Better with Humans Podcast

AI SEO Strategy That Wins Niche Leads

Watch the Better with Humans Podcast on YouTube
Watch and listen to the Better with Humans Podcast on Spotify
Watch and listen to the Better with Humans Podcast on Apple Podcasts
Listen to the Better with Humans Podcast on Amazon Music

 

Agencies are losing clients to AI right now, just not in the way you would assume. Cheryl Luzet, founder of Wagada Digital, hasn't lost a single one, and her AI SEO strategy explains why. If what you hand over still carries real human expertise, the thought that a client could have done it themselves never occurs to them.

The brief that came out of a chatbot

Cheryl's team now gets enquiries where the brief was clearly written by an AI tool. The business has asked what it should be doing, and the answer is everything it could possibly do. So they arrive wanting all of it, with a budget for a fraction of it. Somebody still has to sit down and work out which single thing moves the needle first. That is not a prompting problem. That is expertise, and a conversation, and someone willing to say that isn't where your money should go.

Why an AI SEO strategy still needs a human in the room

One of Cheryl's team queried an AI analysis that had labelled Irish organic traffic as direct. The reason given was simply that it came from Ireland. Pushed once, the model folded and admitted it was wrong. I had the same experience over a typo it argued about before conceding. These tools want to give you an answer more than they want to be right, and marketing lives or dies on testing and iteration. Insight is the bit a human still brings.

Appearing in AI search, not just Google

This is where it gets genuinely interesting. Traffic patterns are shifting, people are having conversations rather than typing keywords, and the AI overview often answers before a single site gets clicked. Cheryl's team now tracks which prompts people use in a sector, where competitors appear and they don't, and how sentiment reads across reviews, digital PR and other channels. Websites still matter. They are no longer the whole picture.

Niche case studies are doing the heavy lifting

Wagada gets real leads from AI tools, and the pattern is specific. Someone asks for the best marketing agency for a cyber security company, or for the nursery sector, and the case studies on the site do the work. Narrow beats broad. There is also a trust shift happening here. A list of ten blue links feels like options. A recommendation feels like advice from a friend who knows you.

The thread running through all of it is authenticity. The agencies struggling are the ones passing on output anyone could generate. The ones doing fine are the ones you can tell a person touched. Worth asking yourself which pile your last deliverable landed in.

Connect with Cheryl Luzet:

LinkedIn
Website
Check out Cheryl's AI optimisation checklist

More conversations like this

This was made with humans.

Your host - Shaun Phillips

Editor: Glen Boswell
astralforgefilm.com
Instagram

Original music composed and produced by Brokli

Original logo illustration and in video podcast graphics by Jesse Rist

Follow us for episode clips on Instagram

Follow us on TikTok

Want to watch episodes instead? Subscribe on YouTube

Connect with Shaun on LinkedIn

Introduction

0:05 agents are losing clients to ai right now just not in the way you'd assume the client gets to work back reads it and thinks i could have done that myself and they're right it came out the same tool they've already got open my guest today runs a digital marketing agency and hasn't lost a single client that way her explanation is simple if what you hand over still carries real human expertise and authenticity that thought never occurs to them my guest today is cheryl luzet

Meet Cheryl Luzet

0:36 cheryl and i first met through business network internet organs around 15 years ago cheryl is the founder of black at a digital an award-winning uk digital marketing agency helping ambitious businesses grow through smarter marketing technology and strategy with extensive experience in digital marketing and business leadership cheryl is passionate about helping organisations now navigate an increasingly complex digital landscape particularly as artificial intelligence

1:03 transforms the way businesses connect with their customers so cheryl it's so it's so great to have you on have you on the show oh brilliant well yeah thanks for having me here shaun it's good to be here i mean i think there's so many just in your bio there's so many things we could we could uh we could jump on but i think maybe a good question to start with would be how have you how have you seen your marketing change in terms of ai and what's the impact of that on your business

Marketing changed more in 18 months than in 10 years

1:29 what what has been happening yeah yeah i think it's i think marketing has changed more quickly than um you know in the past year or so in the past year and a half than it has probably in the past 10 years been my background is very much in search engine optimisation and so i'm kind of used to working in a world where uh things are changing all the time and where you have to keep on top of changes so so i initially set the business up as a search engine optimisation agency and over the

1:57 years we have become more of a full service enterprise and i've been working on a lot of marketing agency um and actually we've we've kept our um kept our roots in seo so i think for us as a business the changes haven't come as much of a surprise as maybe more traditional marketing agencies um but yes it's been it's been really exciting times how how customers are looking for their services and products now how that is changing how that is evolving the tools that

