🎙️ Backstage Tech by George Helgesen

Dig Insights Co-Founder: Bolting AI onto legacy SaaS doesn't work

• George Helgesen • Season 1 • Episode 14

Share episode

Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.

0:00 | 30:24

Ian Ash is the co-founder of Dig Insights — a global market research consultancy with its own SaaS platform, Dig One, trusted by McDonald's, Coca-Cola, Meta, Pernod Ricard and dozens of world-leading brands. What started as a traditional consultancy grew into an AI-native market research company with patented algorithms, a 30% compound annual growth rate over its first decade, and a successful exit to Behringer Capital.

In this episode, Ian breaks down how Dig Insights evolved from traditional consulting into a tech-driven platform, why "service as a software" is replacing SaaS, and why he came back from semi-retirement because AI is moving too fast to watch from the sidelines.

Topics covered:

  • How Dig Insights went from traditional consultancy to building their own SaaS platform
  • How their patented "Tinder for products" algorithm predicts market share from simple swipes
  • Why winning Coca-Cola and Pernod Ricard globally was a step change for the company
  • Why legacy SaaS companies that bolt on AI get punished
  • How AI skills and agents are replacing prompt engineering
  • Why Ian came back after exiting to Behringer Capital

If you're a founder, product leader, or investor trying to figure out where AI fits into your SaaS business — this episode lays out the playbook from someone who's lived it.

👉 Follow George:

👉 Follow Ian:

SPEAKER_01

Welcome to Backstage Tech, a podcast for software founders, investors, and product leaders. Today's guest is Jan Ash, a co-founder of Dick Insights, a global market research consultancy leveraging innovative tech and their own SaaS platform to help world-leading brands like McDonald's, Coca-Cola, Meta, and many more make faster and smarter product decisions. Today we're gonna talk about how Dick Insights evolved from a traditional consultancy into an AI-native research company, the future of SaaS as AI agents take over more and more of the work, and why the next 18 months will decide the winners and losers in every category. Before we start, make sure you hit subscribe to get notified when the next episode drops.

SPEAKER_00

Hey Yen, how are you doing? Excellent, George. Yeah, thanks so much for having me on today, and thanks so much for having us in Lisbon. It's a beautiful city and you've been an incredible host. It's it's been amazing. Thank you.

SPEAKER_01

Thank you so much for joining me on the pod. Ian, you and I spoke the other day that uh one of the hardest things in a podcast is getting a great guest um on the show.

SPEAKER_00

Yeah.

SPEAKER_01

So I'm so excited to have you as a guest tonight.

SPEAKER_00

Oh, thank you. I was I was waiting to hear who the great guest would be.

SPEAKER_01

So we're going to speak a lot about AI and tech today.

SPEAKER_00

Yeah.

SPEAKER_01

But before we do that, I know that you've been running a successful consultancy and also a software business for longer than 15 years. And you've worked with brands known worldwide, such as McDonald's, Coca-Cola, Meta. Huge brands. And I would like to speak about your entrepreneurial journey a little bit. So could we start with an introduction of yourself and how you started the company?

SPEAKER_00

Yeah, sure, absolutely. Um so I'm Ian Ash, co-founder of uh Dig Insights. Uh Dig Insights is a uh a global market research consultancy, but we also have our own SaaS platform. Um and yeah, it was founded in 2010, me and three partners. Uh, and uh originally it was really just a traditional market research company. So we'd already had all three of us, or sorry, all four of us were actually still already quite uh experienced in market research, and we'd worked for a number of large companies over the years. So we were very used to that world, and um, you know, we were quite good at our jobs. As we had founded Dig and as it began to grow, it became very apparent that SaaS was taking over the industry. Uh there was a number of sort of breakouts. So a company called Vision Critical, famously Qualtrics, which you know sold for, I think, eight billion dollars at one point. And we realized that uh the industry was sort of moving into three directions. There was the easy, sort of replicable work at the low end, and that was getting eaten by SaaS. Then there was this big middle, and the majority of the firms in our industry at the time sort of fell into this middle category. So that would be Ipsos, Cantar, very large global companies. And then there was the very high end, and the very high end would often compete with, say, McKinsey or Bain. And so we began at the high end. So we, you know, our work was definitely not value priced, but we also did a lot more complicated data science than most of the companies who competed in the middle. Uh, and I worked primarily in in on the data science side of the business. But as the growth was continuing to happen in the more scalable automated, we realized that we were really missing out unless we also played there. And so we made a really big bet that took multiple years to ever have any kind of return, which was to build our own SaaS platform and our own technology. And that was an incredible journey because honestly, I'd never done anything like that. Uh, and so we had to start from the ground up, starting from originally we outsourced development completely. And then we moved to um to creating an entire dev team internally, which made a huge difference. And then sort of, you know, working with outsourcers as well, like like ProQuitters.

