AI Real Estate Listing Presentation: Build the Seller Deck

Coding Liquids tutorial cover featuring Sagnik Bhattacharya for AI Real Estate Listing Presentation: Build the Seller Deck.
Coding Liquids tutorial cover featuring Sagnik Bhattacharya for AI Real Estate Listing Presentation: Build the Seller Deck.

Supply the verified material yourself: your sales record, adjusted comparable sales and a costed marketing plan. Let AI turn it into a slide outline, plain-English wording and speaker notes. Then check every number and claim against its source before a seller sees a slide. The comps and pricing page deserves the most care.

A seller will remember two things from the meeting: why you chose that price and what happens in the first four weeks. Most AI deck tools, given only an address, produce handsome slides full of statistics nobody can source, which is the wrong way round. Start with the evidence, then ask for layout. Every home, price and fee below is invented for one imaginary house.

Follow me on Instagram@sagnikteaches

Start with an eleven-slide spine and a source for every slide

Structure matters more than design. Write the spine first, name where each slide's facts come from, and only then let AI draft wording. The table shows what you bring to each slide and what the assistant does with it.

Connect on LinkedInSagnik Bhattacharya
SlideFacts come fromThe assistant's job
1. About me and how I workYour own transaction recordsTighten your wording
2. What you told me you wantThe seller's answers to a pre-meeting questionnaireSummarise them in plain words
3. The market in three numbersAn MLS report with its date range and areaExplain what each number means
4. Your home through a buyer's eyesYour walk-through notes and photosTurn notes into a features list
5. Comparable salesMLS records opened one by oneFormat the table; calculate with formulas
6. Price: range and recommendationYour adjusted compsDraft the explanation of the logic
7. The first four weeksYour real service menu and costsDraft the timeline
8. Getting the home readyYour preparation checklistTailor it to the walk-through notes
9. Offers and negotiationYour processWrite it in plain steps
10. Estimated net proceedsThe seller's payoff figure and your fee scheduleLay out the sheet with formulas
11. Fees, agreement and next stepsYour brokerage's approved wordingNothing on the fee text

Notice that the assistant never supplies a fact. It formats, explains and tightens. That division is what keeps the accuracy checks later in this tutorial short, because each figure has an owner and a source before the assistant has touched it.

Subscribe on YouTube@sagnikteaches

The comps and pricing page, worked through for one home

The imaginary subject is a three-bedroom, two-bathroom home of 1,850 square feet with a two-car garage and its original kitchen. The agent pulls four closed sales from the last five months in the same neighbourhood, opens each MLS record to confirm the sold date and price, and rejects a fifth because it has four bedrooms and 2,150 square feet. The adjustment rates are the agent's own: $85 per square foot of size difference, $6,000 for a garage space and $10,000 for an updated kitchen. Replace them with rates your market supports.

CompSizeSold (days listed)Sold vs listAdjustmentsAdjusted
11,800 sq ft$418,000 (24)98.4%+$4,250 size$422,250
21,920 sq ft$436,000 (12)99.3%-$5,950 size$430,050
31,850 sq ft, one-car garage$409,500 (41)98.7%+$6,000 garage$415,500
41,760 sq ft, updated kitchen$431,000 (6)100.5%+$7,650 size, -$10,000 kitchen$428,650
52,150 sq ft, four bedrooms$472,000Not usedDifferent productDropped

The four adjusted values run from $415,500 to $430,050, with a median of $425,450 and a mean of about $424,100. Two cross-checks give the agent confidence. The four sales average about $231 a square foot, which applied to 1,850 square feet gives roughly $428,000, within 1% of the median. The sold-versus-list ratios run from 98.4% to 100.5% and average 99.2%, and the days listed run from 6 to 41, with a median of 18. Only one of the four sold above its list price, and it was the home with the updated kitchen, which is exactly the sort of detail that belongs in the speaker notes.