2:24 they're using and their abilities to do potentially some of the work themselves um and also one thing

Clients arriving with AI-written briefs

2:31 we've definitely seen as a marketing agency is when um companies are approaching us they're quite often um coming to us with a brief that's come out of an ai tool of some sort so that creates a a new challenge for us where people are asking for services and they don't really know what it is that they're asking so we have to start by unpicking that um to support clients to get what they need yeah thank you yeah i can i can imagine that that that coming with a brief that like yes this is what

3:01 i need yes yes because ai has told me yeah and they probably put one line of prompting it's one of the things we comes up a lot in the podcast is actually if we're going to use ai we need to use give ai all the information that it needs but as you just pointed out if i'm not an seo expert how do i know what information i need to give to ai i mean i could ask ai and it could give me some answers but i think that's one of the things that i've been doing for a long time and i've been doing for a long time and i think that's one of the things that we are seeing is that

Knowledge is not expertise

3:26 there might be knowledge we can put into ai but it doesn't replace expertise no i think that's really true um and coming back to these briefs that we've been getting people are asking ai what they should be doing for their business and ai is telling them absolutely everything they could possibly do so that business is saying well we must do all of these things then but they don't necessarily have the budget to do all of those things so you know it comes back to

3:52 that expertise of us knowing exactly what we're doing and what we're doing and what we're doing what they should be focusing on rather than trying to split their budget across every single tactic that you could possibly work on yeah and i guess there's also that human element of where you're able to have a conversation about what's actually the biggest problem what is the best bang for the buck what's the one thing you could do first of all that's going to give you the most opportunity to generate more traffic generate more sales as opposed to as you say

4:20 just pick the first thing on the list because that's why i came out with yeah no you're absolutely right i think that's where we definitely still need humans you know in working out what are the best tactics to to do and how to actually execute those tactics and then that point we can then bring ai into support with with some of the ideation around that but until you've got to that point it's quite tricky to to rely on what ai is giving you so in terms of so

4:47 if there's um people that are actually watching it watching or listening to this podcast and wondering what it means to them to be able to do that then i think that's where we need to be able to do that and i think that's where we need to be able to do that and i think that's where we need to for their you know marketing agency so what are some maybe some advice you can give to give to them in terms of how do they adapt what sort of things do they need to be addressing yeah that's

Why some agencies are losing clients to AI

5:05 a really good question i think there are a lot of agencies that feel a bit threatened by ai and i know there are definitely agencies that are losing work through because of ai because their clients are saying i'm going to do the work themselves um it's quite interesting we haven't experienced that at all we've not lost any of that and i think that's a really good question because the the person the company wants to do the work themselves and i think potentially that

5:31 might be because we're still using quite a lot of human in the work that we deliver to our clients i think where where some marketing agencies are using ai and giving ai generated content for example to their clients their clients are seeing that's generated by ai and saying to themselves well i could do that myself whereas if you're making sure that whatever you deliver to the client still has very much that human side to it that you're delivering to the client that engagement that um expertise then that client can easily see that they can't do it for

6:01 themselves so i think we're very much moving from a world where um marketing agencies focus on executing tactics to really very much um the focus being more around the strategic elements and the storytelling and you know embedding the human and authenticity in into the brand and the work that we're doing for our clients yeah that's really good it makes me think about again this human aspect of it that that what i

6:31 find this if i'm using ai to ask me questions there's no accountability there's no i'm going to tell it what i what i want to tell it there's nobody necessarily coming back and you know querying me on that and saying well particularly if i've got a relationship like i'm you know if i'm a client of wagada and i'm going oh da da da and you're like hold on a minute but six months ago you were saying this if you change that oh no i'm not going to tell you what i want to tell you no i haven't changed that that's actually still still the case and you'll go well that's not then

6:59 that isn't going to be quite right the thing you want to do and i think there's a danger there that we are filling ai with the things that we want to maybe believe about our business but we're not getting the challenge from a human to say well is is that actually accurate or not and as you say storytelling yeah we can get ai to help us write a story but is that story going to be relevant to to to my business is it going to cause more damage

7:24 than actual value yeah no absolutely and the whole piece the whole important thing about marketing is is the kind of testing and iteration and ai is not necessarily doing that so you know when every marketing tactic you should be doing you should be testing if it's working or not working not just talking to ai and getting ai to decide whether that's the best tactic or not and that's where i think the the insight that you get from a human being is is really really valuable

