System1 (SYS1): Unrecognised Growth Potential?

Following Maynard’s recent write-up, I have taken an initial stake (c. 25%) of a full holding in System 1 (S1). (I used to avoid this approach, but I have done it a couple of times recently and I’ll see how it goes.)

As always, Maynard has provided an excellent write up: Maynard Paton | SYSTEM1: Remarkable Q2 May Lead To 8x P/E And Potential Recovery Helped By ITV Progress, New Clients And Net Cash

I’ll try not to rehash Maynard’s work (but, inevitably there is some repetition) and provide a few thoughts by exception.

Investment thesis

I look for understandable smaller businesses, where the principals have significant shareholding (i.e. in same boat as shareholders), that I believe can be materially bigger over a reasonable period (e.g. 5 to 10 years). My approach is probably a bit more similar to Richard Beddard than Maynard.

While S1’s current EV is £27m, I believe it could be a materially larger business of c. £150m or more in the medium term (of course, I may well be wrong!).

William Leverhulme (d 1925), the founder of Unilever, is famously said to have remarked: “Half the money I spend on advertising is wasted and the trouble is I don’t know which half”. S1 seeks to address this issue by quantify advertising effectiveness (see below).

I looked at the business when it was BrainJuicer. At that point, I hated the name and, possibly wrongly, concluded it was unscaleable and bespoke business, more driven by an advertising, rather than hard-nosed business, culture. The costs in these types of business often grow quicker than revenues. Having watched Mad Men it seems that advertising folk prefer to operate (aka consume cocktails) in London, New York, Sydney and Singapore rather than in, say, Bolton or Oklahoma. These folk and their operations do not come cheap. While some of this culture may linger, my investment is based on the belief that S1 is becoming an efficient systematic business exploiting a market-leading niche based on a unique database of advertising effectiveness (i.e. it is increasinlgy replicable, scaleable and profitable than a bespoke project consultancy).

At the current valuation, even if I am wrong, I believe I will at least get my money back.

Activity

As Maynard points out, S1 does not explain itself particularly well. S1 contends that consumers respond to adverts in an instinctive emotional way rather than entirely rationally. Based on Daniel Kahneman and Amos Tversky’s studies, behavioural psychology holds that System 1 (the instinctive and emotional response) is much more powerful that the rational System 2 (which is apparently System 1’s “lazy policeman”).

Unlike many others who analyse adverts rationally, S1 measures consumers emotional response to adverts and gives each ad a star rating. The rating is based on S1’s proprietary analysis of thousands of adverts. We are bombarded with ads and forget most very quickly. S1 measures the consumers’ view on the product’s fame (do I know it? If I don’t the ad is wasted), feeling (do I feel good about it?) and fluency (can I find it quickly?).

Encouragingly, the star rating provides tangible feedback. For a 2 star ad, S1 would expect a 0.5% increase in market share for every 10% “extra share of voice” (i.e. increase in relative marketing spend). A five star ad would increase the share increase to 3% for the same ESOV. Only 4% of ads are rated five star, but the difference between two, three and four starts is meaningful. This provides immediate and quantitative feedback on the creative process.

My initial research could not shed light on how the ad appraisal actually works and I worry there a bit of “black box” magic about this. Maynard’s piece has more on methodology the panel, process etc… but I could not find it on the website. I wonder how robust and consistent the process is. Also, while a lot of the commentary has focussed on TV (and particularly ITV), I think this will work on all video advertising (and I note Facebook is a client). If that’s right, it could be really important.

Mark Ritson (a well-known marketing writer, professor and consultant - see below in people) believes that S1’s product is differentiated.

Clients

I am not sure to what extent the client or decision maker is the advertising agency or the “brand owner”. It appears S1 has an impressive customer list ITV, Adiddas, Facebook, etc., but it is not clear on what basis these clients use S1. I am also not sure about the model. Are the apparently cheap Ad or Brand tests leaders into other business (e.g. how to improve test score)? All insight gratefully received.

Competitors

S1’s materials suggest Kantar is the dominant player in the market, but blogs etc… suggest that its main competitors are Ubiquity (https://ubiquitygroup.co.uk) and Analytic Partners (https://analyticpartners.com). Apparently, S1 is differentiated in the Ad appraisal segment.

People

This aspect may have been underestimated by commentators and bulletin board participants.

