2018 to 2021: enrichment plugs into the sequencer, identity becomes the filter's strongest signal

Between 2018 and 2021 the contact database, enrichment and the sequencer merged into one workflow, so a team could build a list, fill in every field and start sending from the same screen. Mailbox providers answered by leaning harder on identity: domain reputation, IP reputation, authentication and recipient feedback, read together. Who was sending started to count for as much as what was sent, and it has not stopped since.

How did data and sending become one workflow?

The clearest record of this shift is a public filing. On February 26, 2020 ZoomInfo filed to go public, a year after DiscoverOrg acquired Zoom Information. The filing described data on over 14 million companies and about 192,000 paid users who used it to find target accounts, monitor buying signals, craft the message and engage through automated sales tools. It also made the integration explicit.

The vast majority of our customers integrate ZoomInfo with their most-used CRM...

Apollo, founded in 2015, describes itself as a lead intelligence and sales engagement platform: the database and the sequencer as one product. By the end of the era the list was no longer a spreadsheet someone bought. It was a live feed, connected to the CRM and to the sending tool.

For a head of sales the gain was speed. Northwind fits your filter: B2B software, 20 to 80 people, hiring account executives. Enrichment returns Dana Reyes's title, a verified address, the tools Northwind runs. She is enrolled in a sequence before anyone on your team has read Northwind's website. The cost was that every competitor had the same speed and the same records. When every team pulls the same fields from the same vendors, Dana gets the same "noticed you're hiring" email from several of them in one week. The data stopped being an edge at the moment it became a feature.

What did the filters start reading instead?

In February 2019 Google said new TensorFlow models were blocking about 100 million additional spam messages a day in Gmail, on top of protections that already stopped more than 99.9% of spam, phishing and malware. Among the categories it could now catch, it named image-based messages, emails with hidden embedded content, and messages from newly created domains that hide a low volume of spam inside legitimate traffic. A domain registered last week and sending modest volume was now a pattern in its own right. Google also said every email carries thousands of potential signals, and that its models look at them together rather than judging one spammy-looking trait.

In July 2021 Gmail made BIMI generally available. A brand that enforces DMARC and validates its logo with a verified mark certificate can have that logo shown next to its mail. Google framed it as a way to increase adoption of strong authentication. Identity had become something the recipient could see.

The most detailed public list of what a provider measures about a sender is Google's own Postmaster Tools. Look at what its dashboards cover.

Why does a new domain look like a risk?

Because it has no history, and a filter treats a sender with no history with more scrutiny. M3AAWG, the industry's anti-abuse working group, puts it plainly in its current sender best practices: a new domain's reputation starts out unknown, "similar to bad," and a new domain sending a large volume shortly after registration may raise red flags. It recommends warming a new domain by raising volume slowly, about six weeks on average. That guidance is from 2026, but it describes a reaction that took shape in this era: Google's 2019 note on newly created domains is the same idea, seen from inside a filter.

Identity is accumulated. You cannot buy it on the day of the campaign. That is the practical lesson of era three, and it is why the next two parts are about what senders did to get around it and what the providers did in return.

What era three means for a head of sales in 2026

• Your enrichment is everyone's enrichment. The fields a database returns are table stakes. The reason to write has to come from somewhere they do not: a current, specific signal about the account. Part 9 describes how a signal is scored before it becomes a message.
• Treat the sending domain as an asset with a history. Know its age, its authentication status and its reputation before you add a single campaign to it.
• Read the identity rows before the copy. When replies drop, domain, IP and authentication are the first things to check. Part 12 lays out that order as a runbook.

A limit worth stating. Google publishes what it measures, not how much each signal weighs. "Who is sending started to count for as much as what was sent" is this series' reading of what the providers published and built in these years, not a ranking any provider has released.

Next in the series

Part 4 covers 2021 to 2023, when mailbox rotation, automated warmup and generative drafting cut the cost of a new campaign close to zero, and the filters learned faster. Part 2 covers the sequencer era that came before. The series page lists every part in order.