7:50 yeah that's a good point i mean one area that ai could be valuable in is the analysis of those but it's it but actually it's got to know first of all you've got to know what are you doing what are the variations that you want to test and what what outcomes you want to get before you say and then get ai to analyse and give you those results definitely because we definitely do use ai to support us to gain some of the insight from our clients because it can be

8:17 really handy for um you know looking at large quantities of data looking at what's working and what's not working but you do have to really look at what's not working and what's not working and

When AI confidently gets it wrong

8:24 really question that and i've got a really good example from this week actually is one of the team was looking at the results from a client and ai was basically saying that any um traffic any website traffic that came from ireland it was saying that that was um any sort of organic traffic that came from ireland it was saying it was direct traffic and the member of staff queried it and said why why are you calling that organic traffic direct traffic and he said because it's

8:52 from ireland and the person said well why because it's from ireland is it is it direct traffic and not organic traffic and and claude said oh no you're right i'm completely wrong and just completely backtracked so it makes you realise you do you do have to you know you have to question it it's like it's like humans in that way we we get it wrong you know claude gets it wrong you do have to be on the ball really and identify where where something might not be true that um that an ai tool thinks it is yeah absolutely

9:20 and there was the podcast that went out this episode that went out this week with peter facey he talks about where he's in a room full of students and he asked them do you believe everything that chat gpt you know gives you and every single one put their hand up and so yeah that that's a that's a kind of a shock isn't it because not everybody necessarily understands that ai is is giving you the most probable outcome you know what it leaves

9:47 statistically is the best and it might be 85 99 but it's never 100 if you ever query it and say oh well no i'm not or if you go and ask it to check its own work it will give you a different answer yes yeah every time it'll give you a different answer which is really frustrating and i've got an example once where um ai wrote a typo in some work and i said i don't think you spelt that word correctly and it argued with me a little bit and i said no no i really think you

10:15 spelt that word wrong and after a while it said yes you're right i'm sorry i did spell that word wrong but it was it was determined to be right but it wasn't well yeah and i think this is the the challenge we have and it's come a few times in the podcast is that when we're feeding information we make a request it's really important we make the request in a way that doesn't deliver any bias into the request it's clear about what it is that we want to achieve what is the output that we want to give it and what some of the constraints that it has

10:42 so we get the best possible output because once we it's we've given it that request it wants to deliver the request its priority above everything else is giving you the answer and it will find a answer is wrong but it thinks that statistically that's the right answer then it then as you say you're going to get situations where it's going to argue and say no no no no no and until until you actually eventually uh it picks up some other information and goes oh actually that was

11:09 incorrect or even even some of the people use it a lot for copy don't they and and i i've been using some stuff today i've been working on some things today and i don't know i thought that it's not right and i couldn't i don't i'm not a copywriter so i couldn't tell you what is wrong about it i was like reading that just is weird it doesn't say and then if i say to ai can you just go and read that back you know as if you're like a potential client that you read it back and they are yeah that's weird yeah yes yeah i think we do have way too much trust sometimes in these

11:40 ai tools um in that they're just going to deliver exactly um the truth to us and it's not yeah absolutely and i think one of the things that we can look at in terms of

Building checkpoints into your AI work

11:54 in terms of that is what sort of checkpoints do we put in so what sort of things do you do you within your organisation when you're using ai how do you ensure that you're not something that's coming in doesn't end up going out and that is incorrect or is going to be uh not what was intended yeah i think that's a really good shout i think one of the things that i quite often have learned to do and i think you mentioned it just now is um when ai gives you an answer is to put

12:23 the answer back in and then you can do it again and then you can do it again and then you can do it into ai and then it'll it can easily then question itself and then you just keep on putting it back and putting it back until it's run out of questions and i've been doing it i've been using it a lot lately actually for looking at contracts because i find it really useful for pulling apart contracts and what issues you might be be seeing in those contracts um and it will identify some of the issues but then i know that i have to fix those issues and then put the contract back in

12:49 and i keep doing that until it's until it's exhausted itself with all the possible issues that it might have found yeah yeah that's that's really good i think it's a people don't necessarily realise that it's almost as if when you create a new chat and this is something when people will say i chat gpt knows me so well but every time you create a new chat you're having you're making a new connection to a to a to almost to a new chat gpt right each time you do a chat

13:16 which is why when you ask it again it's it's pulling on information memory it's not it's not like a a human where it's not like a human where it's not like a human where it's not like a human where we are having a conversation and it's all going i'm gathering more information about that it's starting off something fresh and i think i think the challenge and the challenge with that is that we think it's going to give us the same answer so as you say for you what you're saying

