How Much Does Bark Charge for Leads? Credits, Cost per Job, AI Replies

Bark has no flat price per lead. Each lead shows its credit cost before you respond, driven by the service, the job's value and local demand, and credits expire after three months. The number to manage is cost per job won, which quicker, better-screened replies, drafted with AI and checked by you, can bring down.

A credit price says what a conversation costs, not how many conversations end in a job. Bark's terms return nothing because a customer goes quiet or because you lose the quote, so every slow, vague or badly aimed reply is money spent for no result. That makes jobs won, the number you divide by, the part a tradesperson can change.

Follow me on Instagram@sagnikteaches

What Bark's pricing page and terms commit you to

You see leads for free from the moment you sign up and pay only to contact the customers you pick. Credits are bought in advance, in packs, and each lead carries a credit cost that Bark shows before you respond. The pricing page says that cost depends on the service, the value of the job and local supply and demand, so a small repair and a full refit are priced differently, and the same kind of lead can cost a different number of credits as demand moves. There is no commission on jobs you win, and the one-off fee gets you the customer's phone number and email, with follow-up messages after that costing nothing extra.

Connect on LinkedInSagnik Bhattacharya

Two kinds of request reach you, and the terms treat them differently. A general request (the terms call it a "General Bark") can be sent by the customer to up to five professionals they have not named, so you are competing from the first message. An "Enquiry" comes from a customer who contacted you directly through Bark Marketplace. For a general request you redeem credits to receive the contact details, and the site tells you how many credits that takes. For an Enquiry, credits come off at the moment the customer acts: when they send a direct message, submit a booking or call-back request, or place a call. That second rule changes your routine, because you have already paid by the time the message lands.

Subscribe on YouTube@sagnikteaches

Wording can differ between Bark's markets, and Bark edits its terms, so read the ones in your own account before relying on any line here. I read the pricing page and the professionals' terms in October 2026. The lead-settings detail further down comes from extracts of Bark's help centre, whose labels may differ on your screen.

One month of leads, priced stage by stage

Cost per job won needs four numbers, and all four sit in your own records rather than on Bark's site:

• Cost per lead answered. Pack price divided by the credits in the pack, multiplied by the credits that lead cost. Work it out from the receipt, because Bark can change pack prices.
• Reply rate. Customers who answered you, divided by the leads you paid for.
• Quote rate. Replies that became a site visit or a priced quote.
• Win rate. Jobs won, divided by the leads you paid for. This is the one that sets cost per job.

Cost per job won is the total spent on leads divided by jobs won. Here is the sum for an illustrative two-person painting and decorating firm. Last month it answered 24 leads, and its receipts work out at an average of $14 a lead, so it spent $336. Nine customers replied, six led to a quote, and two became jobs, each leaving about $620 in gross profit after materials and paid labour.

The cost per job won is $336 divided by 2, which is $168, or 27% of that $620 gross profit. A table of where the $336 went is more useful than the headline figure:

Nearly two-thirds of the spend, $210 of $336, went on customers who never answered. That is where AI earns its place on a lead platform. It will not win a quote for you, but it can help you avoid paying for some of the 15 silent leads and make the first message easier to answer for the nine who did reply.

Here are four versions of the same month. Every movement is an assumption to test against your own log, not a result anyone should expect.

Version B costs nothing to try, since declining a lead is free: the pricing page says you pay only to contact customers you want to work with. Version C costs a few minutes a lead. Version D is the combination the rest of this page builds towards.

How many wins make a lead worth buying

Turn the sum round and it becomes a purchasing rule. The win rate you need equals the cost of a lead divided by the most you are willing to pay to win a job. Choose that ceiling before you look at any lead, and as a share of gross profit rather than of the job price, because gross profit is what pays for the credits.

Take the painters' $620 gross profit and a ceiling of 15%, which is $93. At $14 a lead they need to win 15% of the leads they answer, about one in seven. Last month they won 2 of 24, or 8%, so they were paying $168 for a job they should have paid $93 for. Version D wins 3 of 17, or 18%, and clears the ceiling.

This table gives cost per job won for a range of lead costs and win rates. Swap in your own lead costs before you rely on a row.

Read across the $30 row against a $93 ceiling and the answer is plain: you need a win rate of about one in three. If your own record says one in twelve, a $30 lead is not worth buying however good it looks. Bark's page notes that credit costs follow the value of the job, so the expensive leads are usually the bigger jobs; compare them against the gross profit of that kind of job, not your average. If you buy leads from more than one source, Thumbtack, Angi and Bark compared on cost per booked job runs the same sum across all three.