SPEAKER_01

What was the first version of the platform? The product is called Upside, right?

SPEAKER_00

It's now been once again renamed to Dig One because we merged it with our social listening platform, which we've also uh which we acquired and then added into the platform. So now it's called Dig One. Um the first version was actually on the Apple App Store. And the reason why we named it Upside at the time uh with two eyes was because one I was taken. And we thought, well, this seems very tech forward. Let's just put two eyes in there. Um and it was on the app store, and what it required was you know actual uh downloads and and installs, and we had to pay people credits to take the surveys, and we fairly quickly realized that the economics didn't work, and then we pivoted to an online web app uh SaaS version.

SPEAKER_01

Did that product become your competitive advantage and a result?

SPEAKER_00

Yes, not right away, but it did it did eventually uh become our competitive advantage. One of the sort of concepts that we wanted to bake into it was that we wanted the interface to be simple for the respondents. So the main exercise when we first built the platform uh was basically like Tinder for products. It was swiping, you know, if you like it, swipe left if you don't like it. And then we baked in a trade-off, which was if you liked multiple things, you'd say, which of these do you like more? And what we realized was based on that data, we could actually do uh like what's called share of choice. So we could do some very complicated data science on the back end, which we've actually since patented uh the algorithms around, but that could actually tell you what it's it's would be akin to market share a product could get in market, simply based on a very simple exercise for respondents to take. And so we sort of took that approach with everything in the platform. Simple interface for the respondent, but very deep data science on the back end.

SPEAKER_01

Now that you release DIG One, the new version.

SPEAKER_00

Um how has the product evolved? So the main evolution, well, it's it's evolved in a number of ways. I mean, it it really was at the beginning, it was it was primarily just that swiping exercise. We've expanded it to be a full uh market research SaaS platform or ResTech platform. So you can do any kind of question that you want: grid questions, scaled questions, open ends. We also have open-ended uh AI-driven video responses. Uh, but I think sort of one of the biggest changes just happened this past year, which was we acquired a company called OneClick that was a social listening platform. And then we've merged the two into a single platform. So now within Dig One, you can go in, uh, you can say, for instance, what are people saying online about procoders? And get the detailed explanation of what is being said online, what is the sentiment online? And then you can also then say, okay, this is a very interesting territory that I've discovered through the social listening. Let me ask some questions in the survey platform and get some primary uh data response.

SPEAKER_01

How do you customers use the new platform? Is it a DAY approach, or do you have a hands-on team who can onboard a new client, instruct how to use the interface, uh, how the methodology works? Could you tell me a little bit about that?

SPEAKER_00

Yeah, I think that's a really interesting question because I think it really applies to what's happening today as well, which is that um you can do it DIY, and many of our biggest clients do use it DIY. But you also have the option to have it be sort of partially serviced, meaning that we can program it for you andor we can report on it separately for you. But in addition to that, there's also people who say, I want my data housed there. Um, and we try to do as much of our work through our own platform as possible. Depending on functionality and feature, we may use other platforms as well. Um, but they don't want to actually have to interface. So they it would be a more traditional project done like through a market research consultancy. It's just that we are able to leverage our own technology in unique ways so that we can get better, faster, deeper insights.