13:43 you can do that you can get it to ask that but sometimes i've also seen where it then gives you a complete different answer or it loses something yes yeah it contradicts itself yeah it's got missed out on something and sometimes i mean i do like quite a lot of coding with with ai and i find if i'm using frameworks which i use built-in frameworks to do it it actually does it has a policy where it goes through but it will do a test and then it will do another test and then eventually you'll do it and it'll do a a test run and then it'll go oh

14:12 actually no that's not right that that should be done that that way and it will go off and it will fix it and it'll iterate it iterate it but sometimes as i've as i mentioned you do also lose something you know what happened to that thing in the specification that i mentioned now we're talking about some of the challenges some of the things that

The real opportunity is in the data

14:31 we can use it for what are the some of the things that may be kind of opportunities that excite you about what ai could be capable of or maybe it could do in the future yeah that's a good question um i think i think the whole kind of data side of ai is really exciting to me is how we can gain so much insight by putting data in and um give us actionable

14:57 insights that we can then use to move forward with and i think we're not probably not very good at that right now and i think we need to understand how to get more data out and then you know using ai to to drill through that data to get more more of the insight that i think we need and i think in the past probably we didn't seek out that much data because it would have been too much for a human to to absorb but we can we can do that now you know and i think we need to find ways to to

15:27 really um you know make make most advantage of that yeah absolutely it reminds me of i mean i technically i understand about the tracking pixels in facebook and the google analytics and but but one thing is setting all those things up and it for instance tracking into when what happens if somebody clicks here or an email comes here they click on that and it goes here to track all those things it's even if you set it up so that it can end up also being an awful

15:56 lot of data to determine well okay how do i what does this tell me what does it actually tell me what can i use for it and as you say there is but i'm denying that there's even more data yeah and i think that's really that's really interesting to think about how do we actually get it to what analyse the data but do it in a way that is that is accurate and in a way that's actually going to really going to support us and i think it's i mean it's no surprise coming from an seo background that

16:21 you know you would that would be the thing you come with right you know data is is is

When agents start talking to agents

16:27 so huge and one of the areas that i'd be interested to get your thoughts on is what does it look like in a world that seems to be heading that way in probably much sooner than later where we're actually we've got agents running on our websites and it's the agents talking to the agents so so we're not having the human going onto a website we're having the humans agent in chat gpt having a conversation about you know negotiating some insurance or

16:55 whatever that service might be i think that's quite an interesting proposition actually because we know now that when we're talking to um you know a chatbot on a website you quite often don't get you know that human response you don't you don't you don't get that kind of um like the empathy that you need from from when you're talking to a chatbot and you don't get the sense so actually when you've got two chatbots talking to each other you're even one more step removed from actually gaining sense of

17:26 what's going on so i think that's a really interesting proposition i think that's a really interesting proposition i think that's a really interesting proposition so whilst all this technology is is moving forward really quickly i do think that if we do get to the point where agents are talking to agents i think we're going to lose a lot in the sense of of what is actually happening you know and what what what the actual answer that you're looking for i think we're going to lose that quite a lot um i think there's there's potential issues with that yeah yeah i

17:52 think so too and again it comes back to what you said as well data because it's all well and good if the if the data isn't there i mean your your your yeah your agent on that website is going to need to know everything not just what's on the website but it's going to need to understand what's your sales process it's going to understand understand so many things about the business values if it's having a conversation what what um people talk of agents and having agencies so what

18:20 what is it allowed to do you know what information can it share what prices can it share if it can it offer to the client what can it offer to the client what can it offer to the client what can it offer discounts how does it so there's a and i think i think it's the danger is the challenge is that it probably is going to happen sooner than than we're ready for because it's really the capability pretty much exists existing where we are now and it's going to be and i don't think and but it is

18:46 one of those things it doesn't seem like it would be easy to develop piecemeal so in other words if somebody does that on the website okay great but unless somebody is talking about it talking as an agent that's that's going to do that and if even if you are doing it and not every website responds to that so i think i think it's going to be an interesting challenge but i think again it's one of those things we're going to need to navigate yes yeah i had a really

The AI answering machine at the doctors

19:14 interesting experience actually ringing my doctors recently and they've now got an ai answering machine thing going on and um it kind of tries to get you to tell to tell it what the what the health problem is so you tell it what your health problem is and then it tries to guide you and and help you in the right direction so it's quite it's quite dangerous in some ways um so i tell it what my problem is which isn't the simple thing because it's not necessarily i am ill