Use the assistant for the tedious work and keep the judgement. Paste the four rows, give it your adjustment rates and ask for the table, the median, the mean and the ratios as spreadsheet formulas. Do not ask it to find comps or to choose adjustments. What you check by hand is each MLS record, each sold date, and the sign of every adjustment (a better home than the subject is adjusted down, a worse one up). The tutorial on preparing a CMA with AI covers how to choose comps in the first place.

The Complete Excel Course with AI Integration thumbnail

The Complete Excel Course with AI Integration

From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.

By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner

Buy now

Three list prices and what each asks the seller to accept

A pricing slide that names one number invites an argument. A slide that sets out three options, each with its trade-off, turns the meeting into a decision. For the imaginary home, with a median adjusted value of $425,450, the options look like this.

List priceAgainst the median adjusted valueWhat it asks the seller to accept
$439,9003.4% aboveThe home must show better than the comps, and a price reduction may follow if early interest is thin
$424,900In line (0.1% below)A fair-market price: timing and price are balanced, as the four comps suggest
$414,9002.5% belowA deliberate under-pricing to attract competing offers, which are not guaranteed and may not exceed list

The recommendation goes beneath the table, in words the seller can repeat to their family: "Based on four sales in the last five months, adjusted for size, garage and kitchen, I put the value between $415,500 and $430,000. I recommend listing at $424,900. If speed matters most, $414,900; if you would rather test the market, $439,900, knowing a reduction may follow." Nothing in that paragraph promises a result. It reports evidence, names the trade-offs and leaves the decision with the seller.

Ask the assistant to draft this paragraph from your table and then read it for promises. Words such as "will sell", "guaranteed" and "certain" come out. So do any figures that are not in the table, because assistants like to add a plausible average days-on-market of their own.

The net proceeds slide: the page sellers check twice

Sellers study the net sheet more closely than the comps, because it tells them what they will walk away with. It is also the slide where an arithmetic or sign error does the most damage. Here is the imaginary sheet at an expected sale price of $420,000, which is 98.8% of the recommended list price and a little more cautious than the 99.2% average of the comps.

LineAmountWhere the figure comes from
Expected sale price$420,000Your pricing logic, rounded down
Mortgage payoff-$186,500The seller's payoff letter
Your fee-$21,000Your listing agreement
Title, escrow and closing fees-$5,040The closing company's estimate
Repair credit-$2,500An assumption about inspection results
Property tax proration-$1,150Calculated for the expected closing date
Estimated net$203,810Sum of the lines above

The fee line is whatever your agreement says, so the sheet should use your figure or formula rather than anything the assistant suggests. Fees, taxes and customs differ by place, so ask the closing company for its estimate, and label the slide an estimate, not a guarantee. Ask the assistant to build the sheet with formulas so you can change the sale price and watch every dependent line move, and check each sign yourself. A property tax proration, for example, can be a credit or a charge depending on whether the seller has paid ahead, and a confident-sounding assistant will pick one without asking.

A marketing plan slide that is costed, not decorated

Slide 7 is where sellers decide whether you will do the work. A plan that names dates, owners and decision points beats a page of icons. The sample timeline below is drafted from an imaginary service menu; yours will differ, and every row should match something you actually do.

WhenWhat happensWho does itWhat to verify
Before listingPhotography and floor plan booked, repairs agreed, listing text draftedAgent, photographer, sellerSupplier quotes in writing; listing text checked for fair-housing wording
Week 1Goes live on the MLS and your site; first open house; email to people who have agreed to hear from youAgentThat the email list holds only consented contacts
Week 2Showing feedback collected and shared; second open house if usefulAgentCount real showings; don't estimate them
Day 14Agreed price review against showings and feedbackAgent and sellerThe decision date is written into the plan in advance
Weeks 3 and 4Adjust marketing or price; weekly update callAgentUpdate dates in your calendar before the meeting ends

The assistant is useful for turning your bullet list into this table and for writing the seller-facing wording. It is also where over-enthusiasm creeps in. Asked to "make the plan impressive", an assistant may add drone video, premium portal placement and a reach of 50,000 buyers. If you offer neither service and cannot source the audience figure, those lines make promises you cannot keep. Give it the rule: "List only services in this menu. No reach, view or ranking claims." The four-week content plan tutorial goes deeper on the social and email side, and the tutorial on AI disclosure for listing photos and copy covers what to say when images are edited or virtually staged.

