found it. So this week they do the dull bit. The whole back catalogue gets transcribed. Every episode gets its own page with chapters, and those pages go live under the show's own website. Start the clock. First week, almost nothing. The pages exist. Nobody's visited them except the two hosts, who've each looked at the tomato episode about nine times. And the next? Google turns up. Some of the pages get crawled, a few get indexed. If you checked, you'd see that a search engine now knows the show has an episode about Blight.

02:00 That's it. No visitors to speak of. After that? By week three, they've had a handful of search impressions. Meaning their pages appeared in somebody's results a few times, mostly near the bottom, mostly for very long, very particular searches. One person clicks through. And there's one visit from Perplexity.ai, which they get terribly excited about and which might have been a fluke. So, honestly, slow and slightly disappointing. Yes, that's the answer to your challenge. Three weeks in, that's what it looks like. And anyone who tells you otherwise is selling a course.

02:39 I was braced for a twist. There isn't one at three weeks. The twist comes later, and it's quieter than people want it to be. Here's my problem with that, though. If I'm one of the allotment pair, week three is the week I stop. I've done a load of work, I've got one click in a mystery visitor, and I've got a show to make. Which is exactly when most people do stop. So it's worth understanding why it's flat, because it isn't flat for a mysterious reason. Go on. A new page is a stranger to Google. It doesn't know if it's any good. So it seems to try the page out low down, for odd little searches, and see how it does.

03:20 Nobody outside Google knows exactly what it's watching. That testing takes weeks and it happens page by page. Hang on, though. You said last week a sitemap tells Google about every page. Sure, and it'll probably come and read them within days. Trusting them is a different matter. It's like a new neighbour. You'll know their name in a day. You won't lend them your car for months. So all 140 pages are on probation. Yes, and the site as a whole is on probation too. A brand new section of a website with no history doesn't get much benefit of the doubt.

03:57 That's the domain authority thing we talked about last week. It builds slowly, and it builds from the pages that do well, pulling the rest up a bit. And the AI assistants? Slower still, mostly. They lean on what's already been crawled and trusted, so they tend to show up after search does, not before. That perplexity visit in week three is a bit of a gift. Right. So when does it bend? Let me keep walking. Month two. The impressions creep up. Not dramatically, but the searches start to look interesting. They're not getting found for Vegetable Gardening Podcast, which is what they'd have guessed.

04:37 They're getting found for why are my courgettes going yellow at the end? And Can You Plant Garlic in January? Nobody types that into a podcast app. And every one of them is somebody with an exact problem. Each of those pages is only getting a trickle. Three visits here, two there, but there are 140 of them, and they're all trickling at once. at once. And after that? Somewhere around month three or four is usually where you'd see it, if you were going to. A few pages have passed their probation and moved up. The garlic episode starts turning up properly every autumn.

05:14 The Yellow Courgett's one gets a steady drip every week of the summer. And the site started to carry a bit of weight, so new episodes get indexed faster and tried higher up. Why so sudden, though? If it's all a trickle, why isn't it a gentle slope? Because hardly anyone reads past the first page of results. A page that climbs from 30th to 12th has improved enormously, and it gets no more visits at all because it's still on page 2. Then it nudges into the top 10, and suddenly people can see it. So months of real progress look like nothing, and then one small step looks like a miracle. So the flat bit isn't flat. It's just happening where nobody's looking.

05:58 That's the best description of it I've heard. And it's happening to dozens of pages at slightly different speeds, so they come over the edge one at a time. That's your bend. It is. And what I find genuinely odd about it is this: a normal episode is front-loaded. It comes out, the regulars listen in the first week or two, and then it's basically done. Almost everything it'll ever get, it gets at the start. Sure. That's how every download chart works. A searchable episode is backloaded. Its first month is nearly nothing, and then it earns a bit, and a bit more, for as long as people keep asking the question.

06:38 So a two-year-old episode can end up quietly doing more in a month than it did the month it came out. Can? Or does? Can. Some of them, not most. I'd rather undersell this than have someone at home feeling cheated at month two. I'll be honest. The bit that bothers me isn't the allotment pair. It's me. Why? Because I've made something like 350 episodes over 10 years, and I've always done the maths as hours in, downloads in the first fortnight out. By that maths, loads of them didn't pay. If even a handful could have kept earning for years, then I've been undercounting my own work for a decade.

07:21 That's a horrible feeling. I had a version of it. And? Angry at myself for about a week, then angry at podcasting for about another week. Then it turned into something useful, which is that the episodes are still there. They didn't expire. The launch isn't the whole return, it's the deposit. That's a very marketing sentence. It's also true. Here's my real objection. In month two, how do the allotment pair know any of this is happening? From where they're sitting, a trickle and nothing look exactly the same. They look at two things: which pages people are landing on and where those people came from.

08:03 Start with the first one. That tells you which episodes strangers actually want. That's often not the ones you're proudest of. The allotment pair might think their best work is the interview with the seed historian, and the page that keeps getting visits is 12 minutes on slugs. Of course it is. And the referrers tell you how they got there. Google, mostly. But then chatgpt.com shows up, and perplexity.ai, which is how you'd know the AI stuff from a few weeks back is real and not wishful thinking. Then a Reddit thread where someone asked about clay soil and somebody linked the episode. Then somebody's newsletter, which you'd never have known about otherwise. That last one's the one I'd want to find.