19:41 it was i am ill and this has happened and i need to talk to this person and the lovely advice it said to me is that i should just go and look up on the website for some advice on how to tackle this and actually got a bit annoyed and i actually i have to admit i swore but just thinking i was swearing at an answering machine but the the chatbot actually was was was a feeling chatbot and it got quite upset and he said oh i'm really sorry that you feel like that and then i was really embarrassed because i don't know what if someone's recording this call so then i put the phone down and then he started sending me

20:09 text messages saying oh i can see that you've been upset and you put the phone down can we help you can you call back but it just made me feel really guilty it made me feel like a bad person yeah but i mean it's interesting because ai doesn't do empathy but it can simulate yes it can simulate oh actually yeah there's yeah there's something going on actually there's i think i read it in the past week that things that these some of these companies are

20:39 like like anthropic that you clawed in an opening eye for chat gpt are starting to put in boundaries in terms of like if you abuse so they're putting in in things to stop people abusing they're abusing the agents now i don't and it's a kind of a it's kind of a strange thing but i think it's more to do with actually the way that ai is going to develop and grow is going to be how we feed it how we model

21:07 rather than they're worried about ai being upset or ai it's i think it's more to do with but they're framing it as if are we you know we don't people abusing abusing but i think it's more to do with we need to model the values that we want ai to develop if we want if we were if we're concerned about how ai is going to view humans yeah in the longer term yeah so i think i didn't feel that was empathetic you know i i felt it was a bit contrived this this computer that was

21:37 trying to be sympathetic with me it didn't it didn't wash with me i have to be honest it didn't feel very comfortable yeah i mean it sounds kind of a little bit scary as well messaging you later going oh you know you see those horror movies don't you like where the phone rings and they put the phone down it's not i don't see it runs me war games from the 80s where the guy like you know he hacks in hacks into the uh as into the pentagon and then the computer calls him back

22:04 and he keeps calling him he's trying to they're playing for bits the phone down it rings and it's the it's the uh old modem noise because the computer's kind of calling you back again so i think that's that's um yeah that is going to be interesting i think when we start to get obviously we already have spam type calls that are automated but when we come to a point where businesses or try to send you in to have a conversation with a an artificial intelligence

22:33 to go through your situation and i can see the advantages and why from a health perspective they would want to consider particularly a triage understanding understanding what the problem is but as you say there's a we have to get over the same issue that we've always had with chatbots so far and they still are not good enough which is where you end up with no further along than you need to get over the same issue that we've always had with chatbots so far and they still are not then you wanted to be in the first place and i had this same thing recently round around in a loop

23:01 go to the website go to the website and the website says use our chatbot and go back to the chatbot and then i said okay and thankfully it did respond to the and i didn't i didn't i couldn't i couldn't i go around in a loop and in the end i went actually on the web on the internet and googled one of these websites that finds you phone numbers that aren't like different phone numbers organisations then i phoned them and i got it sorted but i i but i couldn't do that through the normal channels because any of their phone numbers would

23:31 take me into the recorded and then tell me to the website to the chatbot and i was stuck in a loop and i think that's something that is yeah well actually there's a organisation called be human

Bringing humans back to the phones

23:43 and uh and i they they worked with zurich insurance where zurich insurers had this problem where the chatbots were kind of getting in the way and they weren't getting very good feedback so they flipped it so you can you call them now and they've got people actually answering the phones again but they use ai in the back backgrounds to support everything that the you know the those answering the phones need to be able to to do that and they've seen they've seen a huge success in doing that so i think

24:14 there's there's a there's definitely a desire i i think for people to want to have human connection to want to feel a connection to an organisation they want to work with often it depends upon if i'm buying something i don't really feel like i need you know i need a connection with ebay or amazon but if i'm buying a service whether it's training or coaching or marketing i want to feel like i i can build a relationship and i feel like that person will or that organisation will get to

24:42 know me so that then as they get to know me then it works you know the actual relationship works better the business works better because they understand more about what i do and not just what i tell them in the first place so i think that's a really good point and i think that's a and it's all very black and white isn't it with ai but i do think that's how many businesses are going to be able to differentiate themselves going forward by you know if you have a human call centre that's going to make me want to call you as opposed to call your competitor you know and

25:11 if you're calling a competitor i think uh i think that's going to be really good i know for example like when you call the bank you know and i'm monzo bank when you speak to the chat on the app i believe they mix and match a human with a computer so you might be getting frustrated with the ai and then a human comes along and you think oh thank goodness i'm talking to a human now and then the next thing you reply and suddenly it's

25:37 gone back to being a computer again and that is so frustrating and also if you do get the human back i don't think you get the same human back that you were talking before so it's like it's like groundhog day it's like a goldfish in a bowl it's like you're just reliving that moment as you ask the same question over and over again and don't get the same answer and then you're like you know it's very frustrating yeah and i think it can happen with humans as well i can't well i think you say that sometimes we don't necessarily know but i've got an issue where the podcast

