AI in Marketing Attribution: The First Era

Featured Speakers:-

Links and Resources

• The Hidden Gaps in Marketing Revenue Attribution Most Teams Overlook
• Top AI Analytics Platforms You Should Consider In 2026 and Beyond
• Boost Your Marketing Strategy with the Best Predictive Analytics Tools

Transcript

0:18:
Hello everyone.

0:20:
Welcome to our chat show where we are actually discussing something very important, something very important, as the terms, a big shift in the business.

0:30:
Well, our today’s theme is AI attribution in marketing.

0:34:
The first error.

0:36:
Now you must be thinking why we are calling it a first error.

0:40:
Before AI, the attribution was either cut feet or messy spreadsheet, but AI is actually changing it.

0:50:
Well, joining me today are the two experienced and expert who bring immersed knowledge to the table.

0:58:
We have Arpit Srivastava, Vice President of Marketing and analytics at Growth Natives, accompanied by Mr.

1:05:
Rahul Sachdeva, senior director at Growth Natives of the Analytics department.

1:10:
Welcome to the chat show Arpit.

1:12:
Welcome, Rahul.

1:13:
Glad to be here.

1:14:
Well, it is said that experience and knowledge is more enlightened and nurtured when you pass it to the others.

1:22:
Well, in our today’s show, we are actually doing that, sir.

1:24:
So how are you feeling today?

1:26:
Ecstatic, I would say.

1:28:
It was a long time due, I would say definitely.

1:33:
OK, so starting with you Arpit, my first question would be, as we talk about AI in attribution, what actually attribution look like before AI?

1:44:
Well, I would say that, the main thing attribution is trying to solve is what’s working and what’s not.

1:51:
The the question sounds very simple, but it’s very complex.

1:55:
I would say the complexity has, did a bit given all the advent of AI and your ability to really white coat a typical complex dashboard in a matter of hours, and it used to take maybe weeks and months so definitely there’s a radical shift that we’ve seen, if I go back somewhere around 2010 or so when I started off.

2:22:
In analytics, I did my first implementation in Google Analytics where Last Touch was the only model that we used to use.

2:31:
Now you have plenty of options.

2:33:
You have multiple models.

2:34:
You have multi-touch attribution.

2:36:
You have data-driven attribution models.

2:39:
You have geo experiments.

2:40:
You have VDMX modelling.

2:42:
So you have a variety of options here.

2:46:
Now it depends on your organisation maturity, how much you can handle, and can you really make some sense out of it.

2:54:
So I would say.

2:57:
The speed to drive the insights.

3:01:
Has become a lot more easier but the last mile which is taking action.

3:07:
I think over there is still AI is not going to support people like us, like Rahul.

3:14:
They’ll have to come up and really talk to the C-level executives and make those actions because ultimately if you’re spending 10,000 or 10 million.

3:23:
In data doesn’t really matter.

3:26:
What really matters is how much actions we have taken based on the sizes that you’ve generated, and that holds for attribution platform and tool system.

3:35:
Definitely, yeah, I think, you know, picking on your first sign, the attribution remains the same, I think.

3:41:
The very fabric of attribution that we talked about, you know, around a decade back, it’s still the same, but, you know, how we measure it or, you know, the, the complexity on the landscape where, you know, how the user is engaging with your power has changed dramatically.

4:01:
I mean, with the event of especially now AI, you know, earlier we used to go, you know, visit websites or nibble or, or, you know, web browsers.

4:11:
And now, you know, we are using DTI apps and you know within.

4:17:
We are doing all the shopping and browsing WhatsApp.

4:21:
So all those, I think, good old, I would say attribution pointers that we used to look at earlier has somehow gone to, you know, somewhat black box, we, yeah, so it’s, it’s, I think the complexity of measuring attribution has become little.

4:41:
more complex, but yeah, it, it obviously, I mean, we still measure it, we still, you know, check, you know, what you do, what is bringing the revenue, what are the touch points, so yeah, those are all things on the side.

4:57:
OK, that brings me to a question, like, although these developments have been taking place in the last couple of years, now why we call it this particular era, the first era for it?