Prompts that keep the assistant on the evidence

Drafting slides from the verified rows works best when the prompt forbids invention and asks the assistant to say what it was missing. A short, strict prompt beats a long creative one.

Draft slides 5 and 6 of a seller listing presentation. Use only the data below.
Do not add statistics, rankings, averages, reach figures or claims about the
area or the people who live there.
For each slide give: a title of 8 words at most, three bullets at most, a
one-sentence "so what" for the seller, and speaker notes of about 80 words in
plain, calm English.
Finish with a list headed NEEDED BUT NOT GIVEN, naming anything you wanted
and did not have.

Data: [paste the comps table and the three price options]

An invented but typical draft for slide 6 looks like this, and it needs fixing in three places.

Slide 6, "Pricing: three options". Bullets: value range $415,500 to $430,050; recommended list price $424,900; alternatives $414,900 or $439,900. So what: the price you choose decides who sees your home first. Speaker notes: "Homes in this family-friendly neighbourhood usually sell within 14 days, so a fair price matters. Our four comparable sales support a range of about $415,000 to $430,000..."

Three edits are needed. "Family-friendly neighbourhood" goes, because it describes the people who might live there and invites a fair-housing problem; the tutorial on neighbourhood pages that avoid steering explains why. "Usually sell within 14 days" goes too, because the data show 6 to 41 days with a median of 18, and the 14 came from nowhere. The "so what" line becomes "each option trades price against speed", which the table actually supports. The NEEDED BUT NOT GIVEN list is worth keeping: when it asks for local market statistics, you get a prompt to pull the real MLS figure instead of letting the assistant guess one. The same approach works for slide 3, where the market update tutorial shows how to verify each figure.

Choosing where to build the deck

The tool matters less than the spine and the sources, and the best choice is usually the one you already pay for. This table lists the realistic options with what could be verified about each; prices change often, so check the vendor's page before you commit.

OptionWhat is knownWatch-outs
Your brokerage's templateBrand, disclosures and fee wording are already approvedFill it with AI-drafted text; don't let a tool redesign it
PowerPoint with CopilotMicrosoft's documentation (updated 28 October 2026) says Copilot in PowerPoint can create presentations from prompts or existing content, with standard access on the Basic tier and priority access with the add-on licence. Microsoft's Copilot Business page did not display prices when checked in November 2026Check which Copilot access your licence gives you, and get the add-on price from Microsoft or your reseller
Google Slides with GeminiGemini is built into Google Workspace plans from Standard upwards, across Docs, Sheets and SlidesPersonal accounts and lower plans differ; check yours
CanvaAI features run on a plan-based allowance: Pro ($144 a year for one person) gets ten times the Free plan's AI and Business ($250 a year per person) twenty times. Brand Kits hold your colours and fonts (Free allows one, limited to three colours)Allowances have changed recently, so read the pricing page first
GammaBuilds a deck from a prompt and imports PDF or PowerPoint files. Free gives 400 one-off credits and up to 10 slides per prompt; Plus is $9 and Pro $18 a seat a month billed annually; export to PowerPoint, PDF and Google Slides is listedCheck its data settings before pasting seller details; the Gamma plans tutorial compares the plans
Beautiful.aiPro is $14.50 a month billed annually, Team $40 a user a month billed annually ($50 monthly); no free plan; a 14-day trial needs a card; PowerPoint import and export are listed (vendor pricing page, November 2026)Check the cancel-before-trial-ends rule

A simple decision rule works for most agents. If your brokerage supplies a listing presentation template, build inside it and use AI only for the words. If it does not, choose the tool you already have a licence for, because a new subscription and a new interface add cost without improving the evidence. Reach for a dedicated AI deck builder only when design is a genuine bottleneck, and test the export to PowerPoint or PDF with a real deck before you rely on it, since a layout that looks perfect on screen can shift when exported.