26:06 appears twice on spotify i use something called buzzsprout to get the podcast to spotify and i was having conversations with both of them one via email one via via chat and both of them are nice you need to speak to buzzsprout but no no you speak to spotify eventually we got down to it was happened after the conversation like timed out with the person i was chatting to at spotify then

26:32 a new person came on and they said this is what you need to do this is i was like i've just spent half an hour talking to this other person backwards and forwards and you come along and and this was didn't look at that didn't look at the chat just went for the looked at the first bit i'd said i mean oh yes we still and i was like oh my gosh just banging stuff so there's still there's still that challenges now i think we we want to be human but we also want to make sure we're connected to the right human to the right human to the right human to the right human to the right human with the right knowledge and maybe that's where ai can help us but again

26:59 how do we make sure the knowledge that ai is giving us about the organisation or about the answer because one of the things i find most frustrating and i had this yesterday was when i'm asking a chatbot how to use a chatbot so i'm asking chatgpt how to use chatgpt because something isn't working and it doesn't know and then i go and ask gemini or claude and then he gives me the answer and i was like how does that how i can't understand that at

27:25 the moment surely that's got to get better sure it has to get better in terms of understanding that it's but i can't understand part of it is that when it's using its own yeah it's got a fixed amount of knowledge in terms for a period of time so the last time that they had the training data means if it launches something in two weeks you know two weeks in two weeks time but it's still on the data in two weeks time it's still on the same data two weeks ago then it's not going to

27:55 go off unless it goes off and searches for the internet and that's one of the challenges even though i suspect there's going to come a time where that that window is going to be very small yeah no really true really true someone was saying the other day about how how long is the memory of ai and i don't know if you know the answer to that but i think that's quite interesting as as as information evolves is ai storing all the old information as well as the new information

28:23 at what point does it know that this is old information and it's not going to be able to this is new yeah that's a good question i just don't actually know it comes down to how it's being trained i suppose but as you say how does it know and there's other thing about we talk about ai having memory in terms of how we use it but that has a limit as well because ai has something called a context window where it takes each request we make has a limit to how much information it can

28:50 take in a single request so if it's greater than that then it may well split that into different requests or it might just go oh that's that's there's too much information and it will potentially truncate it or ask us to give it less information so when you're building up a memory in your own personal layout or use data you think that oh okay it can consume all that data but really you want it to take all this data kind of summarise that into less days that

29:19 keep summarising it until you get to the point where yeah this idea that it's crunching masses amounts of data it's doing that but it's doing it in iterative process as well which in in some cases can get lost but that's a really good question i think that's going to be a a guy it's the first time i think i'm going to actually have to go away and look at that yeah actually how how do we make sure that because at the moment it seems like that isn't

29:45 the case i suspect the way that it works is that it's often so let's say it's often looking for what is the authoritative authority so for example it's my web site it might have some old information but eventually it's going to say well okay actually actually that this website has got new information on it and now i've got the updated updated information coming through and i suspect that'll be a similar thing if there's say 10 things that

30:12 say something that it picks up but then eventually there's now seven things that are saying something different and there's only three that are saying a different thing then maybe that then starts to steer it towards well okay

Where AI actually gets its answers

30:25 the probability is that those seven are now the same and i i i said to you earlier i was on i was on a radio show earlier today and the the host i said oh yeah so you've got your podcast coming out on friday and i was like no they came out on tuesday and it's so i'll stay on it said your next episode comes out on friday and i said i know that came out on tuesday no interesting i was like

30:51 where did he get it from i had to look when i got got back and i was into the office and i was like checked and i thought i can't find how it got there and then i asked and they first way i come back said oh yes it's every tuesday's come out and it came out on tuesday or the next one's coming out next tuesday and i said oh how did you get the information oh oh actually and it took it like i'd done something wrong because it said oh oh and i i i inferred that i don't really know for sure

31:17 and i was like no it's actually you're correct but how i just want to know where did you get it from oh i got it from your linkedin post and i was like oh so actually it gave me as we talked about insights i realised oh i need to put somewhere on the website some information about future schedules so that so it did help me to realise there's something missing because people are searching for something and and the information but where it got i have no idea

31:44 how it got friday that's that but again partly probably because it's all also about the amount of information if it's only got one reference to the information or not how it hasn't it can't find it it goes well i know that there's a podcast there's enough information for me to tell that it comes out weekly so but there's not enough information for me to tell when statistically it's probably a friday yeah so that's probably what happened