Screening a lead before you spend a credit

The lead list shows the request itself, but not the customer's contact details, until you pay. Search extracts of Bark's help centre (its pages would not load for me) describe a Lead Preferences area with three parts: the services you want, your locations (a radius, a travel time or an area you draw on a map) and filter questions for each service, which narrow leads by how the customer answered them. The same extracts say preferences stay until you change them, while filters on the leads page reset when you leave it, so check both on your own screen. Set the permanent ones first, so the weakest leads never reach your list.

After that, a five-point sheet takes thirty seconds a lead. Each check scores 0 to 2, and the assistant's job is to read the request and mark what is missing, not to decide.

My rule of thumb is to buy at 7 or above, buy a 5 or 6 only when your week is quiet, and decline anything below 5. Lead 1 is an easy purchase. Lead 2 is the typical silent lead from the month above. Lead 3 scores 6 but fails the one check that cannot be fixed, so I would skip it unless you are happy to travel.

The five checks change a little from trade to trade, but the idea stays the same:

• Plumber: scope means a fault or an installation, the number of fixtures, and a photo of the boiler or the fitting with its model plate visible.
• Gardener: lawn and bed sizes, how often the work repeats, and one photo of the whole garden.
• Tiler: floor or wall area, tile size and material, and who is supplying it.
• Electrician: whether it is a fault, an addition or a rewire, the age of the property, and a photo of the consumer unit if the customer can take one.

You are screening a customer request for a two-person painting and decorating firm before we pay credits to respond.
Score it out of 10: five checks, 0 to 2 points each. The checks are: scope clear, start inside the next four weeks, photos or measurements given, inside our travel radius (we work within [distance] of [base]), finish and budget look like a fit.
If the request does not say, score that check 0. Do not guess.
Then list what is missing and the one question that would settle the biggest gap.

Request (details only, with name, phone and email removed):
[paste here]

A reply that takes four minutes to produce

Do the slow part once. Write a fact sheet for the business: what you take and what you refuse, how you price, your earliest start dates, what you need from a customer before you can quote, and any proof you can state truthfully, such as how long you have traded. Save it in a ChatGPT Project or a Claude Project, or in a note you paste each time, so each lead needs only the job details.

Then run the same five steps on every lead you buy:

1. Buy the contact only if the sheet says so. Thirty seconds.
2. Paste the job details into the assistant. Leave out the customer's name, phone number and email. Bark's terms bar making a customer's contact details available to a third party without Bark's written agreement. Whether an AI provider counts is a question for Bark, and the safe routine needs no answer.
3. Read the draft against a short checklist. One specific detail from the request quoted back; one question the customer can answer in a few words or a photo; a start window; a time for the firm price; no figure you have not worked out; nothing you could not prove; under 110 words.
4. Send it yourself. A person should press send. Bark's terms ask you to communicate clearly and promptly through the chat thread, and an assistant cannot vouch for what it writes in your name, so keep a human in the loop.
5. Log the time sent in your tracking sheet and set the follow-up reminders.

You write first replies for a two-person painting and decorating firm. Use only the fact sheet below. Never invent experience, prices, qualifications, insurance or availability.

Fact sheet:
- We do: interior repaints, wallpaper removal, woodwork. We do not do exteriors above two storeys.
- Pricing: fixed price per room after photos or a visit. We never quote over the phone.
- Earliest start: week of [date]. A typical living room takes 1 to 2 days.
- We need from the customer: two photos per room, whether furniture is being moved, whether colours are chosen.
- Tone: plain and friendly. No exclamation marks.

Customer request (job details only):
[paste here]

Write a first message under 110 words that quotes back one specific detail from the request, asks one question that can be answered in a few words or with a photo, gives a start window, and says when they will receive a firm price. Sign off as [your name].

What the assistant gave back, and what I would change

Say the request read: "Repaint living room, hallway and stairs in a four-bedroom house, woodwork too, we will move the furniture ourselves, hoping to start within a month." An illustrative draft:

Hi [first name], thanks for the details of the living room, hallway and stairs. Stairs take longest because of the cutting in around spindles and skirting, so two photos would help: the stairwell from the bottom and the tallest wall. With those I can send a fixed price by tomorrow evening. I could start in the week of [date], and a job this size usually runs four to five days. Is the woodwork staying its current colour or being repainted too? [your name]

Three edits before sending. First, "four to five days" is the assistant extrapolating from the one-to-two-day line in the fact sheet, so replace it with your own estimate or delete it. Second, "by tomorrow evening" is a promise, so keep it only if you will honour it. Third, and this is the one that costs jobs, the closing question was already answered in the request, which said "woodwork too". Asking it again tells the customer you did not read their message. Swap it for something the request does not say: "Are the stair spindles already painted, or bare wood?"