5:09:
Like, are we like moving on a very fast pace to get into it or something has changed over the time period?

5:17:
Yeah, I think, you bang on right.

5:20:
We have not seen such a speed in in the matters of weeks we’re seeing really big updates if you’re not touched, some of the updates in a week, you even feel that you’re almost like a year bored.

5:35:
So the, the advent of the whole AI and the new models come in every week, new players joining in, new collaborations, new partnerships.

5:46:
I have personally never seen, it is more like a.

5:50:
Camel to car moment.

5:53:
So in the middle, it’s a, it’s a pop popular it is that they, they started off with camel and they straight away jump jumped to a car and that’s why we always find them in their driving skills.

6:05:
It is that kind of moment where we have not seen such a shift, and.

6:12:
Everyone has to really do a bit of extra.

6:16:
AI actually promised us that it will automate, but the reality is that you’ll have to, put extra effort to really be up to speed.

6:25:
Yeah.

6:27:
What do you think, I think.

6:30:
I remember a time when Amazon actually launched those, you know, small sleek buttons with a piece of that you can order something online and that was for somebody who’s like not very techy, but, you know, they may not be able to visit websites, but they still want to, you know, do repeat purchases.

6:50:
So Amazon created, I think it was back in.

6:53:
Yeah, same time around wicked bad, but yeah, that’s no button, that instant but, yeah, right, and those, yeah, and those were like physical buttons you could trace and stick on your cable or, you know, your laundromat or whatever, but those buttons, I think that is what we are reinventing now, but with AI and with much more sophistication on the software side.

7:18:
I think, you know, those buttons were a little ahead of their time, I would say, because they, they, I mean, nobody actually, purchased them that time.

7:29:
Maybe a few, but not much.

7:30:
There was not much traction though.

7:32:
But I think what we are doing now is, you know, building agents to do our shopping, you know, be able to buy those items inside Chat GBT.

7:46:
That, that’s crazy.

7:47:
And, you know, yeah, I see, I foresee a time where we would be, you know, asking agents to monitor our, you know, groceries or household item open a recently got the education to Shopify and so it’s coming, it’s there.

8:07:
It’s already there and it’s, it has been so fast, it is just a matter of being up to date and Have that curious mindset, definitely those people who are curious enough who want to.

8:21:
What we call as a.

8:24:
Child in the tiny store, that kind of kind of curiosity, I think those people will be successful.

8:30:
I mean, no one is, I would say 20 years of experience.

8:33:
Everyone is pressure right now.

8:35:
And when that mindset, those people will win who are curious, who want to really reinvent, who want to to reset and restart.

8:44:
Definitely, and it is said that AI is evolving every day, but the reality is every AI is evolving every second.

8:50:
Every moment you see some of the new changes.

8:53:
And with this, that brings me to a question to you, Rahul.

8:56:
You said, that about the black boxes.

8:59:
So how do you think that AI attribution and marketing and everything is not just a black box of guesswork or everything?

9:06:
Yeah, so when I said black box, it means that, you know, earlier we used to get those sketches, right, when we go to the website, so the website recorded the impressionss views.

9:17:
So those things might not be happening now because, you know, all we do is go to publicity as a product click on it, bought it, so you don’t see the whole customer journey there.

9:27:
We just see, you know, somebody bought bought it.

9:29:
They might be, you know, still, Doing research, exploring things, exploring other options for some time, but we, I mean, we, we departed.

9:40:
So it’s, I mean that kind of black box that comes with own set of challenges and opportunities.

9:49:
Now, there’s a new, you know, breakthrough.

9:51:
We definitely get these kinds of challenges, but again, those are points of officially as well.

9:56:
Yeah, I think the importance of, of Brad will also increase.

9:59:
I think we’ve been living in a time where you get some $1000 to Google.

10:07:
They have done a bunch of ads, you’ll get the leads.

10:11:
the whole playbook of that inbound performance marketing that needs to be requested, like Raul mentioned, it’s, it’s more like a zero click word that we’re living in, but again coming back to the attribution and, some of those things, the, the, the maturity model that we have still applies.

10:31:
I mean like you will still track whatever you can track.

10:34:
You will still report whatever we get report.

