As an NBA fan and marketer, there are two “jump shots” I’m longing to have back. While one’s return is only a question of when, the other can seem lifeless and futile. Whether the recency of Easter weekend has me filled with a resurrection optimism or there’s something more to it, I believe there’s reason for hope. Why? I can think of a few possible solutions to replenish at least some of what was lost.
I don’t claim to be an expert in this area. In fact, let me be clear; I’m not an expert in clickstream. However, if I can envision some realistic solutions, I’m sure others who are more equipped could do the same to an even greater extent.
If you’re reading this, chances are you’re well aware of what happened to Jumpshot earlier this year. If not 1, here’s some backstory.
If you want to skip this and jump to my four solutions, click here.
Avast and Jumpshot: From harmonious marriage to lightning-fast divorce
Avast is the world’s largest antivirus software. In May 2015, Avast publicized its new partnership with Jumpshot. This announcement made aware Jumpshot would be aggregating and anonymizing Avast’s user-behavior to sell as market research. Unlike platforms built on similar data, advertisers would not be able to directly target users in the Avast database. It also acknowledged the potential concerns over privacy and outlined the steps Avast and Jumpshot would take to address them.
This partnership proved successful in its first four years, picking up a diverse set of clients like large DTC brands (e.g., Revlon, Home Depot, TripAdvisor) and SaaS marketing products (e.g., Hitwise, Moz, Ahrefs, SEMrush). For SEO tools in particular, this came at the perfect time, as Google’s keyword data was growing less and less reliable.
Starting in mid-2019, however, things began to take a sharp turn.
Timeline leading up to the end of Jumpshot
- June 2, 2019 – The Wall Street Journal mentions Jumpshot as a critical data source in the Google Antitrust DOJ probe. This was the first time many in the general public would have learned about Jumpshot’s existence.
- July 22, 2019 – Ascential plc buys 35% stake in Jumpshot to expand its global footprint and into other verticals. 2
- December 2, 2019 – Wladimir Palant, founder and CEO of Adblock Plus, writes about Avast’s practices and successfully gets Firefox and Opera to remove the Avast and AVG extensions from its add-on stores. 3
- December 9, 2019 – Avast CEO fights back in Forbes article by doubling down on the measures taken to protect the privacy of its users.
- December 11, 2019 – Senator Ron Wyden questions Avast’s practices publicly and privately.
- December 18, 2019 – Avast and AVG extensions are removed from the Chrome Web Store.
- December 20, 2019 – Avast and AVG extensions are added back to the Chrome Web Store after agreeing to stop passing its data to Jumpshot. All Jumpshot data is now exclusively being passed via local software installs.
- January 27, 2020 – Vice publishes scathing review of Jumpshot and Avast, citing concerns around opt-in ignorance and the potential to deanonymize the user data.
- January 30, 2020 – Avast ends relationship with Jumpshot.
The Vice article triggered a 28%, 4-day free fall in Avast’s stock
The events above are annotated in this historical timeline of Avast’s stock. Notice anything unusual about the event right before the plug was pulled on Jumpshot? I’m convinced it was financial pressure and nothing more that caused Avast to cut ties with Jumpshot.
Alternative clickstream ideas worth considering
I’m not here to weigh in on the ethics of what Jumpshot and Avast were doing. I have my own opinions, but I’d rather you make up your own mind. However, what if this isn’t an all-or-nothing scenario? What if we could explore alternatives that give marketers and SEO professionals at least some of the Jumpshot value back, while being much more rigorous on the privacy front?
Here are four ideas I have to do just that.
1. Add noise to Avast’s data
I’m leading off with the idea that just might be my favorite but least likely to happen. The crux of the Avast/Jumpshot criticism, beyond making the opt-in process and its implications more explicit, is the ability for bad players to deanonymize the user data. The fear is that with each device having its own unique ID and all interactions being timestamped, one could match Jumpshot’s data with a source that includes personally identifiable information (PII) to reveal their true identity.
By generalizing the timestamps and clustering devices, you could effectively remove these risks.
Generalize the timestamps
Jumpshot’s timestamps were down to the millisecond. Generalize them to an hourly or even daily specificity, and the ability to deanonymize this data goes down significantly.