Compare that with a generic reply such as "Hi, I can help with this. Call me on [number] for a free quote. Thanks." It uses the same credit and costs the same few seconds, but it asks the customer to do the work, and with up to four other professionals writing to the same person, that is a poor bet.

Three free follow-ups, and when to stop

Messages after the first are free on Bark's pricing page, so the cheapest improvement to your win rate is a short follow-up pattern. Each message should add something the previous one did not. Repeating "just checking in" three times is how a lead goes cold.

Ask the assistant to draft each one from the original request plus the thread so far, with the instruction "add one new piece of information, no repeated questions". Stop after the third message. A customer who has not answered three clear messages is not going to be rescued by a fourth, and the time is better spent on the next lead. If the volume grows to the point where you want the follow-ups sent for you, what it costs to automate lead follow-up with AI is the sum to do before you commit to a tool.

Bad contact details and the first-pack guarantee, in practice

Two parts of the terms can return credits, and both reward people who keep records. Credits that come back are reinstated as credits, not paid out in cash.

The seven-day report. For a general request, you can ask for credits back within seven days of receiving the contact if the phone number is invalid, the email bounces, or Bark can verify that the details belong to someone other than the person who submitted the request. For an Enquiry the grounds are narrower, and the seven days run from when the Enquiry was made: it must look like spam or fraud, or Bark must be able to verify that the details are not the submitter's. A routine that fits: on the day you buy, call the number once and send your first message to the email address; screenshot the failure if either fails; report the same day.

The first-pack guarantee. The pricing page tells new businesses that Bark will return credits from the first pack if you do not win business, with no questions asked. The terms, which call it the Get Hired Guarantee, are more specific. As I read them, it applies once, to the first pack only, and only to credits spent on general requests; you must have used all the credits in it; you must have no contract with any of those customers; and you must wait three days after the last credit was redeemed before you ask. If all that holds, Bark adds the same number of credits as the pack contained.

A timeline shows how it works in practice. Say a newcomer buys the first pack on the 1st and spends the last credit on the 20th. Counting three days from that last redemption, with a day's margin, the request goes in on about the 24th. The conditions require that you have no contract with those customers, so any job won from those leads, even a small one, ends the claim. Keep a simple log from the first lead: date bought, credits spent, customer reference, messages sent, outcome. The log lets you ask the moment the waiting period is up, with the facts to hand, and an AI assistant can turn the log into the short request. Bark also describes a separate guarantee for its Marketplace subscriptions, with its own conditions, which is worth reading if you are offered that route.

Sizing a pack to the three-month clock

The three-month expiry turns pack size into a forecast. Take the painters, who answer about 17 leads a month once they are screening, giving 51 leads in three months. A pack that covers 70 leads would leave about 19 unspent when the credits expire, and at $14 a lead that is $266 gone with no way to get it back, because the terms say expired credits are neither refunded nor extended.

The opposite mistake is smaller but real. Buying small packs more often can cost more per credit if Bark discounts bigger packs, and it can leave you without credits on the day a good lead arrives. I would compare the pack prices on the screen, work out the price per credit for each, and buy the pack nearest to what the last three months of answered leads suggest. Then check the figure again every time, because the price shown at purchase can change.

If you buy through a monthly arrangement rather than packs, the terms say credit subscriptions can be cancelled on 30 days' notice, the monthly number of credits cannot be reduced while the subscription runs, and unused credits roll forward month to month. In that case the sums to watch are your cost per job and whether the monthly credits match the leads you actually answer, because you can stop the subscription but not shrink it.

Where AI replies lose jobs on a shared lead

Four failure patterns are easy to fall into with AI-drafted replies on a lead marketplace, and each has a simple counter.

• Everyone sounds the same. If up to five professionals write to one customer and all use an assistant with a generic prompt, the messages converge on the same friendly paragraph. The cure is the fact sheet: specifics about how you price, what you need and when you can start are what a customer can compare.
• Invented credentials. Assistants fill gaps. "With over ten years' experience" appears in drafts when nobody supplied it. Bark's terms already bar false or misleading information about who you are, and a customer who catches an invented claim will not hire you. Say in the prompt that nothing outside the fact sheet may be stated, then check anyway.
• A price you have not worked out. A draft that offers "from $400" commits you. Keep the fixed-price-after-photos line in the sheet so the assistant never offers a number.
• Automated sending. Some tools can send messages without a person. I would not use that on a lead you paid for: the terms ask for clear, prompt communication from you, and a human read costs ten seconds. The same question comes up on another platform in Thumbtack's own front desk against a third-party auto-responder.