SPEAKER_01

So all these global brands like McDonald's, Coca-Cola, Pizza Hot, if you were to explain in just in one sentence, why do they choose Dick Insights and your technology over the competition on the market?

SPEAKER_00

Yeah, I think it's a combination of speed, but also quality. So, you know, there's that famous saying it's uh faster, better, cheaper, choose two. And I don't think that trade-off is really true anymore. You can definitely deliver high quality at a reasonable price uh very quickly now. And so, you know, that's what upside enabled us to do. And so we were able to deliver turn around very complicated analysis extremely quickly for companies that want to make choices quickly.

SPEAKER_01

I think we're we're getting to the topic of artificial intelligence.

SPEAKER_00

Oh, yeah, absolutely.

SPEAKER_01

And I know that you've been using AI not just during the past a few years, but you also used AI before that. Which role does AI play in your company today?

SPEAKER_00

Yeah, I mean that's that's an that's an interesting challenge that I think we have and everybody has. Um it's definitely evolving and evolving very quickly. It's gone from you know, in sort of trying to embed AI within your product to realizing that it's your product, it can't, it has to be more AI native than that. And so we've seen it in the markets. Companies that are seen as sort of legacy legacy companies or legacy SaaS companies specifically that add AI or bolt on AI get punished. And so we're you know working now uh towards a world, and not just us, but everybody, where if if if SaaS is software as a service, we're moving to towards a world where it's service as a software. Um and people are gonna be switching from uh using your in your interface directly to a world where they're gonna just be expecting outcomes through very simple interfaces like chat. And the agent will take over a lot of the tasks that normally would have done been done either through them DIY or through a service. And that's the challenge that we're all facing, and I think that's that's exactly what we're tackling right now.

SPEAKER_01

Do you have anything like this already built in in Upside or in in Dig1?

SPEAKER_00

Uh yes and no. I mean, I think we definitely have parts of very complicated analysis within the software itself that are conducted by AI, and so they're able to do analysis for you. For instance, one of the features in Upside is called Storyteller. So after you do your study, AI will automatically create an output report that looks through all of the data, actually makes choices around things like which data sets uh should be compared to each other, uh, what are the key insights, and create a full story out of the analysis and out of the data. So that is absolutely already happening in product. I think where it becomes really interesting is the entire process from uh origination of whatever the business challenge is all the way through to delivery. And we're all going to be in a world, I think, where AI will take over multiple steps along that journey. And it'll human beings will, in a sense, be editors and overseers.

SPEAKER_01

When we met in the airport a few days ago, yeah, you mentioned to me this. You said, George, I've been a bit busy in the past weeks. You don't know this, but I'm back back to Dick.

SPEAKER_00

Right. Yeah. After our first um our first exit to uh to Behringer Capital, which is who who've been a fantastic partner, uh, I sort of semi-retired, focused on things like my podcast and and writing. But the pace of change with AI has been so intense, uh, particularly with the uh release of Claude Code, the new releases in Claude Code and Claude Co-Work, um, that have also trickled through other other platform or other agents and LLMs like like Manus, for instance. Uh that you know it's it's apparent that we sort of I'm coming back in a in a more of a project capacity to assist with how we um leverage some of that and and sort of how we can quickly ship and make sure we're capturing value quickly. Because I think a lot of companies, you know, dig is very forward in AI. We have multiple people with master's degrees in artificial intelligence, for instance. Um, but it's gonna be about speed. I think the next 18 months are gonna you know decide a lot of the winners and losers in every category.

SPEAKER_01

Could you tell me a bit about the tools that you're integrating or the tools that you're using yourself and trying to integrate?