08:46 A person going out of their way to recommend you. So would I. If they're on PodHerd, its analytics show them which pages people land on and the sites that sent them, on every tier. The advanced analytics, the search insights, and the Google Search Console integration come in from Grow the Herd Up. Search Console's the one I'd actually open. Why that one? Because it's the only bit that shows you the near-misses. In Search Console, you'd see the searches where your page turned up and nobody clicked. If the slug episode is appearing every day for How to Get Rid of Slugs Without Pellets and nobody's clicking, that's not a search problem.

09:27 That's a title problem. That's exactly it. It's the most useful view in the whole thing, and it's the one people skip. Don't sound so surprised. I'm moved, actually. You've gone from not knowing what a C-name is to diagnosing click-through rates. I've been forced to sit through eight episodes of it. How often would you look, though? That's the other trap. Every morning, if it were mine. With a cost. coffee and a sense of dread. Which is how you go slowly mad. Daily numbers on a trickle are just noise. One day the slug page gets six visits, the next day none, and you start rewriting titles in a panic.

10:09 Once a month is plenty. And the garlic episode will look dead from February to August because nobody plants garlic in June. So seasonal shows have to wait a whole year to see if anything's working. For some pages, yes. Which is one more reason the first month tells you so little. There's another way of looking at it that I like. Instead of asking how the show did this month, you ask how the episodes from a particular year are doing now. All the year one episodes, as a group. Cohorts. Exactly. For a show nobody can find, the answer for year one is basically zero. They had their fortnight and they're done. For a show that's findable, year one is still doing something.

10:54 Less than the new stuff, falling slowly, but not zero. And when you add up every year's not zero, that's the compounding. It's not one episode going viral. It's the whole catalogue earning a little bit all the time. Which is less exciting than the word suggests. Much. It's a savings account, not a lottery ticket. But not every episode's in the savings account. Some of mine are two blokes reacting to the news. And those die in a week. And they're meant to. Nobody in six months is searching for your take on that week's row. Some of my best episodes were reactions, though. I'm sure. And they did their job in the moment for the people who wanted you on the moment.

11:38 I'm not saying stop making them. I'm saying know what shape they are. If the entire feed is reactions, you haven't got a back catalogue, you've got a pile of old newspapers. So what's the difference, in practice, between the ones that earn and the ones that don't? They answer a question people keep asking. Garlic in January gets asked every January. My Soul Trader Tax episode, which I've gone on about enough this season, gets asked every time somebody's first tax bill turns up. Neither of those questions is ever going away. And the ones that die are answering a question nobody asks twice. Or nothing at all. Two people chatting nicely isn't a question.

12:21 Is that the only way an episode dies? No. Some die because nobody can tell what they're about. Remember your old titles? Episode 63, June. I was hoping you'd forgotten. Never. An episode called that might have 40 brilliant minutes on swarm season, and it still dies in a week, because the title, the description, and the page give a stranger nothing to go on. Chapters and a decent transcript can rescue an episode like that. A good title rescues it faster. Now this is where I get twitchy. Because the obvious lesson is only make episodes that answer searches.

13:02 And then the allotment show turns into 40 identical how-tos and loses everything that made the regulars love it. Agreed. And I'd hate that. That's the bloggers with microphones thing we worried about right at the start. The allotment pair should keep chatting nicely. Most weeks, that's the show. The point is just that the how-to bits buried inside those chats are often the ones that'll still be working in five years, so it's worth them being findable. Fine. I can live with that. Can I add something else? Because I think there's a third kind you're leaving out. Go on. Everything you've described depends on somebody typing a search. The question has to exist and keep existing. But clips don't work like that.

13:46 Somebody shares a minute of my show because it made them laugh, and it goes round a group chat. Nobody searched for anything. The trigger was a person. Yes. Which means a clip from year one can go round a group chat in year four. Not because the question's still being asked, but because the moment's still funny. Or still true. That's really good. And it's the bit of all this I'd hang on to hardest. Search Lives or Dies by Google and whatever the AI assistants decide to do next year. Both of those can change their minds overnight. A person sending another person a clip doesn't depend on any of it. It's the one channel nobody can switch off. Pretty much. And it's why the clips have to point somewhere.

14:31 On PodHerd, clips come out captioned in the tall and wide shapes linking back to the episode on every tier. So a clip that travels doesn't strand someone. It leads them home. And the shared answers. From the loyalty episode. Same idea, honestly. When a listener asks ask. the Archive something and shares the answer, it gets its own link. It's kept out of Google, it goes from one person to the one person they thought needed it, and it plays the exact moment that answers them. Ask the Archive is on Grow the Herd, with a limited number of questions, and unlimited on Lead the Herd. Which you sold me on as a loyalty thing back then. And it still is.

15:13 But it compounds too, because the person who receives it didn't have to search for anything. Someone who already loves the show did the finding for them. So the regulars are the compounding. A big part of it, yes. That's been your argument since episode one, and I think this is where it wins. I'd like that on the record. It's a podcast, Tom. It's all on the record. Then I'll give you one back. I came into this expecting you to show me a hockey stick, and I'd have argued with a hockey stick. A dull, flat month that might bend later. That, I believe. That's the nicest thing you've ever said to me. It was about a graph.

15:55 Next week's the last one of the season. What happens if you do nothing? Two identical shows over the same 12 months. One does the work, one doesn't. And we follow them month by month and try very hard to separate the honest bits from the hopeful ones. And we each have to say one thing we got wrong this season. Which was your idea? I'm already regretting it. Good. Thanks for listening to How to Get Discovered. See you next week for the finale. Bye.