10:37:
You cannot just say that, oh, it’s a zero clickboard.

10:39:
I won’t report anymore.

10:41:
That’s not the case.

10:41:
So you’ll have to start, with some kind of foundation, call it.

10:47:
Even the last stick, I, I can tell you there are so many clients that they haven’t even enabled lasttic attribution model, and we are sitting in the AI first era.

10:56:
So you will have to understand where you are right now and how you can move the ladder, what you’re solving for, and there, there, there’s a path that you have to follow.

11:06:
Start with obviously the simple things.

11:08:
One of the biggest thing that currently most of the platform really want is good data.

11:16:
So when, when you’re running ads that in Google, Facebook, or LinkedIn, first thing is have you enabled conversion tracking.

11:22:
If you can enabled conversion tracking, that is step one.

11:26:
And then you’re feeding the, the AI algorithm with right signals that look for similar audience that have done this kind of actions.

11:34:
So if you’ve not done that, I mean like you’re still not in the step one.

11:38:
Itself once that is done, have you really configured your Adobe Analytics or Google and you track those there as well so that you can actually see which pages, which country, which channel are actually getting that traction.

11:54:
And then there will be a lot of dark social and zero play kind of moment for that again, you will have to go to a very traditional approach of measuring marketing, which is just ask customers.

12:07:
UCS guys will still speak to them.

12:09:
So when you’re speaking to them, just ask how we came to know about us.

12:12:
Track that in UCR, and then eventually when you collect this data, you will have really good insights that none of those tools can collect because.

12:22:
They never did, but they can still tell you that, yeah, we actually tie it on chat GPT and we found your citation, and that’s all we care for and now you will be forced to invest in chat GPT, get that exploited, really respect your AI SUD.

12:39:
Exactly.

12:40:
That’s more like, AI has everything, but you need to have a research implementation and then measure everything in order come to you by itself some of.

12:50:
Your efforts are needed, which are bed and Rahul, that was one thing and definitely our audience would like to know that if we are making a marketing strategy, what are the few tips that you would advise to our audience that they should keep in mind while taking the best help of AI.

13:07:
In the marketing attribution could be the quick tips, whatever you would like to think that these are the must things these are the things that you should be careful about, more from the attribution perspective for the attribution perspective.

13:20:
I think, the, the first step I think we were actually discussing this.

13:26:
Somehow, so look at what has converted in the past.

13:30:
Look at your existing customers.

13:31:
That’s the research.

13:33:
You need a bit of.

13:34:
I think you always have that data that these are my customers and then they are repeat customers.

13:40:
One thing we generally miss is that people who have bought.

13:43:
From us for a one time that could be a chance, but people who have continuously purchased, have done the renewals with your company for 3 years.

13:55:
What are the patterns that you see across those kind of customer boards which industry?

14:01:
Like what revenue size, what kind of buying committee were part of those, be closed, and that will help you really look for similar kind of audience.

14:12:
And that’s how you can make more impact in your marketing strategy and not just marketing strategy, the whole, I would say the revenue team call it marketing, which really helps you drive those initial conversations.

14:24:
The sales who closes it and the account management team to actually helps you resell, upsell, cross sell, expand the revenue, and in that Excel is a good funnel.

14:34:
So I think generally we, we are too much focused on acquiring new logos, but what we really miss is that there’s a cool sitting in your CRM system in your, Shopify systems about your existing customers that can help not only drive revenue from.

14:53:
them, but also help you get more simple loans.

14:56:
Definitely I think that that is a strategy that I generally keep on saying to customers, but you know there’s always like that one team that is more focused on customer acquisition and less focused on customer retentions, loyalty, those kind of things.

15:14:
And while you’re saying all this, not another thing that came to mind as a data guy I would say.

15:19:
You know, have a data check or data quality check in place.

15:23:
I mean, you cannot just, take whatever you have in your system first of all, because, I mean, you, you might cut all that it can be garbage and garbage out, right?

15:33:
Definitely, yeah.

15:34:
So I mean that is a must.

15:37:
So whenever I look at data inside, you know, our trans CRMs marketing automation, you know, we find so much of.

15:46:
It’s always a mess.

