I’ve been testing Twitter ads for the last ten years, and here’s what I’ve learned.
The magic of Twitter campaigns
Engagement campaigns (boosted posts) are the only worthwhile campaign objective.
For example, in growing small Twitter accounts, don’t run follower campaigns with only a 0.3% follow rate.
Run engagement campaigns on high-value posts, giving you a 25% follow rate. Your cost per follower will be almost 100 times better.
Only follower look-alikes are worth targeting. Don’t use interest or demographics, which are faulty and sparse.
I typically get a 10% engagement rate on my ads, meaning 1 in 10 people will click. It is 50 times better than average ads. My video campaigns get a 50% or higher watch rate. Not because my video is killer or edited fancy, but because I target the followers of the people in the video.
People will watch those they’re already familiar with.
Think about what this means for your podcasting strategy, especially to repurpose clips of people who have interviewed you and vice-versa. Thus, leverage your podcast and those you’ve been on.
Twitter has been notoriously bad for referral traffic (website clicks). You could choose website clicks or slightly better visits (pixel fires to confirm visits).
But you’ll still be better off with an engagement campaign. The CPC (Cost per Click) will be slightly higher, but it’s worth the extra engagement and organic reach.
Do you think this will change with Elon making adjustments on the platform?
It will get better!
Has immediate ad monetization on Twitter ever worked?
This pinned tweet got half a million impressions, 80% of which were paid.
Of the 50,000 interactions, 98% were from paid, but only an admin would be able to spot this.
Whatever works on Facebook, replicate to Twitter with the same content and targeting combos.
Boost for a dollar a day, except you must be extra careful on tightening audiences down, since the Twitter algo isn’t as smart. It will waste your money on giant audiences at high auto bids if you let it. I manually bid to 2 cents per engagement or a penny a video view.
I don’t see anyone doing this except for Nick Venezia, who is the smartest advertiser in Twitter I know.
Last week, Twitter announced they’re cutting off API access to providers like Datasift, forcing everyone to go through GNIP, the data reseller service they purchased.
Full disclosure, we were a customer of GNIP (not Datasift), and were forced to pay between $5,000 to $20,000 per month for access to Twitter data. And Datasift via Mark Suster is disappointed that Twitter didn’t make the split amicably. People are wondering if Facebook will yank everyone but Datasift’s access in retaliation.
My bet is no.
Why does this matter to you?
If you want to grow your community, perform social analytics, or run ads– and if that has something to do with Twitter– then you’ll need Twitter data to make decisions. Twitter folks, if you’re reading this, making decisions = spending money. Without analytics, brands can’t understand where Twitter campaigns are performing.
They are in the dark.
So while forcing everyone to go through GNIP may yield Twitter a few million dollars, this is penny-wise and pound foolish. Twitter’s potential advertising business is not worth maiming for data access fees.
That’s why Facebook has never charged for their APIs and likely never will. Facebook even made their ads API open to everyone. Anyone can get basic access without having to jump through hoops. Facebook has even free training in Facebook Blueprint, Facebook for Business, the various Facebook developer garages, free training materials, and so forth. All of this is free– since it builds the ecosystem of developers, brands, and consumers.
Google doesn’t charge for Google Analytics or the API because they know that stronger intelligence leads to more ad dollars. In fact, Google has been investing heavily in free tools for marketers: Think With Google, Google Tag Manager, Google Webmaster Tools, Google Translate, Google My Business, Google Keyword Planner, Google Trends, and so forth.
If I were Dick Costolo…
I’d find out what Datasift customers have been doing with the Twitter data. If the metrics lead to analysis and then action, they’re saving pennies by throwing away dollars.
The last thing customers want to do is get into the business of building data infrastructure. I know this. The very reason we play in the data space is because of the complexities of dealing with multiple APIs from multiple vendors.
Not an option for small businesses and a shaky play for even enterprise-level brands to have their data strategy reliant upon just a GNIP or Datasift. The Adobes of the world will eventually release something here, likely via acquisition, while you see other third parties building logic layers (data models) via platforms (like Pylon from Datasift) so expensive that if you have to ask, you’re missing a few zeros in your budget.
Maybe the marketing automation or tag management players will come in to save us, for the geeks that care to discuss. But for the rest of us, we mourn the slow irrelevance and demise of Twitter at their own hand.
Maybe Twitter with smart folks like Adam Bain will wake up to see that analytics (or at least data) needs to be free.
Nobody knows, despite what they may claim, which accounts are real versus robots.
Even twitter doesn’t know.
Now that twitter has come out with organic reach reports, the house of cards comes crashing down– for the 3rd party tool vendors and for people using these reports.
One of our friends runs social for one of the largest agencies in the world– his screenshot above.
Instead of the 27,000 reach from Sysomos, the actual reach was 1/18th that.
Because twitter doesn’t provide organic impressions in their API, even if you pay for the reporting, no tool vendor can accurately report here.
We’ve told twitter a few times that they need to get organic reporting right to be able to drive ad sales.
Certainly hiring a ton of sales people after the IPO is a normal thing to do. But we recommend educating clients instead of just selling to them.
Some people will be shocked at getting numbers 1/18th what they used to report.
Will they restate historicals so it doesn’t look like their current performance stinks?
Will they keep using the old method (assuming 100% coverage), since those numbers are “better”?
Will they decide that reporting impressions is a shell game, abandoning reach for engagement metrics?
Will they be unaware or apathetic, saying social metrics are hocus pocus anyway, since revenue is what counts?