I think Stefan Barden’s appointment as CEO is hugely significant. John Kearon’s LinkedIn feedback on Stefan is fascinating. Stefan Barden appears to be very different from the typical advertiser. I think he was a Unilever fast-track, McKinsey consultant with significant CEO experience mainly in large food-related multinationals. He was a very successful CEO of Wiggle, where I understand he made tens of millions. After that, he went plural mainly advising interesting private businesses – most of the fun, with much less of the responsibility. I am not sure his c. £0.5m investment is that significant – while it not “chump change”, I don’t think it is a material percentage of his net worth. It is much more significant that he has taken the title of CEO of a (small) public company after effectively going plural. The financial investment may be de minimis but the reputational one is significant. I doubt he would do that if he did not think S1 had material growth prospects and that his structured approach was both welcome (i.e. from the founder and the team) and important in the value creation. The impact is already visible: office rationalisations, systems’ implementation, net over gross profit, cost control and a data/system based strategy.

Mark Ritson is a well-known figure in the marketing world: he has been a leading business school professor for over 20 years (he is a brand expert who wins awards at every school he teaches at - he even taught me when he was starting out), he has consulted to many major corporations and he writes for both leading professional and academic publications. Notwithstanding that Mark appears to be mates with a few of S1’s principals, if he believes S1 is differentiated, it probably is.

Valuation

Maynard covers this in some detail, so I will not repeat.

At the moment, at the tail of Covid, it is often difficult to establish sustainable earnings. For what it is worth, I think S1 makes c. £4m of pre-tax operating profit purely on the basis of adjusting for the dire FY22 Q1. S1 had not lost money in a quarter for the last five or six years, so even adjusting that quarter to £0.5m profit (low by historical standards) seems conservative. There are lots of arguments both ways: revenue momentum is positive, but costs may be under-stated and the advantages of government support are arguably not sustainable.

If you accept £4m, an EV of 7x is not punchy especially given increased earnings from database and lack of appreciation of ITV/Euros benefit.

Conclusion/Questions

If S1 becomes a better version of its previous self, I suspect it will be worth more than it is now. If it transforms into a unique data led service, it could be worth much more.

I still have many unanswered questions that include:

  • How does the ad rating database/system really work?
  • How difficult is it to replicate S1’s system? [They have spent £5m+]
  • Who is the ultimate client and how much is the S1 analysis in an entire project?
  • How big is S1’s actual market (ad testing)?
  • What is realistic long term cost base?
  • Has the culture really changed to a hard-nosed profit/cash focus?
  • Should S1 invest in exploiting the unique market opportunity rather than buy back shares or paying dividends (i.e. is founder putting his needs ahead of business)?
  • How related are consulting sales to the system? If so, what proportion are systematic rather than bespoke?
  • What happened in 2018 (pre Covid)?
  • Is labour properly allocated to projects? (i.e. what is project contribution margin rather than GP?)
1 Like

Hi Mike,

Thanks for the great write-up. SYS1’s results are out on Wednesday, which may help answer some points.

These links might help:

https://system1group.com/ad-of-the-week/60-heineken

https://report.testyourad.com/report/e14e94fe-c1c3-44f8-8917-4e9bb1eec81c

Smart Working_-NatRep PDF.pdf (759.3 KB)

Essentially SYS1 has an online panel of tens of thousands of people and picks 150 to judge the advert. The 150 watch the advert and click a button to represent their mood (happy, sad, anger etc). I think another 150 may watch it and click when they recognise the brand involved (fluency).

The tests should work on all video advertising, though I am not sure whether online ads can build a brand as well as TV. Note that the largest advertisers of the last few years include US techs such as Microsoft, Amazon, Apple etc. You would have thought their target audience was online, not watching TV.

I don’t believe anyone is doing exactly what SYS1 is doing. Some firms rate adverts, but not in the same way. If a rival were to rate adverts in exactly the same way, it would need to construct a database of tested adverts (SYS1 has been testing every UK/US adverts broadcast since 2017/18), and then correlate the findings to subsequent changes in market share. This is the hard part – convincing CMOs the test results do lead to greater ESOV etc. SYS1 has been doing this style of testing for 20 years, so should have a decent head-start on a complete newcomer.

The likes of Kantar could get involved in System1-type testing, but according to SYS1’s management, that would mean throwing out years of System2-type work, which might be a hard concept for the clients to accept – i.e. after being charged big £££ for years for research that would no longer be relevant.

Oh this is interesting. Do you work in marketing?

SYS1’s tries to deal with the company direct. I have been to past AGMs and management has said getting the attention of CMOs has not been easy. Agencies apparently hate SYS1, as SYS1 often tells the client direct that their ads are poor and the agency has to go and redo the ad.