15:47:
It’s always a mess, definitely, yeah, and yeah, and these are not like, one-off cases.

15:54:
Everywhere, any time we go we find similar things.

15:58:
So it’s definitely, I mean, start with, the basics.

16:02:
You start catching the right data, the correct data, the standard or normalised data.

16:06:
I think that’s first, and, yeah, and build upon it.

16:12:
across your systems, you have to have a stitched story together, you know, which gives you a good feeling of, you know, where you’re starting, where you’re ending, what does the whole cycle look like.

16:25:
So that’s, and then one good thing is with AI coming into the picture, some of that foundation which is required for ensuring your data is in good shape, it’s harmonised, it, it is consistent, it is complete.

16:38:
AI can really help you some of that enrichment as well.

16:41:
It is not that you have to have like 50 people sitting doing internet research and a pending records creating that older record that we always want to have, but AI is now your co-pilot in that space as well.

16:54:
So be it import fees like industry, company size, radio size, number of employees.

17:01:
We have even agents who can really fetch the right information on the flight, get you signals around whether the company is growing, whether they are actually.

17:11:
Got funded recently and all those signals are very important to really understand when it is the right time to reach out the prospects because ultimately marketing is all about reaching the right people and trying to the right message so AI can support in all these three aspects.

17:28:
It’s a matter of applying those foundations and fundamentals and be that AI be the force multiplier in your strategy.

17:37:
Exactly.

17:37:
We have got someone to make channelize the first, yes.

17:41:
Which there’s one question that everybody asks and when then everybody hears a term called AI, the first thing comes that is this a danger for the traditional marketers or are they being more empowered with AI attribution and marketing?

17:55:
What will you take?

17:57:
Well, I think, AI is definitely.

18:01:
Can be your leverage if you take it positively because now what it used to take maybe hours it is now a matter of minutes so I don’t think it it is going to be your competition it is it is just matter how you’re leveraging it.

18:18:
Are you really applying some of the repetitive tasks and really looking for some alternative using AI.

18:28:
don’t consider it as it as a competition.

18:33:
rather, look out for new projects and new tasks where AI can really help you.

18:41:
it could be you pick any stream.

18:43:
I’m like our design here.

18:45:
I mean you can.

18:47:
Imagine like it used to take 5 to 10 days to get a full-blown video production like this.

18:54:
Now it is a matter of a couple of hours.

18:57:
I mean, like, we can do so much more.

19:00:
I mean, AI is obviously overwhelming, but at the same time, I think the whole playbook that year used to be there now needs to be revisited, and it’s an exciting time for all of us because we are all now into it.

19:13:
And if you’re not into it, definitely that there’s a danger for you.

19:17:
For becoming here, yeah, and that’s more like it’s evolving together it is, yeah, and, you know, I believe and I’ve seen that you know the digital marketing, I would say space is one of the, early adapters of AI, I would say, because, I mean, especially in terms of, you know, using tools, or, you know, combining creative with technical with, you know, doing just talk base, so all that is merging together and, you know, earlier we used to say, back in our, you know, beach group that, marketing is all about numbers.

20:02:
It’s not about, you know, it’s not qualitative, it’s contributed, everything of it.

20:07:
And, you know, combining that with digital marketing, it even makes more sense.

20:14:
so AI is, I would say it’s, you know, I would say it’s the most powerful innovator yet, you know how we do or how we deploy, marketing, so I would say it’s definitely a plus for Marketplace, definitely not a competition.

20:33:
OK, now as we are moving toward the last section of our.

20:37:
The chat show, and there’s one question that I really think that everybody would love to know.

20:43:
As you two being the flag bearer of something that is very important at the present time.

20:49:
What do you think?

20:50:
What’s next after the first era, as we see, what after the first era?

20:55:
What do you see as a future goal or as a future accomplishments?

21:00:
Again, for the attribution, yeah, for the attribution.

21:04:
OK, I think for to understand the future you’ll have to understand the present first.

21:09:
So in the current scheme of things, the, the, the standard frameworks around attribution, I would say there are 3 buckets.

21:18:
And that were approved by the researchers and doogle as well.