32:12 in that scenario because it didn't necessarily pick up what was on linkedin yeah but it should have said it didn't know right it should have said i don't know but i am going to make a guess of this because of this you know rather than just trying to pretend that it knows the answer when it doesn't yeah it should but i think that's just the way it's designed isn't it to to it to come up it it's it has to have it has to give an answer yeah unless of course there are circumstances where you're trying to push it for medical information so i was doing

32:41 that i was doing that the other day and i just couldn't get it to i was i just want you to i'm not not saying that you need to give me something that's definitive i just want to understand this it's like no no we can't do that you know you're gonna you're gonna have to go and do this about dosages of dosages of dosages of vitamin d it was but all right yeah and in the end i was kind of i gave up in the end i got rent works around it but was that on claude because i feel like claude

33:07 often judges me and doesn't it doesn't let me um doesn't let me ask questions that maybe are slightly dodgy you know you just want to hear an opinion even if it's an opinion of an ai bot you know um but i feel claude is is it is quite judgmental more than chat gpt that's right i mean chat gpt cat has and it still is a bit big sycophantic so there was the chat which was a chat gpt 440 that uh was the one that people ended up you know having relationships with

33:37 some people because it was just so like oh yeah it's so sycophantic oh you're so amazing and great job and but actually whenever i've kind of i don't use it so much now but there was times when like oh i just need to make this line a bit snazzier or a bit more like you know and i always found chat gpt was a was had that element of being able to be expressed i used chat gpt to come with the name of the podcast so that was one of the things that i used it for early use of it was to get it to

34:05 go off and well build me some code so go and check available websites and give me options and then just kept chucking it through until until it popped out options that are available that i liked nice yeah that's really good usage of it because it's so hard to find a company name along with an available domain name so that's actually a really good shout yeah it's definitely got harder my first domain name i bought in 1998 and and it's definitely got

34:30 harder since then then i didn't really care i was just buying stuff i liked but then i think it was probably in around 2010 when i first wanted to find a web a domain for a business you know where i could get the and then back then the twitter was a reference right it's like okay what is the twitter yes you have to get twitter as well yes it's the domain available yeah and uh and that

34:55 works but now yeah now it's it's i still managed to get some good ones but i've had to really dig dig deep and use ai because it was just even with ai it probably took me hours but but without it as you say it's it's impossible so you think it's available i know it's not it's gonna be three thousand dollars i mean i've got domains that tell me they're worth three thousand dollars but they're not worth anything if nobody buys them so yeah yeah so true so true

35:22 so what is what is um

How to show up in AI search

35:25 it can be ai or not about ai but what would be some advice that you'd like to kind of give give those that are watching in terms of marketing and technology and how they might approach that what we're seeing clients really focusing on at the moment is about um ai search and sort of appearing in ai search so um obviously a lot of what we do for our clients is around search engine optimisation and we're seeing very much that traffic is dropping on tools

35:55 such as google and obviously people are moving towards having more of a conversation rather than with google traditionally you'd put a keyword in and you'd get a list of websites even when they're going to google these days it's the ai overview that's coming back to them so they end up having a bit more of a conversation but there's lots of different ways that you can kind of support these ai tools to um allow you to kind of rank or appear for different prompts so we use a tool called

36:25 that um allows us to identify what prompts people are potentially using in your industry and what prompts your competitors are using and so you can identify where your competitors might be appearing for prompts where you're not appearing for prompts and also um what i really like about it is the kind of sentiment analysis because ai tools are very much trying to assess um you know kind of the quality of what you are offering or the authority of the sort of tools that you're

36:55 so whereas google relied a lot on the website ai tools still rely on websites but they're also looking a lot broader across um lots of different channels it wants to find you know ai tools want to see you mentioned um in a positive light across lots of different channels so whether that's in reviews whether that's in digital pr or or all sorts of different ways so um i think that's something i would definitely recommend is sort of looking into how you can

37:25 um really feed ai some of the information that that it needs in order to see your relevancy to certain topics um and i know very much at wagada we get really good leads from from ai from ai tools so um particularly um using our case studies so we we have various niches that we've fallen into so we work with quite a lot of cyber security companies we work with a lot of tech we also

37:51 have a bit of a niche in the sort of children's nursery and the sort of elderly care sectors and where we've got case studies on our websites we're finding that um we're getting found for companies that are putting into you know ai tools such as chat bt touch it touch gpt give me the best marketing agency for you know a cyber security company and these sorts of things so we're getting we're getting found for more relevant kind of niche phrases that are really