(AGM 2019 notes, AGM 2018 notes)

https://system1group.com/wp-content/uploads/2020/11/2020-Interims-results-presentation.pdf
“Every year $900bn is spent on Advertising. Over 50% has little or no effect & yet only 0.1% is spent testing whether an Ad will work”

I am not sure as yet. Future results should help us answer those questions.

Dividends are not on the cards just yet. The buyback could be a useful way to create value for shareholders, assuming the shares are indeed inexpensive and the future of the company is indeed bright. I am hopeful future product investment can be funded from annual cash flows, but I may be wrong on that. I would like SYS1 to maintain a cash-positive balance sheet.

The last statement said 15% of Q4 revenue was from automated-data products, so the rest I guess is consultancy. Mind you, I am sure some consultancy revenue may not be that ad-hoc, such as ‘Brand Tracking’, which SYS1 has said in the past is quite a reliable income source.

Certain large clients cut their marketing budgets. SYS1 cited the mooted bid for Unilever, which I understand is/was a SYS1 client, and the possible bid prompted cost cutting throughout the sector to shore up earnings and dissuade activists. I think that event prompted SYS1 to start transforming into a more reliable data business.

I don’t think we can know this for sure. But I would like to think revenue per employee will improve over time as the ‘scalable’ nature of the data business emerges.

Maynard

1 Like

Maynard,

Thanks - super-helpful. The adoption challenges were particularly interesting.

I will continue my homework, including reviewing the results on Wednesday (MP edited).

I don’t work in marketing, but know many that do - some who may be particularly relevant. I will see if I can find some scuttlebutt!

MTIOC

Any thoughts on yesterday’s trading update?

The shift towards data from traditional practices appears to be accelerating but overall growth still seems relatively slow.

I attended the AGM in the hope that there might be some form of presentation and/or Q&A session. It was just a dull session with only the agenda items/votes covered with a kafkaesque voting system on Teams.

Stefan Barden wasn’t in attendance at all which was disappointing as I was hoping to see how he and JK inter react.

1 Like

Hi longshanks,

Thanks for the message and welcome to the forum.

Difficult to know for sure about growth rates given the weak comparatives from H1 2021 and the business being in transition.

But Data revenue is accelerating and this is important I feel, as the uptake of the new-style Consultancy assignments will be dependent on the initial volume of Data activity.

Some of the old-style Consultancy work is expected to be dropped, and so Consultancy revenue as a whole may look like it is contracting for a while. The question I guess is whether the new-style Consultancy products can grow as quickly as the Data services, and what the eventual economics of both sources of income will be.

At some point I will review the annual results and the AGM statement for my blog.

Yes, poor show really, especially as two days prior I attended a ‘physical’ AGM that hosted a near 1-hour Q&A. So proper AGMs can be arranged at present.

Mind you, SYS1’s management has hosted lengthy Q&A sessions at previous ‘physical’ AGMs, so I am reserving judgment as to whether the directors have suddenly decided to ignore ordinary shareholders. The voting showed notable protests about the LTIP change and board remuneration, so perhaps the board did not want to risk a protracted debate about those matters.

I think the chairman said at the start of the AGM that the questions submitted to the board prior to the meeting had all been answered (in private).

Ah-ha, here is the AGM notice, dated 20 July, stating no questions could be asked during the meeting. Wish I had read that first!

Shareholders attending virtually will not be able to vote or ask questions during the meeting.

Shareholder questions
Shareholders can submit questions to the Board in advance of the AGM by emailing them to legal@system1group.com by no later than 09:30 a.m. on 2 August 2021. We will consider all questions received and seek to provide a written response in due course.

Maynard

I also made enquiries regards attending the agm but didn’t log in after informed it was just formal business. I did note from the subsequent RNS a few of the resolutions had significant opposition. Remuneration, rebasing of the LTIP etc

I can’t say I am too bothered by the remuneration policy and certainly didn’t vote against it.

If there is anything that grates, it is the £220k paid to James Geddes for “loss of office”. The RNS in 3rd April suggested it was a resignation and not with immediate effect. Looks like it was more a case of paying him “clear off money” to allow Stefan to develop the business his way.

I smell something in all this. Not necessarily negative, it could be they are working on a merger or trade sale of the business, but I sense some board awkwardness with where they are in the transition of the business.

Hi longshanks,

Some might argue the changes to the LTIP will cost shareholders more that the £220k paid to Mr Geddes! Definitely looks now as if he was paid off to leave as Mr Kearon reshuffled the board.

The remuneration voting may also reflect Mr Kearon’s additional pay during a year when the company received government benefits, did not declare a dividend, reduced the workforce and Mr Kearon seemingly delegating some of his executive duties to his new chief exec.