Accuracy checks before the seller sees a slide

Print the deck and go through it with a pen, writing the source page for every number in the margin. These are the claims that most often need correcting, with the check for each.

What the slide claimsHow to check itA failure to expect
A market statisticOpen the MLS report; confirm area, date range and whether it is a median or an average"Sells in 14 days" taken from a wider area or an older period
A comparable saleOpen each MLS record; confirm it is closed, not pending, and check the date and priceA pending sale used as a sold comp
An adjusted valueRecompute in a spreadsheet; check the sign of every adjustmentA better home adjusted up instead of down
Your track recordRun your own transaction report for a stated periodA round number, or a "top agent" claim with no dated, sourced ranking behind it
Any promise of speed or priceDelete it; show the comps instead"We'll have it sold in a week"
Neighbourhood wordingRead it for descriptions of people; keep to facts about the property and nearby amenities"Family-friendly", "safe", "perfect for professionals"
Edited or AI-generated imagesLabel them; compare with the unedited photosA virtually staged room shown without a label
Fee and agreement wordingUse the brokerage's approved text onlyAn assistant paraphrasing a clause
Net proceedsGet the closing company's estimate; label the slide an estimateA sign error on a proration

Keep a source file for each presentation: one folder holding the MLS reports, the payoff letter, the closing estimate and your transaction report. If a seller or your broker asks where a figure came from, the answer takes ten seconds. The folder also makes the next listing quicker, because the market and track-record slides need only an update.

Speaker notes and the questions sellers always ask

With the slides settled, ask the assistant for speaker notes in your voice. Paste a paragraph you have written yourself so it has something to match, and tell it to keep each slide's notes to about 80 words. Then ask for the questions sellers tend to ask, each with an answer scaffold that points back to a slide, so you can practise out loud.

  • "Why not list higher?" Slide 6: the comps and the three options, including what a higher price asks the seller to accept.
  • "What if it doesn't sell?" Slide 7: the day-14 price review, agreed in advance.
  • "How will you find buyers?" Slide 7: the services you actually provide, in the order they happen.
  • "What will I walk away with?" Slide 10: the estimate, and who produced each line.
  • "Why that fee?" Slide 11: your agreement's wording, read aloud, never improvised.

The notes are practice material, not a script to read. Run through the deck twice with the assistant playing a hesitant seller, then once on your own. The aim is a deck in which every number has a source, every option has a trade-off and every promise has been deleted, so the seller is deciding on evidence rather than on polish.

Questions agents ask about AI-built listing decks

Can AI choose my comparable sales?

It should not. Comparable sales come from your MLS records, opened one by one, with sold dates and prices checked. An assistant can format the table, compute price per square foot and ratios with formulas, and explain adjustments you chose. If it supplies comps of its own, treat them as invented until you find each in the MLS.

Should the deck state an expected sale price?

Show a value range with its evidence, a recommended list price and the trade-offs of each option, and say plainly that the market decides the sale price. Avoid promises such as a specific price or days on market. The seller chooses the list price, and your comps explain why you recommend the one you do.

What about AI-generated images in a seller deck?

Label any AI-generated or edited property imagery clearly, and never use it to suggest condition, views or features the home does not have. Check your MLS and brokerage rules on virtual staging before using such images in marketing, and keep unedited photos available for the seller to compare.

Can I paste the seller's details into an assistant?

Use the tool your brokerage has approved, and give it only what the slide needs. Names, loan balances and contact details are rarely required for wording help. Business plans such as ChatGPT Business or Claude Team don't train on business content by default, but your brokerage's policy still decides what may be uploaded.

Further reads

Sources: Microsoft Learn's Copilot overview (updated 28 October 2026), and the pricing pages of Google Workspace, Canva, Gamma, Beautiful.ai, ChatGPT Business and Claude, all checked November 2026. Every home, price, fee and cost in the examples is invented for illustration.

Want your listing presentation built on solid evidence?

On a 1:1 call we'll look at how you prepare for seller meetings, where AI can draft the deck safely, and which numbers must always come from your own records.

Book a 1:1 call with me