A lead list for a web design agency is easy to build and hard to keep. Most people get the collection right and lose the value in the six weeks afterwards, when nothing is re-checked and nobody removes anyone.
Five stages, of which the last two are where the work actually is.
1. Define what a good prospect looks like
Not a paragraph about ideal customers — three or four checkable conditions, so that building the list is mechanical rather than a judgement call each time.
For local first-website work the useful conditions are: a category with a real gap, a review count above the category norm, a rating above 4.0, and a location you can reach. Everything else is refinement.
The mistake is defining this by industry alone. "Small businesses in Jaipur" is not a profile; "hardware and electronics retailers in Mansarovar and Sanganer with 40+ reviews and no website" is one, and it produces a list you can work on foot next Tuesday.
2. Source it
Map listings, for the reason that runs through this whole corpus: it is the only source that contains a business regardless of whether it has a website, a domain or an email. B2B databases are built around a domain and a named person, and a business with no website has neither.
Whether you gather it by hand, use a tool, or buy an export is a volume question rather than a quality one. Below fifty leads a week, by hand in Maps is genuinely correct and costs nothing.
3. Qualify before it enters the list
The single highest-leverage habit, and the one most people skip because it feels like throwing away work.
A list of eight hundred unfiltered businesses is worse than a list of a hundred qualified ones — not just less efficient, actively worse, because the close rate on the unfiltered version teaches you the wrong thing about your pitch and your market.
Qualify on what the listing shows: the website field is really empty, the review count clears the category's norm, the rating is above 4.0, and the business is in an area you will actually visit. Four checks, under a minute each.
| Stage | Typical count | What happens here |
|---|---|---|
| Collected | 800 | Everything in the category and city |
| After website check | 300 | Genuinely no site, or social-only |
| After review and rating | 120 | Proven customers, no obvious problem |
| After geography | 100 | Reachable in a clustered round |
Illustrative. The shape matters more than the numbers — most of what you collect should not survive.
4. Deduplicate on the ID, never the name
Duplicates arrive constantly and from a specific place: searching by neighbourhood means businesses near a boundary appear in two searches. That is not a flaw in the method, it is the cost of getting past the result cap, and it has to be handled at ingest rather than later.
Deduplicate on the place identifier rather than the business name. Names collide far more than people expect — chains, franchises, family businesses using the same surname, and two unrelated "Sharma Electronics" on opposite sides of a city. Names also differ trivially between records: an extra full stop, "and" versus "&", a suffix.
The failure this prevents is the expensive one in a local market: two people from your agency approaching the same shop in one week, or worse, you approaching a business you already pitched two months ago and forgot.
5. Maintain it, which is the whole job
Collection is an afternoon. Maintenance is the reason a list is worth anything in month three.
Re-check the website field on a cycle. This is the field that decays in the direction that embarrasses you — a business that built a site is a business whose problem you can no longer describe. Published B2B decay runs 22–30% a year; treat a list older than three months as needing a re-check before a round.
Remove people properly. A no should come off, or be marked so clearly that it never enters a round again. Lists that only grow stop being trusted, and a list you do not trust is a spreadsheet.
Record what happened, in one line. Not a CRM — a column with the date and what they said. "Nov 12, said maybe after Diwali" is the difference between a follow-up that works and one that reads as a stranger's cold message.
A list, already filtered
Live from the same index the product searches — 1,917 businesses in Indore currently have an active Google listing and no website.
Novelty Market
The Waterfall Restaurant Indore
PRINCE IT SOLUTION
Shree Chotiwala Restaurant
Skip to the qualified hundred.
Category, city, review count and rating — filtered before it reaches your list rather than after.
Build a listFrequently asked questions
How do I build a lead list for a web design agency?
Define three or four checkable conditions rather than an industry, source from map listings, qualify before anything enters the list, deduplicate on the place identifier, and re-check the website field on a cycle. The last two are where the work is.
How many prospects should a lead list contain?
About 100 qualified prospects supports four builds a month. It matters that they are qualified — an unfiltered list of 800 is actively worse than a filtered 100, because its close rate teaches you the wrong thing about your pitch.
How do I deduplicate a lead list?
On the place identifier, never the business name. Names collide constantly — chains, franchises, two unrelated Sharma Electronics — and differ trivially between records. Duplicates arrive because neighbourhood searches overlap at boundaries.
How often should I refresh a lead list?
Treat anything older than three months as needing a re-check before a round, because the website field decays in the direction that embarrasses you. Published B2B data decay runs 22–30% a year.
What is the most common mistake with lead lists?
Never removing anybody. A list that only grows stops being trusted, and a list you do not trust is a spreadsheet. Declines should come off or be marked so clearly they never re-enter a round.
Related reading
where the listings come from · the qualification checks · why the field decays · how large the list should be