SPEAKER_00

It's it's I think it was the release of skills in both Claude and I I like Banis, which is a separate agent. So I that's the one I tend to use more because it it uh it plugs into every LLM. So I can, for instance, say uh I want Claude to do the analytical work on the CSV file. Uh and I also, but I want Gemini to use nano banana to create images related to it. So I can I can plug into multiple AIs without having to create my own harness. So I'll I'll speak to that one a little bit. But it was it's really the advent of skills and how easy it is to create skills in these platforms now that has made such a an incredible change and a need for a pace of change to really escalate. So I'll give you an example both in Claude and Manis as well. It used to be about prompt engineering. So you would you would create a prompt, uh, you would see what the outcome was, you would take that prompt, you would refine it, and then you would go back and you'd try a new prompt again. And so there was sort of this iteration and cut and paste experience that was happening in prompt engineering. And so that's why people were calling themselves prompt engineers, and they were saving out prompts and saying these were best practices, like CoStar. Um yeah. So now the process has become so intuitive. So that if you're in Claude or if you're in in Manus, for instance, you can go through your entire process. You can say, okay, first I want you to process this data file, then I want you to create a PowerPoint, then I want that PowerPoint to include uh nano banana images, then I want them resized, then I want you to cut it down to 12 slides. Whatever that process is, you can go through it and at the end of the entire process say, now turn that into a skill and save it to my library. And so you don't need to think about what's the perfect prompt. AI does that part for you. Right. And that is such a fundamental change in the way that we think about how we can save these skills. And then once you have these skills perfected, which no longer requires this iterative process of back and forth, you can chain them together. And and I think this just changes the game for everybody.

SPEAKER_01

How do you document the processes? I wonder, for instance, what what I do with with Pro Coders, we do a lot of work around product documentation, all these PRDs, test cases. Right. We just take the recordings of the calls, all the meetings, we transform it into prompts and the skills, and then we use uh clothed for it, for instance. Right. How do you document your processes?

SPEAKER_00

At this point, it's we're early days in in sort of rolling this out uh company-wide. But it's it's a matter of that that initial process of refining a skill and getting it to the point where you want it to be is relatively iterative, and we don't and you don't necessarily have to um document all of that. You you can if you want to, but it it seems somewhat unnecessary. Once you actually land on a skill that is that you're happy with, you definitely need to save that in well in this in our case, we're gonna be saving that in a GitHub uh repo so that you can now recall it, but you can also use it through other through other um platforms. So for instance, if I build a uh a skill in Manus, I save it to a GitHub repo, it's also accessible to Claude. They're actually interchangeable. Um, so that would be a good way to have a proper central uh repository in whatever repository you want, that uh is the one you know truth as to what the skill should be, um, and also has proper version control.

SPEAKER_01

Now that we're speaking about AI tools and agents, what is your opinion on AI assisted development, all these uh tools as Cloud Code, cursor that help engineers, or maybe Vipe code in platforms like base44, lavable, raplit? What do you think about them?

SPEAKER_00

I think it's amazing. I think it's such a multiplier of what people can do on their own, and it's creating this very interesting situation where you can have relatively small teams like like we talked about Lovable on my podcast with you. Uh, you know, they have like a million dollars ARR per employee. It's insane. Um, and and so I think it's that multiplier effect that AI gives to developers that is just unbelievable.

SPEAKER_01

Uh I had a question on my list about I probably mentioned this to you before, about your first big deal. It's a business question. Would you like to speak about this?

SPEAKER_00

Can you Yeah, I mean, I our first big deal. It was it's interesting because we started like in traditional market research consulting. And so that actually was a it wasn't like this one moment inflection point. It wasn't like we suddenly, you know, we were this sort of mid-sized company, and then suddenly uh this one big client came in the door. It all sort of happened gradually uh over time. We we were we were growing very quickly. I mean, our first 10 years of business, we grew at a 30% uh compensated annual growth rate. So we were holding on for dear life anyways, but it wasn't just sort of just one, but with our with the software, I could say that there was a couple of real inflection points. So with with upside, what was called upside and now called Dig One, uh, you know, I think it was Pruneau, Ricard, and Coca-Cola. And when those companies determined that they were going to use us globally, um, that was a major step change for us in terms of the way that we thought about our software, in the way that we had to be enterprise grade and enterprise ready, the way that we needed to have SOC 2 and get proper pen testing every year and all that sort of thing. That was a major step change. So when when we won the global work uh for both Prono Ricard and uh Coca-Cola, uh those were that was a major inflection point for the company.