38:20 helping us to then sort of convert those companies um you know when they when they when we have those and i think i think there's a trust there as well i think when you get found through ai people feel like they're talking to a trusted friend whereas i think the trust has kind of gone a little bit with google i think i think we were all aware that google is just giving us websites whereas i think we we may be kind of you know we're a bit more trusting that the ai tools kind of you know are

38:46 making a recommendation they're not just giving a list of possibilities yeah absolutely it's really interesting what you said about the the authority element of it because yeah in historically you would go on to google and if you saw saw something listed three or four times in the first page then that probably could give you some confidence to think okay that this is actually a kind of a going thing whereas if it's one and you'd have to go and click and you'd have to go and look around it yeah and something i know

39:16 that um other people have found valuable is awards and so i think and again i guess that's another thing that i probably would use in terms of it's going oh well oh they're award winning therefore yeah that they also are going to be better than better than others because that's the other thing is what i that trick i use if i'm looking to buy something then i'll and i ask ai for recommendations i'll then get it to come up with a criteria

39:42 and score each of that criteria by 10 and give me a total score love and so and then that's really helpful because but also then i can query it and go well okay why why did you score that and or i can say actually that criteria is not important to me and then it can recalculate the scoring and help me find it but i wonder if that's a similar way that it's determining determining the right answer because what's different about let's say i i had an assistant and i said okay go and find me some websites for

40:13 this they're going oh yeah that that seems like you know there's not that they're going to find an answer but with ai because of the way it's built it as we said it wants to give you an answer i've got to find you an answer and you've asked me for the you know you've asked me for the best answer you've asked me in the in my location for a particular thing then it's going to look it's going to work harder in trying to try to determine that than necessarily we would so i think that's i think that's a fantastic bit of

40:41 advice the whole looking at searchable and looking at how we how do we rank because i mean it's how we've done it before well not everybody but we've done it before with keywords but now i understand we've got to start moving in a different direction because we are seeing there is a shift in people as you say look you know going to google and going to ai and of course even though that brings up its own challenges some would argue

41:07 it's good to use ai some use google sometimes because you do have to think about it a little bit more in terms of you're not just relying upon oh yeah great you know you you found me the the exact and i probably still use it maybe it's just because i'm you know because because i'm older but i still do tend to use google a little bit more of my searching than i do ai unless of course as i say i'll probably use ai to give me find me the products and i'll potentially we're going to use google because i don't find ai as good at finding the

41:37 right deals or yeah and that type of thing so that's just quite interesting i have um i've used it quite a lot to kind of buy computers and it was i found it's on one side it's really good because when you're buying a computer you've got a list of features and you say what features you want you say what you want to use it for and ai is brilliant for kind of you know assessing what's on the market and giving you a list of different computers and then you say where can i buy these computers and it tells you where you can buy it and they are never available where they

42:03 say they are available and then you go back and you say well you told me it was available there and it's not available and they said it says oh yeah i'm very sorry it's out of stock why don't you try here so it's it can be really frustrating in that sense and i think it can be very helpful but also if it if it doesn't if it doesn't check its information properly it can be quite frustrating yeah and interesting that's kind of that reminds me as i said about the domain name system and i think it'd be quite simple probably to build a tool that actually that can with and

42:32 then maybe it's something they'll integrate eventually into the the chat box we currently use but where you could get it to go off and it actually reads the page so so part of its job is not just looking at its knowledge but it's going because even if you say go and check i don't it it will still potentially go well i've got knowledge here why would i bother going why would i bother going checking or the other thing you probably found the self as well it comes back and gives you something else that's oh yeah i just

42:59 thought that you might like an e-bike because you know that that's going to be something that's going to be yeah instead of a the laptop yeah maybe you want to i think you know or a phone you could just e-bike to go and speak to the person that's a bit of an extreme example but yeah yeah no really true i think it's been it's been a fantastic conversation show i've really it's been great to get that perspective from uh from the marketing side

Where to find Cheryl

43:25 from some someone that's dealing with these sort of things from a day to day and if if people want to get into we're going to touch connect get further further support from you where can they get in contact with you um well they can go on the wagada website which is uh w-a-g-a-d-a.co.uk or find me on linkedin okay amazing and for those watching hope you've enjoyed this conversation if you want to unpack other things then other topics you want us to speak about on the on the show then

43:55 just please get in touch with the website betterwithhumans.com and just like follow on youtube apple apple spotify wherever you want to watch or listen you can find us and yeah thanks again cheryl for for a fantastic fantastic lovely great chat thanks for having me on

Follow the Pod