21:23:
The first bucket is obtained multi-touch attribution.

21:26:
I know, always debatable, yet still used.

21:29:
Still, you need that answer, what’s working, what’s not, whatever you can track.

21:33:
So make sure that you’re good there, you at least have some foundation there.

21:38:
Second is, more like mediumic modelling, which is again, not relying on cliques and cookies, but still doing some kind of regression analysis, and that can create some correlation between how much you’re spending.

21:53:
And how much you’re getting in terms of sales and ravelling.

21:57:
And then third level of sophistication is your geo-based experiments where you do come some kind of EV testing and understand, if I, let’s say, stop spending on a particular state or city.

22:13:
What is the incremental impact?

22:15:
Are we still getting the same level of revenue, or there’s a tip, and then from there you decide whether the campaign or the channel that you want to invest mode, it is working for your favour or is it against.

22:31:
That is where we are right now.

22:33:
I think what will happen very soon and it’s happening right now as well is, can we add a layer of explanation.

22:41:
Across all these three buckets and really help business users understand what are these systems are telling you and how quickly they can take initiative helping them do the scenario planning because even if you have set up these complex systems.

23:00:
Deriving insights.

23:02:
That can be actionable and that can even challenge the politics generally you would see across teams.

23:08:
It is a big thing, right?

23:09:
Can we simplify that?

23:11:
So that’s where I think AI can is, it’s already helping, but this thing I would see some level of work required in order to make it a lot more trustworthy and really cover all the real complex scenarios.

23:26:
If you have, let’s say one data source is straight, you can add a layer of AI and then have a conversation with your data.

23:34:
But if you have tonnes of other data sets and there’s a complex join that needs to be created, I would say it’s still not there that you just add a layer of LLM and then it will chat and tell you everything that you all know.

23:46:
But pretty soon I think that that will also get covered.

23:50:
So that will happen and then.

23:53:
After that, the only thing I wish that would happen is the old Cajun buzz that we’re hearing.

24:00:
Can we have agents who can actually take a call?

24:04:
And you were just talking and Adam today.

24:05:
Yeah, no, we should have an agent, but which accidently heals a pipe and a broken pipe, yeah, yeah, yeah.

24:13:
So that’s more on a data engineering side.

24:15:
And even from the, the, the decision making, can we trust the system and let It take a call maybe with some people in the loop in terms of let’s say increasing budget decreasing budget deciding your bonus based on your performance such that it becomes a lot unbiased let the media decide things so that’s where I’m thinking hopefully we as real human being we have not solved those complex political challenges that always we get.

24:48:
when we show the data, they don’t trust the data.

24:50:
When we show the model, they want to know more about the model.

24:53:
But They’re lazy.

24:56:
They’re lazy and to take the action, and can we solve that last mile is, is my hope.

25:02:
Yeah, I think North Star marketers would love to see on their dashboards.

25:09:
You know, not just how they’re performing, but how to perform.

25:12:
Exactly.

25:13:
That is one good, good question that they would want to know, you know, where should I invest, you know, what, what are my drivers, so they would, they would want to know that, and, you know, at some pace, and I would say to some extent we are there.

25:30:
I mean we are getting, you know, those kind of insights with LLMs today, but I think it’s more, And like we were talking about, you know, how we can.

25:40:
Automate or, you know, make it more whole in the sense that, you know, it, it, it drives itself.

25:51:
So I think that’s where we are headed for sure.

25:55:
Definitely.

25:57:
Well, so.

25:59:
Today we actually came across this, and we knew, and we actually understood that the AI in attribution is really working.

26:08:
It’s basically history unfolding from foggy guesses to crystal clear, from static model to the living attribution.

26:16:
And for this, thank you so much, Arpit.

26:18:
Thank you, Rahul, for joining us and to our audience.

26:21:
Until we meet next time.

26:23:
Remember, AI has got you covered.

26:26:
You all you need is to nurture it well.

26:28:
Thank you so much.

26:29:
Thank you so much for joining us today.

26:31:
Definitely we should do more, more.

26:33:
Definitely.

26:34:
The world needs us.

26:37:
Thank you.

26:37:
Thank you so much.

See more webinar