A related risk is data. The customer's phone number and email arrive only after you pay, and they are the part of the lead Bark's terms protect. If you later keep notes in a CRM or a spreadsheet, treat them the way you would any customer record, and switch off the model-training setting in the privacy settings of any consumer AI account you use for drafting.

Checking the routine after four weeks

Keep a one-sheet ledger with a row per lead: date bought, cost in dollars, screening score, time of first reply, replies received, quotes, outcome, and gross profit if won. Add one row for each pack: date, price, credits, expiry date. Four weeks of rows is enough to see whether the routine helps.

Compare three numbers with last month's: cost per lead answered, reply rate and cost per job won. If the reply rate rose and cost per job fell, keep the routine. If the number of leads fell but cost per job did not, your screening is too strict and you are declining jobs you would have won; lower the threshold by one point and run another four weeks. If nothing moved, look at the quotes themselves, because no first message can rescue a price that is too high or a visit that went badly.

The firm's rule fits on one line: buy the next pack only if cost per job on the last pack stayed under the ceiling, and write down why whenever you break that rule. If the ledger will cover several lead sources, plumbing lead generation with cost per job tracked by AI summaries takes the same habit across more than one source.

Bark credits: what tradespeople ask next

Does Bark take commission on jobs I win?

Bark's pricing page says there is no commission and no hidden fees. Your cost is the credits you spend to contact customers, plus any subscription you choose to take. What you charge a customer for the job is yours, because Bark's terms say you set your own rates. Check the current terms in your account, as wording differs between markets.

Do I get credits back if the customer never replies?

No. Bark's terms say credits are non-refundable except in the cases they list, and they say nothing is returned just because a customer does not respond or you do not win the job. The listed exceptions are narrow, such as an invalid phone number or a bouncing email reported within seven days.

How long do Bark credits last?

Credits bought on or after 1 November 2025 are valid for three months from the date of purchase, according to Bark's terms; the pricing page simply says three months. Credits bought before that date had twelve months. Expired credits are not refunded or extended, so buy packs sized to the leads you expect to answer in three months.

Can I use ChatGPT or Claude to write my Bark replies?

The terms I read have no rule about templates or automated replies. They do ask for clear, prompt communication, bar false or misleading information about your identity or location, and stop you making customers' contact details available to third parties without Bark's written agreement. Paste only the job details into an AI tool, review every draft, and send it yourself.

How many other professionals get the same lead?

For a general request, Bark's terms say a customer can send it to up to five professionals they have not named, so assume you may be one of five. Enquiries sent to you directly through Bark Marketplace are different: the customer chose you by name, and the credits come off when the customer contacts you.

Further reads

• Angi Leads for Plumbers and HVAC: Worked Cost per Job and AI Follow-Up — The same cost-per-job sum for a different paid-lead source.
• Deck Builder Leads: What They Cost and How to Reply Fast With AI — Speed-to-reply routine for a trade where quotes need photos.
• Insulation Leads: What They Cost and How AI Qualifies Them — Qualifying questions that screen out leads before you spend on a visit.
• AI Lead Generation vs Hiring a Salesperson: A Cost Comparison — Puts paid-lead costs next to the cost of a person chasing work.
• Pool Service Leads: Compare Paid Leads, Local Services Ads and Fast Replies — Compares paid leads with pay-per-lead ads and fast follow-up.
• Junk Removal Leads: Follow Up Bought Leads Before a Rival Does — Text bought junk removal leads in two minutes, ring in five, then log cost per booked job to see if buying leads pays. Scripts and duplicate rules included.
• MyBuilder vs Checkatrade for Plumbers: Cost per Job, With AI Replies — How MyBuilder and Checkatrade bill, one month costed per job won on both, and a six-week test of whether faster AI-drafted replies pay.
• Boiler Replacement Leads: Follow Up in Minutes Using AI — Follow-up timings for boiler replacement leads, AI first-reply examples, quote-visit booking wording and a three-message chase that still sounds like you.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Bark's pricing page for professionals and its terms for professionals, including the clauses on credit validity, returns and the Get Hired Guarantee (read October 2026); Bark help-centre descriptions of lead preferences, taken from search-result extracts because the help pages would not load.