SPEAKER_01

I get it. When you work with these big boys, you know, the SOC security, you need you need to be compliant, you need to prove that you're uh that your software is enterprise ready. Yes. I I wonder from the product standpoint, what what challenges did you have building building the features that Coca-Cola needed, for instance?

SPEAKER_00

I think every company has that challenge when they're looking at uh managing the roadmap, for instance. So you have a few big clients and they have specific demands, and obviously they get their weight matters, and so they have more impact on your roadmap than a lot of other comp a lot of other clients do. But you still have to manage the overall roadmap of the product. And if you personalize it too much to any one single client, it's no longer scalable. So it's a it's a balancing act. You know, there's what you try to do, or what we try to do, is say, what are the features that this client needs that we're gonna need across clients? So those become highest priority. So if your biggest client says you need to be SOC 2, and you're like, well, we should be SOC 2 anyways, then you're doing SOC 2. Um, but there'll always be requests that are specific to individual clients and it's a balancing act. And that's where actually companies like yours, like Pro Quoders, actually can really come in handy because you need to keep on your traditional roadmap. You you have a goal for your software, and if you just you know meander off into every feature request from every client, it never gets built properly. Uh, but you can look at that list of priorities and you can you can actually give an entire feature set to a company like Procoders that does it well and can integrate with your team, and then everything gets built and everyone's happy. So, I mean, that's you know, that's one of the reasons why we love working with you guys.

SPEAKER_01

Exactly. Some of our Clients even call it an elastic team approach where we just work as your remote IT department, just integrate it into your processes, working, going to the same stand-up meetings, following the same guidelines, the same code style, the same architecture. You mentioned your podcast. Yeah. And actually, I was really excited to be a guest on your podcast yesterday. I will leave a link for it in the description of this episode. I never asked you this question, but I'm curious, how did you start your own podcast?

SPEAKER_00

Yeah, so actually I I had the idea. I'd started writing a newsletter that I'd called Biz Unsolicited Business Advice or Unsolicited Biz Advice, and where I was actually writing letters to CEOs of large companies and giving them unsolicited advice, saying this is what you should do. And I first started doing it for fun. And I noticed that actually it was getting fairly good traction. People liked it. And it's kind of like being an armchair athlete for business. It's not that my idea is necessarily right or wrong, just like when people have an opinion about, you know, which football player should be moved to a different team and which one should be uh you know kept. But that's strategy. Strategy is a set of trade-offs. So you, you know, I think the discussion around strategy is is is actually fun for a lot of people, including myself. And so I realized that uh I was getting good engagement in the newsletter, but that for it to really get break through, you needed to have a more personal element, and that was the that was the podcast. And so I talked about it with uh my podcast partner, Steve, Steve Mast, and uh he agreed to do it with me. Yeah, and we're about we're I think we're seven episodes in now, six episodes, seven episodes in now, after the one that we just completed with you. So yeah, it's great, it's a lot of fun, and it we're really having a good time.

SPEAKER_01

I know it's not easy sometimes to transition from one format to another, because I started my podcast on Spotify only, because uh I used to be camera shy, and also I didn't know if I would be ready for it because all this video production, audio production, it's this is complicated stuff.

SPEAKER_00

Right.

SPEAKER_01

Uh for you to transition from an article to a YouTube and also Spotify podcast, what was the biggest challenge?

SPEAKER_00

Uh yeah, absolutely. I mean, the technical side of it, I think, just like you, it was um I had done podcasting in the in the past for dig. So dig has a podcast called Dig In. Uh, and I had I had started that actually for Dig years ago. Uh it it was just primarily odd audio at the beginning, but it transitioned to a YouTube uh podcast as well. And uh so I I'd had experience in the having the conversation on camera, but I didn't know how to do any of the technical myself, and I didn't want to suddenly just go out and uh and outsource all that because I wanted to understand what I was doing uh at a deeper level and understand and and sort of see how I could make the product better. So learning that you know, learning how to use new platforms is always challenging. So learning how to record and edit and host and all of those things, but it wasn't that hard. So, you know, it was it it was it was fun actually.

