In 2017 I was an outbound BDR at a retail and restaurant analytics company, targeting North America enterprise accounts from Bengaluru. My job was to find companies that were likely to buy — the right size, the right vertical, at the right moment — and get someone on the phone.
The tools available to do that were a fraction of what exists today. Not slightly worse. Fundamentally different in capability.
Nine years later, I build the systems that do this work at scale. The contrast is worth documenting — not as nostalgia, but because understanding where the friction used to be explains why the current tooling is so significant.
What finding accounts actually looked like in 2017
The list problem
The starting point for any outbound campaign was a list. In 2017, building a good list meant:
- Exporting from ZoomInfo or DiscoverOrg — which existed but were much more limited in data freshness and coverage than they are now
- Cross-referencing LinkedIn manually to verify that the contact was still in the role
- Checking the company website to confirm they were still in business and hadn’t pivoted
- Guessing at email format and running through a validator
A list of 200 reasonably verified accounts took a serious chunk of the week. And “reasonably verified” still meant 20-30% bounce rates, wrong numbers, and contacts who’d moved on months ago.
There was no real-time enrichment. There was no automatic deduplication against what was already in the CRM. There was no scoring. You built the list, you worked the list, you discovered its problems as you went.
The timing problem
Timing is everything in outbound. Calling a company that isn’t in a buying cycle is a waste of everyone’s time. Calling one that just started one — that’s where pipeline comes from.
In 2017, finding buying signals required manual effort that most BDRs didn’t do because it wasn’t scalable. I built my own system using Feedly — RSS feeds from trade publications, company news pages, and press release wires, organised by vertical. Every morning before calls I’d scan for mentions of expansion, new leadership, funding, construction, new store openings — anything that suggested something was changing at a target account.
It worked better than cold volume dialling. But it was entirely manual. It covered maybe 50-60 accounts actively. It missed things constantly. And it was entirely dependent on companies issuing press releases or being covered by trade media — which smaller private companies often weren’t.
Intent data as a category barely existed in a usable form. Bombora launched in 2014 but enterprise pricing made it inaccessible for most BDR teams. The idea that you could see which companies were actively researching your category of software in real time — that wasn’t a practical reality for most outbound teams in 2017.
The firmographic problem
Even basic firmographic filtering — find me companies in retail with 500+ employees and $100M+ revenue — was less reliable than it sounds. Revenue data in particular was largely estimated, often wildly inaccurate for private companies, and months to years out of date.
Employee count was more reliable but still imprecise. Headcount from LinkedIn was the more accurate source, but pulling it for hundreds of accounts wasn’t automated — it was manual research.
The result was that ICP filters were coarse. You’d define a target profile, export a list that roughly matched it, and accept that maybe 40-50% of accounts genuinely fit once you looked more closely. The rest you’d discover didn’t qualify after you’d already spent time on them.
What the same problem looks like now
The gap between 2017 and now isn’t incremental. The underlying capability has changed category.
Account universe definition is now systematic
ZoomInfo‘s current database covers 100M+ companies with NAICS/SIC codes, verified employee counts, revenue bands, technology stacks, and direct contact information. Filtering to a precise ICP — healthcare systems with 1,000+ employees, operating in North America, running a specific ERP — returns a list in seconds rather than days, with meaningful confidence in the data accuracy.
Clay can take that list and automatically enrich each account from 50+ data sources simultaneously — pulling LinkedIn headcount trends, recent news, job postings, technology changes, funding events — and return a scored, ranked account universe with a research brief per company. What took a BDR team a week in 2017 takes an n8n workflow 20 minutes today.
Buying signals are now detectable at scale
The Feedly system I built manually in 2017 now exists as an automated layer. The signals I was trying to surface — executive hires, facility expansions, funding rounds, new construction projects — are now captured through multiple automated channels:
- ZoomInfo Scoops — captures company announcements, leadership changes, expansion signals, and purchasing intent, updated continuously
- Intent data — Bombora, G2, and similar platforms now show which companies are actively researching specific categories of software, based on content consumption across thousands of B2B sites
- Job posting signals — a company actively hiring capital project managers or construction directors is signalling CapEx activity more reliably than any press release
- Building permit data — commercial permits are public records, now aggregated by services like BuildZoom, giving direct evidence of committed construction spend
- SEC filings — for public companies, the cash flow statement on SEC EDGAR discloses CapEx investment in real numbers, audited and exact
The difference is not just coverage. It’s that these signals can now be monitored automatically across thousands of accounts simultaneously and fed into a scoring system that surfaces the warmest accounts to sales without any manual research step.
AI has changed the research layer entirely
The step that consumed the most BDR time in 2017 was manual account research — reading company websites, scanning LinkedIn, piecing together a picture of whether an account was worth calling and what angle to use.
Claude and similar models can now do this in seconds. Feed in a company name, domain, and a structured prompt, and get back a research brief covering what the company does, why they might need your product, who the likely buyers are, what signals indicate current buying intent, and what the personalised outreach angle should be. Not a generic summary — a brief framed against a specific ICP and value proposition.
This is the research step that I was doing manually for 50-60 accounts in 2017 using Feedly, now running automatically across thousands of accounts in an n8n workflow.
The buying committee is now mappable before first contact
In 2017, identifying the right person to contact at a target account was a research task that could take 20-30 minutes per company. LinkedIn search, cross-reference against ZoomInfo, verify current role, find a second contact in case the first didn’t respond.
Today, ZoomInfo contact search filtered by title, seniority, department, location, and tenure returns verified contacts with direct dials and email addresses for the entire buying committee in under a minute. Clay can then enrich each contact with recent LinkedIn activity, thought leadership signals, and work history context — giving BDRs genuine personalisation material before first contact.
What hasn’t changed
The tools have changed. The underlying problem hasn’t.
Finding the right account at the right time still requires someone to define what “right” means — what the ICP actually is, which signals genuinely predict buying behaviour for your specific product, what the trigger event means in the context of your value proposition. The tools surface the data. Humans still have to define the logic.
And the BDR who picks up the phone still has to have a conversation worth having. All the account intelligence in the world doesn’t change the fact that the first 30 seconds of a cold call determines whether there’s a second conversation. The research layer creates the conditions for a good call. It doesn’t make the call.
What’s changed is the ratio. In 2017, maybe 10% of a BDR’s time was actually selling. The rest was list building, research, admin, and chasing bad data. The tools that now exist — when properly configured into an automated pipeline — can flip that ratio significantly. More time on what actually moves pipeline. Less time on everything that used to be a precondition for doing the job.
That’s the real shift. Not that AI has replaced BDRs — it hasn’t. But the BDR who understands how to use these tools, or the MOps team that builds the pipeline behind them, creates a structural advantage that compounds with every account worked.
I’m Ajay Kumar — Senior Marketing Operations Analyst based in Bengaluru, with 9 years in RevOps spanning outbound BDR work and full-stack MarTech architecture. I now build the automated account intelligence pipelines that do at scale what I used to do manually. Find me on LinkedIn.
Leave a Reply