I think the board awkwardness has now been and gone. Two chief operating officers left between 2017 and 2018, and Mr Barden’s appearance as a board adviser I am sure ruffled a few feathers. My notes say

"He supports management teams deliver step change value by aiding their setting of clear strategies, creating a positive performance culture that ‘gets work done’, and in helping build strong teams.”

…and his ‘gets work done’ attitude may not have gone down well with typical ‘marketing types’.

The recent results had a ‘Group Overview’ statement signed by Mr Kearon, Mr Barden and Mr Willford, the new FD, so I get the impression they are all in tune with the new business direction.

I did wonder whether the business would be snapped up by ITV during the first lockdown when the market cap was just £12m, but I reckon the board – with its new £1b market cap ambition – may want to keep the business independent for now.

Maynard

PS No bad language on the forum please. Aim is to maintain a higher standard of discussion than found elsewhere.

They have a whole book on what they do:

https://system1group.com/book

Lots of background details - chapter 3 is on advertising…

More here:

https://report.testyourad.com/report/e14e94fe-c1c3-44f8-8917-4e9bb1eec81c
https://system1group.com/test-your-ad

But don’t really understand why this can’t be replicated by others - where’s the moat here - maybe they have patents?

Hi Colin,

Thanks for the post and welcome to the forum.

It could be replicated by others. But to sell these services to a CMO, you have to prove your ad data is super predictive and matches up to subsequent changes to market share etc. That is the hard part. And even if you could replicate that proof as well, you would have to sell the service cheaper than SYS1 to ‘disrupt the disruptor’. And perhaps get a deal with ITV to help drive sales. Some more from my earlier comment:

Maynard

Hi Maynard,

Thanks for the welcome! Only found your forum recently, really like what I’ve seen so far - a high quality initiative.

And thanks for your response, it really does look intriguing and the ITV hook-up (and linked in) as well as the other recent wins (adidas, Expedia, Danone, Sky, Boston Beer, Carlsberg, Kellogg’s, Globo) does seem to show that they have a real, serious and seemingly unique product / proposition.

Still don’t understand how they can scale? They talk about automation, but there’s still the need for 150 humans to watch and score each ad (if I understand correctly) - although they do talk about researching AI solutions - but I think that’s gonna be really hard to get to work anywhere near as well as humans.

So not at all clear to me how this can scale…

Although sounds like they have figured it out … still don’t understand how…

https://system1group.com/blog/what-happens-if-you-test-everything

How did Ad Ratings happen?

"It started with a very simple question from our Labs team – what’s the true market picture of advertising? If you tested every ad that aired, how many of them would be good, brand-building ads, and how many of them would be basically a waste of money?

We’d managed to automate our testing process so well that scaling it up to tens of thousands of ads wasn’t a problem. But as soon as we started getting the data in we realised how powerful it was."

1 Like

Hi Colin,

The 150 people are sourced from panel organisers Toluna, Prodege and NetQuest. I can’t find the quote in my notes, but I am pretty sure management said at an AGM the company has 100,000+ people registered through those suppliers. So SYS1 is not using the same 150 people every time, just 150 picked/volunteered at random (I presume) to meet the default ‘national representation’ or a specified client demographic.

The IT cost of maintaining and developing the Test Your Ad platform ought to be broadly similar whether the service is handling 10 tests or 1,000 tests a day. That is the ‘scalable’ attraction. Panel costs will of course rise in line with the volume of testing.

The automation refers to clients being able to upload adverts themselves, click a button and have a report produced within 24 hours. The bottleneck could be the panel supplier selecting the 150 people, but that does not stop the automation process I think.

I do wonder if SYS1’s follow-on Expert Guidance consultancy services are as scalable as the Automated Predictions. Surely some human input is required to give the client some advice on how to improve the ad.

Like you I am not sure how AI could provide an ‘emotional’ verdict on adverts!

Maynard

PS Long blog post here on SYS1’s FY 2021 results.

Just got around to reading your post on the FY 2021 results - excellent work IMO - it still looks intriguing and the rapid growth in Data Revenue as well as signing up some big names… (adidas, Danone, Sky, Boston Beer, Carlsberg, Kellogg’s, and Globo) seems quite promising…

But I’m a little wary of what seems to me to be a bit too much hype, for example “we have challenged ourselves to deliver them at1/100th the cost and 100 x faster than traditional methods”

Again, given they need actual humans to rate each ad, the above does appear to me a bit far-fetched - OTOH, any competitor is gonna have to deal with the same constraint.

But I’m still not convinced that it would be as hard, for a competitor to replicate what they’ve done, as they try to make out…

1 Like