SPEAKER_01

And I hear you're a supporter of a hands-on approach where I heard you're building skills for manos, uh claude yourself, you're editing the videos yourself. That's that's amazing.

SPEAKER_00

Yeah, it's you know, it's great what you can do with AI these days. I mean, I think I would have been really challenged to do all my own editing before all of the all of the new AI tools that are available, because it just takes so long. But um now it's so simple. I mean, you can even and and if you build separate skills in in things like Claude or Manis, uh, it can handle whole workflows for you. So the amount of work that I can get done on a podcast by myself is in in just a matter of days is would have used would have taken weeks before, I'm assuming. How many percent does AI do for you in video editing, for instance? Probably half of the work. Because I can simply say, for instance, uh in the platform that I'm using, it's called Riverside, I can say remove all ums, ahs, and it will just automatically do that. And it'll often that'll chop off like between five and ten minutes, just right there. Uh, I can do things like uh use create AI B roll on the fly. Um there's just so many things in there that I can do quickly. I'd say it probably takes about half the work off my plate. It also does things like automatically creates show notes, chapters, all those things. Yeah.

SPEAKER_01

Yeah, I also use a lot AI for editing, specifically for for the short clips.

SPEAKER_00

Yeah.

SPEAKER_01

So I export the script, I put it on Claude, somehow I I prefer Claude. Yeah. And then it gives me options. I read them and then I manually chop them. I'm still a little bit of an old school guy in the meaning. So I I prefer doing it myself manually. Yen, I have a rapid fire round for you. Okay, 10 questions. Do you mind doing that? I'm I'm game. Okay. Don't overthink it, just pick one of the options. Oh, okay. Okay. Claude or Chad GPT? Claude. Joe Rogan or Stephen Bartlett? Stephen Bartlett. Motley Crew or Motherhead? Motley Crew. Nike or on running? On. Beer or port wine? Port wine. Oysters or barnacles?

SPEAKER_00

Oh. That one's impossible. I still can't believe you ate those barnacles, man. Um, definitely oysters. I I never want to look at barnacles again.

SPEAKER_01

I hope I didn't look like you weirdo when I took those uh delicacies back home with me.

SPEAKER_00

No, I understand. But you know, every delicacy is a little strange, but yeah, I couldn't do it. I didn't have the I didn't have the strength of will to try the barnacles.

SPEAKER_01

We'll have a picture of those barnacles right here. Okay. Listening to a podcast or recording a podcast.

SPEAKER_00

Oh, that's a tough one. Uh I guess recording a podcast. Yeah. Hiring developers or vibe coding. Oh, I'm you know, I'm relatively new to vibe coding. I've I've I've been trying it. I've um yeah, I think hiring developers still. Okay.

SPEAKER_01

Nate Bargatze or Jimmy Carr. Nate Bargatsu. Yeah. And the last one, advice that somebody asked and paid for, or unsolicited business advice.

SPEAKER_00

I'm really enjoying the unsolicited biz advice because when someone pays for advice, uh, you you're you're beholden to what they want advice on. And when I get to do the podcast, I get to talk about anything I want and give them any advice I want. And, you know, they're not paying for it, so there's nothing they can say. Perfect.

SPEAKER_01

Yen, thank you so much for being my guest today. I was also really happy to be a guest on your podcast. We'll leave a link for it in the description. Uh, the episode where we discussed lovable and gave advice to them. So thank you so much once again for visiting me here uh in Lisbon and Portugal. And I hope to see you again soon.

SPEAKER_00

It's been an amazing experience. Thank you so much, George. Uh it's been a lot of fun. And it's been great to meet you too. Thanks again.

SPEAKER_01

If you enjoyed the episode, hit like, subscribe, and drop a comment. And by the way, check out more backstage tech episodes on my channel. If you'd like to join as a guest, send me a DM on LinkedIn or email at George at GeorgeHelgison.com. Thanks for watching, and I'll see you in the next one.