You may have heard of Google’s Privacy Sandbox, which is its proposed solution to killing third-party cookies. At its core, the Privacy Sandbox is meant to add just enough noise to browser behavior to protect individuals, while still allowing for aggregated targeting for advertisers. This idea is a spinoff of that.
Why not group all Avast devices in cohorts of five to ten? The cohorts could be random or perhaps clustered by similar behavior.
This would still give SEO tools like Ahrefs, Moz and SEMrush enough data to refine monthly search volume estimates and discover new keywords. It would also give marketers enough to generate valuable market research.
2. Bring back exact match search volume
This solution does little to replace the market research or expose Google’s thumb on the scale. It could also dampen a unique selling proposition from SEMrush, Ahrefs and Moz (more accurate volume estimates) but would be extremely beneficial for the SEO community at large.
Google used to provide average monthly search volume for specific keywords. Now, these estimates often represent a grouping of keywords 4. The grouping is entirely inconsistent and often disregards differing search intents, which could lead to ill-informed decision making.
I suggested it to John Mueller here, but alas, the separation of organic and paid search at Google stopped it in its tracks.
The grouped keyword volumes can lead to bad decisions, especially because what does and doesn't get grouped is highly inconsistent, and the grouping regularly pulls in irrelevant terms.
— Tylor Hermanson (@MyNameIsTylor) March 3, 2020
3. Resurrect Google Correlate
Google Correlate was discontinued in December 2019. I wrote a long guide to Google Correlate that has since been taken down on Search Engine Journal, but you can find it here thanks to our friends at the Wayback Machine 5.
While Google Correlate has its blind spots, and spurious correlations were a huge risk from such an incredibly large dataset, it at least helped give us unique insights into how people searched cyclically, after important events, as well as geographically. Whether you were a marketer, data scientist or anthropologist, Google Correlate had something for everyone.
4. Be transparent with the remaining clickstream sources
I’m going to be honest; the three solutions above are probably more unlikely than I’d care to admit. However, I have some hope for this next one. Let’s explore the companies still leveraging clickstream and consider the value of them being more transparent with their user bases.
SimilarWeb remains the sole widely-known clickstream provider. To be clear, it has some major differences when compared to Jumpshot 6. Here are a few:
|Offering||Dashboards of aggregated data||Raw device-specific data|
|Volume||Undisclosed||100 million users|
|Source||Undisclosed (multiple sources)||Avast & AVG|
Assuming SimilarWeb gets its data via above-the-board methods, why not be more transparent around it? It would help current subscribers understand any potential biases that could present themselves in the data. Plus, if they implemented my grouping and timestamp generalization ideas, they could suddenly generate more revenue by meeting the need previously met by Jumpshot.
Moz, Ahrefs and SEMrush
All three of these tools used to leverage search data from Jumpshot. They all used it to refine search volumes, and at least Ahrefs used it to discover new keywords 7.
Similarly, if they shared their sources, it would help SEOs understand what biases may be present in the data. It could also allow marketers to engage with some of those sources directly for market research. I’ve reached out to all three companies and have been underwhelmed with their responses. It’s clear they want to keep their remaining sources close to their chest, but can I really blame them?
But let’s be honest. SimilarWeb and these SEO tools probably aren’t going to be more transparent. There’s too much to lose. After all, this situation was at least partially due to Avast and Jumpshot’s transparency in the first place.
Shouldn’t we just move on from clickstream altogether?
Perhaps, but I’m not quite there yet.
- I find keyword-level search volume still an incredibly useful metric.
- I think having visibility into Google’s hoarding of clicks is something worthwhile.
- I think search behavior can be an incredible market research tool. Like Seth Stephens-Davidowitz says, it’s digital truth serum.
But to be clear, however, if we can only resurrect one ‘jump shot’, my heart is with hoops.
or if you could use a pre-pandemic refresher↩
AVG merged with Avast in 2016 and delivered clickstream data to Jumpshot the same way. ↩
and if you don’t pay Google enough money, the volumes are in large ranges↩
beyond, you know… existing↩
That’s why at the time of writing this in late April, Ahrefs doesn’t show any new keywords past December 2019↩